GTM Operator Field Manual

The step-by-step system for delivering client pipeline. Written for an operator on day one with zero prior context.

Start here — your first day

This manual is big, but you do not read it all. On day one, read these six, in this order, and you can run:

1. The map below (how the parts connect)  2. The campaign runbook (the order you do things)  3. Part 0 (the rules)  4. Part 18 (how to open a DM)  5. Part 17 (how to run the chat to a booked call)  6. Part 12 (your daily routine).

Your actual daily job is three things: engage and post, send a few good DMs, book meetings and log them. That lives in Parts 12, 17 and 18. Everything else is setup you do once (Parts 2, 3, 16) or reference you look up when you are stuck (Parts 13–15).

Currency note. LinkedIn and WhatsApp do not publish exact limits, and the algorithms shift. Every number here is a safe ceiling, not a target. Re-verify quarterly, and when in doubt, go slower.

THE MAP — HOW THE PARTS CONNECT

STARTING A NEW CAMPAIGN — THE ORDER TO DO THINGS IN

Rough timing: Parts 3–5 in the first two weeks, content live from about week 2, first conversations from about week 3, and reporting every month. Guides, not deadlines.

First, the thing that is NOT part of the campaign: the accounts are already set up. The advocate LinkedIn profiles and WhatsApp numbers are a standing, pre-warmed pool (Part 2, with the click-by-click build in Part 16). They take about 30 days to warm, so they are ready before a client signs. A campaign never begins with "set up accounts." If none are free, warm them first — that is a prerequisite, not Part 3.

Part 3 · Onboard, kickoff & decode

Part 4 · Define & map the ICP

Part 5 · Build the target list & committee map

Part 6 · Content & channel activation

Part 7 · Warm the network & harvest

Part 8 · Validate messaging & convert

Part 9 · Track & recalibrate

Reference — look these up when you need them, not in order:

Parts 0–1 orientation, the golden rules, and the engine map (read once at the very start).
Part 2 + Part 16 the advocate-account infrastructure and its click-by-click build SOP (the standing setup above).
Part 17 the deep conversion/messaging SOP — this powers Part 8, step 12.
Part 18 the offer + first-DM SOP — this powers Part 8, step 11.
Part 13 tools, Part 14 pre-flight checks, and Part 15 troubleshooting when a step stalls.

PART 0 · ORIENTATION (read this first)

0.1 What XpandEast does

Foreign B2B tech companies fail in SEA, ANZ, and the Gulf not because their product is weak, but because buyers reject a stranger. Enterprise committees here buy names their own network already trusts. Our job: make the client an insider, a recognized, trusted name, then convert that trust into booked sales calls. We hand off at the first confirmed meeting; the client's team closes.

0.2 What YOU (the operator) do, every month, per client

1. Keep a roster of advocate LinkedIn profiles healthy and posting (Part 2).
2. Run ICP intelligence: learn what your buyers read, follow, and engage with (Part 4 / Part 4).
3. Maintain a scored target account list with the buying committee mapped (Part 5 / Part 5).
4. Ship content through advocates that the right buyers actually engage with (Part 6 / Part 6).
5. Harvest the engagement, turn engagers into warm leads (Part 7 / Part 7). Most of your meetings come from here.
6. Run the outreach ladder to convert warm leads into booked calls (Part 8 / Part 8).
7. Report and recalibrate (Part 9 / Part 9).

0.3 The one-sentence engine

We study what our buyers trust and build advocate authority on those exact topics. The people who engage become the warm path: we qualify them, engage, and book the meeting. The ICP who never engage become the cold path: harder, more layers of value across formats and a longer cadence, before any ask. Both walk the same channel ladder (LinkedIn, then earned WhatsApp, then a call).

0.4 Two "doors", know which one this client uses

The engine is identical; only the front door changes.

• Door A, Demand-gen led. Warm an end-customer committee from zero. Content leads; outreach follows engagement.
• Door B, Channel/source led. Reach buyers via partners/resellers or a pre-qualified registry. Outreach off a scored list leads; content gives air cover.

Most clients are mostly Door A with some Door B. You set the mix in Part 3.

0.5 The golden rules (breaking these is how engagements fail)

0.6 The four approval gates (never automate these)

• Gate 1, XpandEast Founder / Top Management signs off the ICP and roadmap before work begins.
• Gate 2, the client approves every post published on their official company channels within 24h. Content on advocate profiles does not need this sign-off.
• Gate 3, the GTM operator reviews and edits every outreach message before it sends, even when a custom LLM prompt generated the draft.
• Gate 4, XpandEast Founder / Top Management reviews the monthly report before the client sees it.

PART 1 · THE ENGINE MAP (how the parts connect)

1.1 Why Month 2 is the inflection (the "winning advocate" rule)

You will run several advocates at once and you will not know in advance which one the ICP responds to. 360Brew takes time to categorize a profile's expertise (it can take ~90 days of consistent on-topic posting to build a clear "semantic fingerprint"), so reach and engagement per advocate only become legible after a few weeks. By end of Month 2, measure ICP-engagement per advocate, identify the 1–2 winners (the advocates whose content the right buyers actually engage with), and concentrate content and effort there. Until then, treat Months 1–2 as a discovery period across the roster.

PART 2 · INFRASTRUCTURE, the channels & accounts you run on

2.0 The WhatsApp sending SOP (follow this exactly)

WhatsApp is where SEA and Gulf deals actually move, and it is the easiest channel to get a number banned on. These rules keep a number alive. They are not optional.

If accounts get restricted or throttled, nothing works. Operate a roster of real, consenting advocate profiles so each looks like a normal human, while publishing content and opening conversations. The advantage is real, but it is not "post once a day on ten profiles." In 2026 that pattern destroys reach (6.1, 6.8).

The advantage is: each profile posts 3–5 quality posts per week on a tight topic, engages far more than it posts, and stays inside one safe session. Ten healthy, engaging profiles multiply presence; ten daily-posting, throttled profiles multiply nothing. "Healthy" is the whole game, and two things kill health most often: the wrong session setup (2.2 / 2.5) and posting too much without engaging (6.8). Read both before activating a roster.

The full click-level setup-and-operation procedure (proxy + GoLogin configuration, the day-by-day 30-day warm-up, the daily routine, example comments) is Part 16 (SOP-A), this part is the strategy, Part 16 is the exact procedure.

2.1 The SSI reality (important)

Most of our advocates sit below 40 SSI (Social Selling Index, LinkedIn's 0-100 score of how established and active an account looks). Do not build the plan around high-SSI thresholds — advocate accounts realistically will not hit 70. Still, always nudge SSI upward where you can: complete the profile, post on-pillar, and engage daily. Two implications:

• Good news (360Brew): content reach now depends far more on topical relevance and semantic match than on SSI or follower count, small, low-SSI accounts can still reach the right buyers if the content is genuinely relevant and on a consistent topic. SSI is no longer a reach gate.
• Caution (outreach safety): low SSI = low trust history = a smaller safety buffer for outreach. So keep outreach volumes at the conservative end (2.4), warm carefully, and lean on relevance, not volume.

2.2 Infrastructure (one-time setup per advocate)

LinkedIn builds a behavioral fingerprint per account from device, IP, login pattern, and action speed, and it now flags multiple concurrent or alternating sessions (logins from different devices/IPs) as a primary sign of bot activity or account compromise. Isolate each advocate and keep each one on one consistent session:

1. One GoLogin browser profile per advocate. Never two advocates in one browser/session.
2. One dedicated residential proxy/IP per advocate, fixed, and geolocated to the advocate's real city. The IP must stay consistent; hopping IPs/countries, or using free/datacenter proxies (already blacklisted), triggers distrust and verification prompts. (A likely cause of past bans: accounts sharing a network fingerprint, jumping IPs, or running multiple sessions, see 2.5.)
3. Consistent device fingerprint in GoLogin, don't keep changing it. Changing the email/password does not reset risk; LinkedIn identifies the environment (IP + device + fingerprint), not the credentials.
4. One session active at a time. Close one session before opening another. Do not have the GoLogin profile, an automation tool, and the advocate's phone all logged in at once (2.5, the three-session trap).
5. Profile completed to 100% before anything: real photo, banner, a keyword-rich headline that names the advocate's topic pillar (this feeds 360Brew's semantic read of the account), full About, 3+ experiences.
6. Don't churn profile fields (frequent name/title/company changes raise risk flags).

2.3 The 8-week warming ramp (new/cold advocate accounts)

The first ~20–30 days are a behavioral-modeling window. No automation in this window. Ramp like a human; cap increases at ~30%/week per action type.

Health gates during warming, if any fail, stop increasing and reduce volume:

• Acceptance rate above 35–40%.
• No repeated CAPTCHAs, forced logouts, or "verify identity" prompts.
• (SSI will likely stay sub-40; that's fine, judge health by acceptance + no warnings, not SSI.)

2.4 Safe operating limits per WARMED account (conservative, our reality)

Behavioral hygiene that keeps accounts alive:

• Distribute across the week and business hours. All requests Monday 9am = automation signal. Spread the 15–20/day.
• Weekends at 30–50% of weekday volume; take one full day off per week. Real humans are inconsistent.
• Personalize. Beyond reply-rate, generic outreach that matches spam patterns is now actively flagged by LinkedIn, personalization is a safety measure, not just a performance one.
• Avoid "I don't know this person" reports. A few of these degrade an account fast. Only request people who plausibly recognize the advocate or topic, warm them with genuine content engagement first (Part 7).

2.5b Extra automation-safety rules (added 2 Aug 2026)

• Disable Grammarly and ad-blockers on LinkedIn while automating. Their injected scripts clash with the automation and read as an unusual environment.
• Anti-detect browser is non-negotiable for a fleet. GoLogin is our standard; Multilogin is a fine alternative. One profile per advocate, never shared.
• Avoid cheap sub-$50/month cloud automation tools. They run the session from their own servers (the exact thing LinkedIn hunts, see 2.5) and get whole tool providers banned. If a tool is that cheap, you are the product being burned.

2.6 Tier the roster so effort matches output

One operator can run the daily routine for roughly 8–12 warmed accounts if Tier-1 content is produced centrally (Part 6). Stagger activity windows so the pod never acts at the same minute. Keep a one-row-per-account health sheet (last login, acceptance %, pending count, warnings) and check it daily (Part 12).

PART 3 · ONBOARD, KICKOFF & DECODE

Goal: take a signed client from "deal closed" to "engine ready," inside the first 7 business days, so that everyone knows what is happening, who owns what, where progress is visible, and what we need from them. Outputs: a run kickoff, the access and infrastructure set up, the inbound assets collected from the client, the ICP, the regulation classification, the channel map, the 6-month roadmap, and the config sheet.

Nothing in Part 4 onward begins until Gate 1 (XpandEast Founder / Top Management signs the ICP and roadmap) and the client's access/assets are in.

Why this part decides the engagement. Projects with a properly run kickoff are markedly more likely to hit their objectives, and the client forms their opinion of us inside the first week. The two failure modes to avoid: starting execution before access, assets, and approvals are sorted; and treating the kickoff as a social call instead of the first real delivery meeting.

3.1 The onboarding timeline (first 7 business days)

If onboarding regularly runs past two weeks, it is a process problem, not a client problem.

3.2 Before the kickoff: internal handoff + welcome

1. Internal handoff (sales -> delivery). The person who sold the engagement transfers to the operator: scope, what was promised, pricing/commitments, the meeting target, the priority market, and any context on stakeholders. The most dangerous moment in the relationship is this handoff, because if sales promised something the operator never heard, trust is lost before work starts. Capture it in one handoff note.
2. Welcome email. Thank them, confirm the kickoff time, name the delivery team, and attach the kickoff agenda plus a short "what we will need from you" list (3.4) so they arrive prepared.
3. Internal risk rating. Flag onboarding risks before the account starts: missing ICP/account data, no named content approver, multiple decision-makers, unrealistic deadline, fragmented brand assets, or a client switching from a frustrating prior agency. Seeing the friction before kickoff lets you manage it before the client feels it.

3.3 The kickoff meeting (video, ~45 minutes)

This is the first real delivery meeting, not a demo and not a recap of the sales calls. Send the agenda 24 hours ahead. Attendees: from XpandEast, the account lead + the operator/PM; from the client, the project sponsor / executive buyer (to confirm commitment) and the day-to-day owner / content approver (who you will work with weekly). Run it lead-with-outcome.

Run of show:

1. Welcome + intros (keep to ~60 seconds each): name, role, and what each person wants from the engagement.
2. Confirm scope and success. State explicitly what is in scope and what is not (this is where scope creep is prevented). Confirm the meeting definition (what counts as a qualified call), the monthly target (5–10), and the framing we hold all engagement to: content = trust + visibility; pipeline = sales calls + positive replies + newsletter subscribers warmed from content.
3. Walk the 6-month roadmap (3.9). Set the expectation out loud that Months 1–2 build trust and visibility with no meetings promised in Month 1. Trust and meetings compound across Months 2 and 3, proposals start Month 4, Months 5 and 6 iron out proposals in physical meetings, and closes then run 6 to 9 months. A visible path to the goal is what makes the client confident they chose right.
4. Communication norms. Agree the weekly sync cadence and that the Infinity workspace is the single source of truth (share screen, walk the boards). Confirm response-time expectations and that content is approved by the named approver within 24 hours (Gate 2).
5. What we need from you (3.4). Walk the request checklist and assign an owner + due date to each item, so assets do not arrive piecemeal across email and chat.
6. Access & infrastructure (3.5). Walk the Infinity access, the communication email handle, the WhatsApp setup, and the CRM-sync arrangement, assigning owners.
7. Surface risks/dependencies. Missing data, approver availability, legal sign-off, brand-asset gaps. Name them now.
8. Next actions + first check-in date. End the meeting with the client knowing three things: what happens in week one, who to contact, and when the first sync is.

Within 2 hours of the call, send a written summary (decisions, owners, due dates, next sync). Not the next day. Two hours. Then send a short progress note in week one even if it is small, so that by day 30 the client has never wondered what we are doing.

3.4 What we request from the client (the inbound checklist)

Collect these in the first few days (assign owners at kickoff). Pre-fill anything sales already captured so the client only fills gaps.

3.5 Access & infrastructure setup

Onboarding is a security event as much as a workflow event. Use secure, role-based access, document who has what, and revoke promptly when team access changes. There are four things to set up, in this order.

3.6 Classify the business, regulated vs. not (do this early)

This decides how carefully you must source and handle data later.

• Highly regulated (fintech, banking, insurance, healthcare, gov/defense): personal-data sourcing must come from legitimate, documented sources only. A fintech client can be fined for data sourced improperly; it also exposes XpandEast. Flag the client as REGULATED in the config sheet and apply the data-legitimacy rules in Part 5 (5.3). Note the key compliance fact: cold outreach to personal mobile numbers sourced from third-party databases (Clay, Airscale, Cognism and similar) defeats the B2B exemptions most people rely on, so warm, earned, opt-in-adjacent contact is the safer path.
• Standard B2B tech: still respect PDPA/GDPR/Spam-Act basics, but with normal commercial latitude.

Write the data rule explicitly for this client (e.g. "REGULATED, source only from public professional profiles, opt-in directories, accreditation registries, and the person's own published business contact info; document provenance for every record; no scraped/purchased databases of unknown origin").

3.7 Build the ICP (qualifiers AND disqualifiers)

The disqualifiers keep the list clean.

ICP, [Client]    | Regulation class: [REGULATED / STANDARD]
COMPANY FIT (qualifiers): industry/vertical · size band · the pressure signal that means "need this now"
COMPANY DISQUALIFIERS (remove if true): too small / wrong segment / no digital readiness / wrong geo …
PERSONA / COMMITTEE SEATS: Champion (feels pain) · Influencer/architect (shapes shortlist) · Economic buyer (signs) · Gatekeeper (blocks)
TRIGGER EVENTS (reasons to reach out that aren't a pitch): recent hire / funding / launch / regulation / public post …

3.8 Decode the channels per market

First-touch and second-touch differ by country. Write the explicit rule:

Rule everywhere: LinkedIn starts relationships; WhatsApp closes meetings (SEA/Gulf); WhatsApp is never cold. (Note: this table is incomplete for Thailand and Facebook-active markets. See the channel-map gap flagged for v3.)

3.9 The 6-month roadmap + monthly deliverables

Set expectations; protect Months 1–2 (no meetings). Agree the meeting definition up front (what counts as "qualified") and split end-customer vs partnership meetings in the count so the metric can't drift later. Present this in the kickoff as the client's visible path to the goal.

Targets in brackets are illustrative examples, set the real numbers per client at kickoff.

3.10 The internal config sheet

One sheet per client every later part reads from: ICP + disqualifiers, regulation class + data rule, committee seats, channel rules, the 3–4 content pillars (filled in Part 4), the active winning angle (updated monthly), the give-to-get assets, the meeting target, the email-handle/domain decision, the CRM-sync method, and the advocate roster + who's the winner.

Deliverables: run kickoff + 2-hour summary · client access to Infinity + assets collected · email-handle/WhatsApp/CRM set up · ICP + channel map + 6-month roadmap + config sheet. → Gate 1 (founder approves ICP + roadmap).

PART 4 · DEFINE & MAP THE ICP(the step most operators skip)

Goal: before building lists or writing a single post, understand your buyers cold: what they read, who they trust, what content they actually engage with, the language they use, and where they form opinions (including AI tools). This output defines the content pillars and angles and sharpens the sourcing. Skipping this is why content misses.

This is a Week 1–2 activity at kickoff, refreshed quarterly.

4.1 Map what the ICP consumes, SparkToro

SparkToro reveals the websites, podcasts, YouTube channels, publications, social accounts, keywords, and topics a defined audience actually engages with (and, as of 2026, you can describe the audience in plain language, e.g. "Heads of Retail Marketing at mid-market Malaysian retailers").

Steps:

1. Run the audience query for the ICP (use the natural-language description from the ICP).
2. Pull the top results from each tab: Websites, Social Accounts, Podcasts, YouTube, Publications, Keywords, Topics, and Search & AI Tools.
3. Use the "Hidden Gems" filter to surface high-engagement accounts with smaller, focused followings, these are the niche thought leaders your buyers actually trust (more useful than the obvious big names).
4. Use "Compare Audiences" to analyze a competitor's or a key influencer's audience and find overlaps and white space.
5. Note the Search & AI Tools data: B2B buyers increasingly form opinions via an LLM (Claude, Gemini, ChatGPT or Perplexity) before they ever reach the client, capture this for the AEO angle (your content and the client's presence should be answerable by AI engines; see 6.7).

Output a Consumption Map: the 10–20 sources (sites/podcasts/creators/publications), the recurring topics, and the keyword/language bank your buyers use.

4.2 Mine the ICP's LinkedIn behavior directly (manual, high-signal)

SparkToro is aggregate; this is specific.

1. Take 20–30 representative ICP prospects (from the list, or hand-pick exemplars).
2. Open each one's Activity tab (Posts / Comments / Reactions).
3. Log: what topics they react to and comment on, which creators/voices they engage, which formats (text/carousel/video), what language and pain points they use in their own comments, and what they complain about or celebrate.
4. Synthesize the patterns: the 5–10 themes that reliably get this buyer to stop and engage.

4.3 Track the thought leaders who already reach your ICP

1. From 4.1 (Hidden Gems) + 4.2, list 5–15 thought leaders who write to our ICP and whose topic overlaps ours.
2. For each, study what content and angles they use that get ICP engagement, the hooks, the formats, the recurring arguments, the comment debates.
3. Purpose is twofold: (a) understand what resonates so our pillars are grounded in reality, and (b) find the white space, the angle they're not covering that we can own. We learn from them; we don't copy them.
4. Add them to a tracking list (a saved LinkedIn search or a simple sheet) and check periodically, they are an early-warning system for what's landing with the ICP.

4.4 Track the top engagers in your ICP

1. Identify the most engaged people in the ICP, those who comment substantively and often on relevant content (you'll see the same names recur in 4.2/4.3).
2. Track what they engage with, they are a live signal of what's resonating and a warm-lead source (they're active, opinionated, and reachable).
3. Tag them in the CRM as "high-engagement ICP", these are priority targets for the harvest and outreach (and natural candidates to pull into comment threads on advocate posts).

4.5 Synthesize into the content strategy (with the LLM)

Feed the Consumption Map (4.1), the LinkedIn behavior patterns (4.2), the thought-leader analysis (4.3), and the engager signals (4.4) into the LLM and produce:

• The 3–4 content pillars for each Tier-1 advocate (narrow topics 360Brew can categorize, see Part 6).
• The angle hypotheses (grounded in what the ICP actually engages with, not guesses).
• The channel priorities (where the ICP actually is, often LinkedIn + a specific podcast/publication + YouTube, per the map).
• The language/keyword bank (the buyer's own words, for semantic match and message personalization).

Deliverable: the ICP Intelligence Brief → feeds Part 5 (sourcing) and Part 6 (content). Refresh quarterly.

PART 5 · TARGET LIST & COMMITTEE MAPPING

Goal: a scored account list with the buying committee mapped and reachable, built with an LLM (Claude, Gemini or Kimi) as the reasoning layer, not a scraper, and, for regulated clients, sourced only from legitimate origins. This is the system of record everything downstream uses.

5.1 Why the LLM, and how we use it

You do not build the target list by hand. You use an AI model (an LLM like Claude, Gemini or Kimi) as the brain that does the thinking, and you do the pulling.

Here is the split of work. You give the LLM two things: the ICP (who we are targeting) and the Intelligence Brief (what we learned about them in Part 4). The LLM then does four things. It tells you where to find these people. It writes the search strings. It cleans and organizes the results. And it scores each account for fit.

Your part is the pulling. Using the one Sales Navigator account and the sources the LLM recommended, you run the searches and collect the list. In short: the LLM reasons and structures, you gather.

5.2 The step-by-step sourcing process

Step 1, Write the sourcing prompt. Give the LLM: the ICP (qualifiers + disqualifiers + committee seats), the regulation class + data rule, the priority market, and the Intelligence Brief. Ask it to recommend where to source, tailored to this client's category and market.

Step 2, the LLM returns a sourcing plan across source types, for example:

• Accreditation / industry registries (e.g. for cyber: CREST, NACSA-certified lists; for finance: regulator/association member lists), highest trust, and legitimate for regulated clients.
• Events, conference/expo/summit attendee, speaker, and sponsor lists in the category.
• Business directories, Clutch, industry directories, chambers of commerce, trade associations.
• Content engagers, people engaging with our advocates' posts and with competitors' posts (warmest; ties to the harvest).
• Competitor followers, people following a competitor are in-market.
• Association / membership lists relevant to the vertical (e.g. PIKOM, MRCA, fintech associations).
• LinkedIn / Sales Navigator boolean searches, the LLM writes the boolean strings for the ICP titles + industries + geos + keywords.

Step 3, Pull the lists from those sources (Sales Nav search for the boolean strings; the registries/directories/event lists manually or via export). For REGULATED clients, only pull from legitimate, documented sources and record provenance per record (5.3).

Step 4, Clean + structure with the LLM. Paste the raw pulls; have the LLM dedup, normalize, and structure into one sheet with columns: company, name, role, LinkedIn URL, (WhatsApp where appropriate), business email, the pain/scenario signal, source/provenance, and LinkedIn activity status (active/recently posting vs dormant, see 5.4).

Step 5, Score A1–C with the LLM against the ICP rubric (fit × intent). The team works A/B seats first.

Step 6, Map the committee with the LLM: 2–3 seats per A/B account (champion, influencer, economic buyer). Single-threading kills consensus deals.

Step 7, Enrich + verify. Confirm LinkedIn URLs; verify WhatsApp numbers where used (bad numbers burn warm-up touches and trigger WhatsApp bans later). Keep sourcing legitimate per the data rule.

Step 8, QA a 10% random sample by hand: real ICP? Real contacts? Correctly graded? Legitimate source? Catch systematic errors cheaply before activation.

5.3 Data-source legitimacy (mandatory for REGULATED clients)

• Source personal/contact data only from: public professional profiles (LinkedIn), opt-in/consented directories, accreditation registries, event lists where contact is reasonable, and the person's own published business contact info.
• Record provenance for every record (where it came from).
• Do not use scraped or purchased databases of unknown origin for regulated clients, fintech/finance/health clients can be fined, and it exposes XpandEast.
• Respect PDPA / GDPR / sector rules; in Malaysia, business contact info is PDPA-exempt (lighter), but document anyway.

5.4 Prioritize ACTIVE prospects (important targeting rule)

When ordering who to pursue and send connection requests to, prioritize ICP people who are active / recently posting on LinkedIn, and deprioritize accounts that look dormant. Active people will actually see the advocate's content, engage, reply to DMs, and convert; dormant profiles absorb your daily limits and rarely respond. Tag activity status in the sheet (e.g. posted/engaged in last 30–60 days = active) and work the active ones first.

Deliverable: the scored, deduped, committee-mapped, provenance-tagged, activity-flagged account sheet (system of record).

PART 6 · CONTENT & CHANNEL ACTIVATION

Goal: content that 360Brew matches to the right buyers and that feeds the harvest. This section is rewritten for the June 2026 algorithm.

6.8 Diagnosing dead engagement (the #1 operator bottleneck)

The situation: an advocate posts on a specific, validated topic (other profiles went viral on it) for weeks, and engagement stays near zero. This recurs across our profiles. It is almost never a content problem, a validated angle proves the content is fine. It is a visibility, behavior, and network-fit problem. Work this checklist in order; the first two causes account for most cases.

1. Posting too often (the usual culprit). Daily, or more than once a day, cannibalizes each post and pushes the account toward throttled "publisher" distribution. Fix: drop to 3–5 posts/week, never more than once/day, spaced 18–24h+. Fewer, better, spaced.
2. Not engaging. The account just posts (and runs outreach) without genuinely commenting on others' content, so it has no inbound affinity and no warm audience to seed the post. Fix: 15–20 min/day of substantive comments on ICP and thought-leader posts before posting. This is the highest-value single change.
3. Dead-post decay has compounded. Weeks of dead posts cause reach decay that deepens if you keep posting through it. Fix: pause posting for 3–5 days, run 2–3 weeks of pure engagement (comments/saves on others, no posting or light posting), then resume at 3×/week. Recovery takes 3–4 weeks of consistent engagement, there is no instant fix.
4. The first-degree network is cold / non-ICP. Advocate accounts built for outreach are full of recently-added cold connections who don't engage, so the post never gets the initial dwell/comments 360Brew needs to expand it. Fix: build the network with real ICP and warm them via engagement (7.3); don't expect reach from a network of cold automated connections.
5. The account is throttled by the session/automation setup. If LinkedIn views the account as automated or compromised (the three-session trap, 2.5), it suppresses content reach and restricts actions. Fix: correct the session setup (2.5), one IP, one session, no phone concurrency, and reduce automation footprint.
6. Comment-to-like ratio too low. Reactions but no conversation (ratio below ~0.10) signals shallow content and gets demoted; a healthy ratio is ~0.15+. Fix: end posts with a real, expert-level question or a credible contrarian take that peers debate, and reply to every comment within ~30 min to build the 3+ person threads that earn ~5.2× amplification.
7. Mechanical suppressors. External link in the body (−40–60%), more than ~3 hashtags, engagement-bait phrasing, or templated/AI-filler content all suppress reach. Fix: link in first comment, ≤3 hashtags, no bait, specific/expert/in-the-buyer's-language copy (6.3, 6.5).

Do not respond to dead engagement by posting more, that is the one move that deepens the decay. Cut volume, raise engagement, fix the session setup, and give it 3–4 weeks.

PART 7 · WARM THE NETWORK & HARVEST

Goal: publish across the roster and turn engagement into warm leads. The harvest is the highest-ROI step and the one most operators skip, skipping it severs content from BD and wastes the entire content spend.

7.0 Getting accepted in cautious markets (the sideways move)

In markets that are allergic to sales, the real champion will almost never accept a cold connection from someone outside their network. This is part of the Month-1 warm-up you run on every LinkedIn profile deployed for the campaign. First you understand why you get rejected, then you use the sideways move to get inside the network the healthy way.

7.0.1 The sideways move: connect into sales/BD first, then the champion accepts

Because the champion won't accept cold, we do not go to the champion first. We connect sideways into the target company's sales and business-development people, whose instinct is to accept an unknown request (they assume you might be a lead for them). Once we are connected to their colleagues, we now share mutual connections with the champion, and the request reads as healthy networking, not spam.

7.1 Publish across the roster + find the winning advocate

• Tier-1 advocates post on-pillar; Tier-3 genuinely engage (no pods).
• Through Months 1–2, measure ICP-engagement per advocate. By end of Month 2, identify the winning advocate(s), whose content the right buyers engage with, and concentrate content and effort there (Part 1.1). Under 360Brew this is also when each advocate's semantic fingerprint starts to resolve.

7.2 THE ENGAGEMENT HARVEST (run within 48h of every post)

Why it matters: outreach tied to a recent trigger event, like engaging your post, lifts response rates by roughly a third, and under 360Brew consistent engagement between two people makes the model keep showing them each other's content (behavioral affinity), so harvesting compounds.

Daily harvest routine:

1. For each live post, pull everyone who engaged: reactions, comments, saves, reposts, CTA-comments/DMs.
2. Filter against the ICP (the LLM scores each engager fit/no-fit using the client rubric). Drop non-ICP.
3. Enrich + rank the ICP engagers (a substantive comment or a CTA comment is hotter than a like; cross-reference the "high-engagement ICP" list from Part 4).
4. Push to the CRM warm queue, each with: the post they engaged on, the engagement type, and a suggested ladder stage + opener context for messaging.
5. Hand the warm queue to conversation operators every morning.

Build pragmatically: even a manual version (export → the LLM ICP filter → CRM) delivers immediately. Automate the export/sync later with no-code. Don't wait for automation to start harvesting.

7.3 Engagement-as-warming (before any cold-ish outreach)

Before a conversation operator sends a connection request to a high-value target, have an advocate genuinely engage that target's content (a substantive comment, a profile visit). Under 360Brew this builds behavioral affinity so the target starts seeing the advocate's posts, so the later connection request and DM land warm, protecting acceptance/reply rates and account health.

7.4 The amplification flywheel

After a meeting, the advocate publishes an on-pillar post drawn from the insight and tags the prospect (sparingly). The prospect engages, exposing the client to their network and surfacing new ICP people who re-enter the harvest.

Deliverable: the daily warm-lead queue in the CRM, with context attached.

PART 8 · VALIDATE MESSAGING & CONVERT

Goal: walk warm leads up the ladder to a booked meeting, then hand off. Fixes the messaging slop, the channel discipline, and the WhatsApp ban problem. The full click-level procedure, with every connection note, DM, follow-up, WhatsApp transition as copy-paste assets, are in Part 17 (SOP-B), including what to do when the prospect has no content to engage with.

8.1 Kill the AI slop — anchor every message in five inputs

The slop comes from the team free-typing into a general chat with no strategy and no memory of the content the prospect engaged with, and generic outreach that matches spam patterns is now actively flagged by LinkedIn, so slop is also an account-safety risk. Fix it with a simple rule: no message goes out unless it is anchored in the five things below, so the strategy is baked in, not improvised.

The rule — every message must be built from all five of these:

1. The prospect + the exact post/asset they engaged with (from the harvest).
2. The active winning angle for this client/market.
3. The give-to-get offer.
4. The ladder stage + the market's channel rule.
5. The advocate's voice profile.

Forbidden: pitch in the opener; meeting ask before value; identical text across prospects; ignoring the engagement context; connection notes over 300 chars.

Build it as a locked LLM project (call it "XpandEast Outreach") with the playbook, the voice rules, the phrases-to-avoid list, and the five-input rule (8.1). The operator drafts from the five inputs so the message follows the strategy, then edits and sends it. Gate 3 still applies: you always review before sending. This makes good strategy the default and removes the slop.

8.2 The ladder (never skip a rung)

1. Rung 1. Advocate genuinely engages the prospect's content (warming), see 7.3
2. Rung 2. Connection request plus a light personalized note (under 300 chars)
3. Rung 3. After they accept: a value-first DM referencing the post they engaged with, plus the give-to-get
4. Rung 4. Peer-level conversation (substance, no pitch), they set the pace
5. Rung 5. Earned WhatsApp (SEA/Gulf), move here once they have replied or engaged
6. Rung 6. Human-recorded voice note to deepen warmth
7. Rung 7. Book the call, confirm, then hand off to the client

Benchmarks: a connection note + then a DM lifts post-acceptance reply meaningfully; a profile visit + DM pushes reply rates higher; outreach tied to a trigger event (their post engagement) lifts response ~32%; sequenced follow-ups 2–5 business days apart lift conversion ~49%, it takes ~5 contacts on average, yet ~48% of reps never send a second message. Always follow up. Good B2B reply rates: 10–25% average, 25–35% strong.

8.3 Connection request SOP

• Prioritize active prospects (recently posting/engaging), they see content, reply, and convert; dormant profiles waste your daily limit (5.4).
• Target 40–60% acceptance via tight targeting + warming (7.3), not spraying.
• Note ≤300 chars, light and relevant (reference their content/context); no pitch.
• 15–20/day per account; withdraw unaccepted after 2–3 weeks; keep pending under 500.

8.4 The DM conversation — see Part 17

The first DM (the opener) is in Part 18. Everything after they reply — exactly how you chat your way to a booked call — is the single linear flow in Part 17. Do not run a separate sequence here: Part 17 is the one source of truth for the conversation, so it can never conflict with itself.

8.5 WhatsApp operation (fixing the ban / SIM-churn problem)

Why numbers die: WhatsApp scores accounts for spam; the top triggers are messaging people who haven't saved your number (weighted heavily) and blocks/reports, as few as 3–5 reports can suspend a number. Virtual-number/eSIM ranges are often pre-flagged, and clustering accounts on one device/IP compounds it. Our numbers die because they're used cold. The fix is the method, not better SIMs.

Rules:

1. Earned-only. Only message people who already engaged/replied on LinkedIn and expect you. High replies + near-zero reports = numbers survive. This one rule fixes most churn.
2. Warm a new number 2–3 weeks before business use: don't scan the QR for 24h+ after registering; set a real photo/name/bio; message 5–10 genuinely warm contacts, spaced 2–4 hours, and receive replies; build real saved contacts; ramp slowly.
3. ~50 conversations/day/number ceiling even warmed, but earned warm conversations stay well under it.
4. Don't cluster, separate device/network fingerprints; don't run the fleet off one wifi/IP.
5. App for warm 1:1, not the API. The Business Cloud API sends template messages that read as a blast, wrong tool for warm conversation. The human voice notes that convert so well are app-only, human behavior, keep them, recorded by a human, in earned conversations.
6. Never modified clients (GB WhatsApp etc.) or grey-market bulk tools, instant detection.

8.6 Cold email and cold calling, verdicts

• Cold email: not a core pillar. Low SEA response in our experience; restricted in Australia (Spam Act) and consent-gated across APAC. Don't fund Smartlead-at-scale, it tunes the wrong thing for a warm motion. Narrow use only: in Malaysia (business email PDPA-exempt), a low-volume, hyper-personalized email as a 2nd/3rd touch on an already-warm contact. No cold blasting.
• AI cold calling: no. DNC/telecom rules apply; an obviously-AI cold call to a SEA enterprise buyer is brand-negative; and our own data shows meetings come from LinkedIn→WhatsApp. The voice advantage we want is human WhatsApp voice notes in earned conversations, scale that.

8.7 The handoff

On a confirmed meeting: capture context (who, role, pain, the post/thread that warmed them), confirm the invite, hand to the client with a short brief, log it, and feed the amplification flywheel (7.4).

Deliverable: booked, confirmed meetings + handoff briefs.

PART 9 · TRACK & RECALIBRATE

Goal: show the numbers in the locked format, read the outlier, reconfigure next month. Also the renewal engine.

9.1 Assemble the month's data

From the CRM/sheet: content published per pillar/advocate, impressions, ICP-weighted engagement (saves, substantive ICP comments, thread depth, DMs), warm leads harvested, conversations, meetings (split end-customer vs partnership), newsletter growth, pipeline. Under 360Brew, weight saves + thread depth + ICP-comment quality over raw impressions.

9.2 Read the outliers (content AND advocate)

State explicitly: which pillar/angle/format converted, and which advocate the ICP engaged with (the Month-2 winner, refreshed). These are the recalibration signals.

9.3 Honest assessment

Keep the "what underperformed + the specific fix" section, it builds trust and drives renewals.

9.4 Recalibrate next month's config

Update the config sheet: pillar/angle changes, advocate concentration, channel shifts. Every report ends with an explicit config change so next month isn't a repeat.

9.5 KPI scorecard

Score against the agreed monthly target (the 5–10 calls). Show the funnel: impressions → ICP engagement → warm leads → conversations → meetings (end-customer vs partnership).

Deliverable: the monthly client report (your locked HTML→PDF stack). → Gate 4.

PART 10 · ACTIVITY LOGGING (the two logs that make conversion legible)

You cannot improve a funnel you don't measure. Two logs run continuously and feed everything downstream, the weekly sync (Part 11), the monthly report (Part 9), and the recalibration. Without them, "we're busy" never becomes "here's the rate, here's what to scale," and you can't tell whether WhatsApp or LinkedIn is actually producing meetings.

10.1 Why logging is non-negotiable

Conversion only becomes visible when you count the funnel per channel. Example: 100 first-contact WhatsApp DMs → 4 positive replies → 1 meeting tells you WhatsApp converts at ~4% reply and ~1% meeting, so to land 5 meetings/month on WhatsApp you need roughly 500 DMs/month through it. Run the same math for LinkedIn and you instantly see which channel deserves the effort, and how much volume each monthly target actually requires.

Logs turn activity into rates, and rates into decisions.

10.2 Log A, The Outreach Funnel Log (GTM executive, daily)

Who: every GTM executive logs their own outreach, every day, by channel. This is mandatory, the conversion picture is only as good as the logging discipline.

What to log (per channel, per day): connection requests sent, first-contact DMs sent, positive replies, conversations opened, meetings booked, tagged by channel (LinkedIn vs WhatsApp) and by source (cold list vs warm-from-content).

Weekly/monthly roll-up computes the rates per channel:

• Reply rate = positive replies ÷ DMs · Meeting rate = meetings ÷ DMs (and ÷ positive replies).
• The speculated flow per channel, e.g. WhatsApp: 100 DMs → 4 replies (4%) → 1 meeting (1%) vs LinkedIn: [its own numbers], so you allocate effort to the higher-converting channel and back-calculate the DM volume needed to hit the monthly meeting target.
• Segment cold-list vs warm-from-content. Warm-from-content should convert several times better, this is the data that proves the content→BD funnel works and justifies the harvest.
• Safety check: the log also exposes if anyone exceeded 15–20 connection requests/day or 50 first-contact DMs/day per account (account-health risk).

10.3 Log B, The Content Traction Log (content team, weekly)

Who: the content team logs every piece created, and reviews traction at the end of each week.

What to log per post (advocate, pillar, format, angle, link) and the engagement signals. Under 360Brew, weight the meaningful signals, comments (especially substantive ones and thread depth), reposts, and saves, and treat raw likes/impressions as vanity.

End-of-week review:

• Flag the winners by comments/reposts/saves (not impressions).
• Feed winners into: (a) double-down content next week, (b) the harvest (their engagers → warm queue), (c) the winning-advocate read (Part 1.1 / 7.1).
• Feed flat content into: kill or rework the angle.
• This makes the through-line explicit: traction → engagers → warm queue → meetings.

10.4 The two logs are one funnel

The monthly report and weekly sync stitch them together: "content produced X engagement → Y warm leads harvested → Z DMs → W meetings." This is also your hardest proof of content ROI, showing warm-from-content outreach converting at a multiple of cold-list outreach.

PART 11 · CLIENT COMMUNICATION (the weekly sync + the monthly report)

How you communicate is half the renewal. The rule is simple and absolute: lead with outcome, not activity. The cadence and framing then shift as the engine matures from content to pipeline.

11.1 The cardinal rule: outcome first, activity as evidence

Clients don't buy "we sent 400 DMs and posted 12 times." They buy trust, visibility, and meetings. Every sync and every report opens with the outcome, then shows the activity as the evidence behind it, then what's next. Structure always: Outcome → the activity that produced it → next.

Translate activity into the client's language, every time:

• Content = trust + visibility. Always frame content as building trust and visibility with the right buyers, never "we posted." Repeat this framing relentlessly so the client values the content work before meetings appear.
• Pipeline activity = sales calls + positive replies + newsletter subscribers being warmed from content. Frame outreach/engagement outcomes as pipeline: booked calls, positive replies, and the owned audience (newsletter subscribers) being nurtured toward becoming buyers.

11.2 The weekly sync flow

Cadence: a weekly call (or async update). Always lead with the outcome.

The shift by month:

• Month 1, mainly CONTENT activity. No meetings promised in Month 1; trust and meetings compound from Month 2. The sync focuses on trust + visibility being built: content live, advocates warming and posting on-pillar, early engagement signals from the ICP (comments/saves), newsletter launching. Frame: "We're building your trust and visibility with [the ICP], here's the early traction and who's engaging."
• Month 2 onward, PIPELINE activity + CONTENT activity. Now you share-screen the pipeline: warm leads harvested, conversations in flight, positive replies, meetings, newsletter subscribers warming, plus the continuing content traction. Frame: "Here are your conversations and meetings (pipeline), and the content engine still building visibility behind them."

Share-screen in Month 2: walk the client through the live pipeline (the CRM/sheet), the warm queue, conversations, meetings booked, subscribers, so they see pipeline activity, not just hear it described.

Weekly sync agenda (in lead-with-outcome order):

1. Outcome headline. Month 1: visibility/trust traction. Month 2+: meetings + positive replies + subscribers warming.
2. Pipeline view (Month 2+): share-screen the CRM, warm leads, conversations, meetings, subscribers.
3. Content traction: the winning posts this week (framed as trust/visibility), with the engagement signals that matter (comments/reposts/saves).
4. The connection: how content engagement is feeding the pipeline (engagers → warm queue → conversations), this is the story that makes the whole retainer make sense.
5. What's next: next week's focus, and any client input needed (approvals, warm intros, attending a booked meeting).

11.3 The monthly report (formal version, ties to Part 9)

The monthly report (the Part 9 deliverable, in your locked HTML format) is the formal version of the same outcome-led story: outcomes (meetings split end-customer/partnership, positive replies, subscribers, ICP-weighted engagement) → the activity evidence → the outlier read (winning pillar/angle/advocate) → next-month config. → Gate 4 before it goes to the client.

11.4 What the client should always hear

• Month 1: "We're building trust and visibility with your buyers." (content = trust + visibility)
• Month 2+: "Here's your pipeline, the calls, the positive replies, and the audience we're warming, plus the visibility engine behind it."
• Always: outcome first, activity as evidence, and the through-line that content visibility is what feeds the pipeline. Never let a sync become a list of tasks.

PART 12 · THE OPERATOR'S ROUTINES (keep open while working)

Daily routine (about 35 to 45 minutes per account)

Do these in order every working day, for each client. The whole point of the order is that each step feeds the next.

Step 1. Check account health (5 min)

Open your one-row-per-account tracking sheet and check each advocate account:

• ☐ Does it log in cleanly, with no CAPTCHA, warning, or "verify your identity" prompt?
• ☐ Is the connection acceptance rate above 35 to 40%?
• ☐ Are pending (unanswered) connection requests under 500?

If any account fails, stop and fix it before you use it (see Troubleshooting, Part 15). A fried account cannot be undone, so this check comes first every day.

Step 2. Harvest yesterday's engagers (5 to 10 min)

• ☐ Pull the list of everyone who liked, commented on, or saved an advocate post yesterday.
• ☐ Run that list through the LLM to keep only the people who match the ICP.
• ☐ Add the matches to your warm queue. (Full method in 7.2.)

These people already engaged, so they are your warmest leads. This is where most meetings come from.

Step 3. Warm up your high-value targets (5 min)

• ☐ For the top accounts you want, have an advocate genuinely engage with the target's own posts first (a real comment, a profile visit). This makes your later message land warm, not cold. (Full method in 7.3.)

Step 4. Work the warm queue (10 to 15 min)

• ☐ Send connection requests, 15 to 20 per account, and prioritize people who have been active recently.
• ☐ Send value-first DMs that reference the exact post the person engaged with. Never pitch in the first message. Keep to a maximum of 50 first-contact DMs per account.
• ☐ Send any follow-ups that are due (space them 2 to 5 days apart). Most operators skip follow-ups, and following up is the single biggest lift in replies, so do not skip them.
• ☐ Advance any WhatsApp conversations you have earned. Use a human voice note where the relationship is already warm.

Step 5. Publish and engage (10 min)

• ☐ Publish the day's on-pillar post for each posting advocate.
• ☐ Reply properly to comments to build real threads (aim for three or more people going back and forth). Do not judge a post by its first hour.

Step 6. Log everything (5 min)

• ☐ In the Outreach Funnel Log (Part 10), record what you sent: connection requests and first-contact DMs by channel (LinkedIn or WhatsApp), plus positive replies, conversations, and meetings. Tag each as cold-list or warm-from-content.
• ☐ Move any booked meetings to handoff.

Weekly routine

• ☐ Prepare next week's content, on-pillar, from the Intelligence Brief and the winning angle. Send it to the client for approval (Gate 2).
• ☐ Withdraw any connection requests older than 2 to 3 weeks, across all accounts (this keeps the pending count healthy).
• ☐ Refresh your target list in the LLM with new engagers and new buying signals.
• ☐ Run a team sync on what is working: which topic, angle, and advocate is converting.
• ☐ Review the Content Traction Log (Part 10). Flag the week's winning posts by comments, reposts, and saves, not by impressions. Double down on the winners and rework the flat ones.
• ☐ Run the weekly client sync. Lead with the outcome. In Month 1 that is trust and visibility; from Month 2 on, share-screen the live pipeline and content (Part 11).
• ☐ Skim the thought-leader and engager tracking lists from Part 4 for fresh angles that are landing.

Monthly routine

• ☐ Produce the monthly report (Part 9), get it signed off (Gate 4), and send it to the client.
• ☐ Recalibrate the config for next month: the topic, the angle, which advocate to concentrate on, and the channel mix.
• ☐ Review the roadmap against what you promised for the month.
• ☐ Review the advocate roster: are you concentrating on the winner, and is any account at risk?

Quarterly routine

• ☐ Refresh the ICP Intelligence Brief (Part 4): re-run SparkToro, re-mine LinkedIn behavior, and update the topics and keywords.
• ☐ Re-verify the platform limits and benchmarks in this manual, since LinkedIn and WhatsApp change them.

PART 13 · TOOL STACK & WHERE THE AGENTS PLUG IN

Build the agents in this order (ROI): (1) Engagement-Harvest, (2) constrained Messaging, (3) the LLM sourcing + Intelligence synthesis, (4) Reporting. Never automate the four gates.

PART 14 · ERROR-PROOFING PRE-FLIGHT (the 5 seams)

• ☐ Seam 1 (bad list): ICP has disqualifiers; sourced with an LLM from legitimate sources (provenance recorded for REGULATED); scored A1–C; active prospects flagged; committee mapped; 10% QA done.
• ☐ Seam 2 (off-target content): Part 4 intelligence done first; content on locked 3–4 pillars; built for 360Brew (specific/expert/save-worthy/thread-starting, in the buyer's language); no generic AI, no bait, no pods; native vertical for video reach.
• ☐ Seam 3 (messaging slop): messages built from all 5 inputs (8.1); tied to the engaged content; no pitch in opener; follow-ups scheduled; ≤50 first-contact DMs/day.
• ☐ Seam 4 (account death): 1 account = 1 profile = 1 fixed proxy; warming ramp followed; 15–20 conn/day; no InMail; Leadspicker on warmed accounts only; WhatsApp earned-only on warmed numbers; no grey-market tools.
• ☐ Seam 5 (content↔BD disconnect): harvest runs within 48h of every post; warm queue lands with BD daily; content KPIs and BD KPIs are one funnel.

PART 15 · OPERATOR TROUBLESHOOTING (when your flow stalls)

These are the problems that stop an operator mid-flow. Each one has a fast diagnosis and the immediate fix. When something breaks, come here first.

15.1 "I post a validated topic daily for weeks and engagement is near zero."

The most common stall. It is a visibility/behavior/network problem, not content. Immediate moves: cut to 3–5 posts/week, never >1/day, spaced 18–24h; add 15–20 min/day of substantive comments before posting; if posts have been dead for weeks, pause 3–5 days and run 2–3 weeks of pure engagement before resuming; verify the session setup (2.5) isn't throttling the account. Full checklist: 6.8. Do not post more to fix it, that deepens the decay.

15.2 "A LinkedIn advocate account is getting CAPTCHAs, forced logouts, or a 'verify your identity' prompt."

Almost always the session setup or volume, not the content. Immediate moves: stop all automation for 1–2 weeks; use the account only through its single GoLogin session (one IP); complete the verification; resume manually, then re-enable automation at ~50% of prior volume. Check for the three-session trap (GoLogin + Leadspicker + phone all live, 2.5). Changing the password does not fix it. Take a screenshot of any warning for support.

15.3 "Reach suddenly dropped about 50% across the LinkedIn advocate accounts."

Usually a platform recalibration plus a behavior issue. Immediate moves: confirm you're not posting daily; confirm engagement time is happening; check for dead-post decay (15.1); confirm no automation/session flags (2.5). Reach now favors saves, comment depth, and niche relevance over breadth, shift content toward save-worthy, debate-starting, pillar-true posts (6.3).

15.4 "Connection acceptance rate is low (under ~30%)."

Immediate moves: tighten targeting to active, in-ICP prospects (5.4); warm them first via advocate engagement (7.3); add a light personalized note; reduce daily volume (15–20/day, conservative for our sub-40-SSI accounts); withdraw pending requests older than 2–3 weeks so the pending pile (a low-acceptance signal) shrinks. Low acceptance compounds into restrictions, so fix it before scaling.

15.5 "WhatsApp numbers keep getting banned."

You're using WhatsApp cold. Immediate moves: make WhatsApp earned-only (message only people who already engaged/replied on LinkedIn); warm new numbers 2–3 weeks before use; don't cluster numbers on one device/IP; use the app for warm 1:1, not the API. Full rules: 8.5. The fix is the method, not better SIMs.

15.6 "The target list stalled, most accounts have no named decision-maker."

This is a data-ops / enrichment task, not more searching (≈70% "Unknown" is the normal state of a raw list). Immediate moves: run the bare company names through the enrichment stack to fill decision-maker, role, and contact, then promote the ones that qualify (Part 5 / 5.2). Don't keep re-running searches; enrich.

15.7 "We're getting engagement on content but no meetings."

The harvest isn't running, or BD is ignoring the warm queue (Seam 5). Immediate moves: within 48h of every post, capture engagers → filter to ICP → push to the warm queue with the post context → BD works it the next morning (Part 7 / 7.2). Content KPIs and BD KPIs are one funnel; if they're split, meetings leak.

15.8 "Messages keep coming out as generic AI slop / different from the content."

The operator is free-typing into a general chat with no strategy or post context. Immediate moves: build every message from all five inputs (the engaged post, the winning angle, the offer, the ladder stage/channel, the voice), with the forbidden list enforced (8.1). Generic outreach is also now flagged by LinkedIn, so this is a safety fix too.

15.9 "Content is stuck waiting for client approval and publishing has stalled."

Do three things right away.

First, confirm who approves content and how fast. At kickoff you agreed on a named approver and a 24-hour approval window (Gate 2, see 3.3). Remind them of it.

Second, send content for approval in batches. Send a full week at once, not one post at a time.

Third, keep a small buffer of already-approved posts ready. That way a slow approval week does not stop publishing.

15.10 "A LinkedIn advocate account got restricted, should we spin up a new one?"

No. A new account from the same environment gets linked and re-flagged, and creating accounts to dodge restrictions violates LinkedIn's terms. Immediate moves: fix the environment (the proxy/session/fingerprint, 2.2 / 2.5), run the recovery in 15.2, and let the account age back. The fix is the setup, not another signup.

15.11 "Low reply rate despite sending 100+ DMs a day."

The problem is the messaging, not the volume. A healthy reply rate is 10 to 25%. A strong one is 25 to 35%. If you are under 10% even at high volume, your messages are landing as spam. They are too cold, too generic, or they pitch too early.

Fix it in this order:

1. Stop cold-blasting. Warm the target first by having an advocate engage their content (see 7.3).
2. Use the DM flow (Part 17): give value, ask one question, and never pitch in the first message.
3. Make every message personal and relevant. Reference their specific post or situation.
4. Read each message out loud before sending. If you would not reply to it, do not send it.

One more thing: high volume with bad messaging also gets the account restricted. So slowing down and personalizing fixes your reply rate and protects the account at the same time.

15.12 "No-show rate is over 30%."

There are two causes, and both are fixable.

Cause 1: a weak reminder cadence. A meeting booked days ago, with no contact in between, gets forgotten. Fix it with a set sequence of reminders: a confirmation the moment you book, a helpful reminder the day before, and a short WhatsApp message the morning of the meeting. This is exactly why you moved them to WhatsApp in 17.5.

Cause 2: a weak offer. If the reason for the meeting was vague, like "a quick chat," the interest cooled and they deprioritized it. Fix it by tying the meeting to a specific outcome they actually wanted, based on the pain they admitted in the conversation.

If no-shows stay high, the booking itself was soft. Go back to the segue in 17.5: you should book off a real, stated pain, not just a polite yes.

PART 16 · ADVOCATE ACCOUNT, ZERO TO READY (deep SOP)

What this is. The complete, do-this-exactly procedure to take one advocate from "real person who agreed to let us run their profile" to "campaign-ready account," and then run it daily without getting it throttled or banned. Written so a new operator with zero context executes it without asking a question.

Owner: the operator. Time: ~2 hours of setup work, then a 30-day warm (≤30 min/day), then the daily operating routine (~35–45 min/account/day). Done-state: defined in A7.

Why this is step one of everything. Every result in the engine sits on healthy accounts. The two things that kill accounts, wrong IP/session setup, and posting-without-engaging, are both fixed here.

A0 · Prerequisites (have these before you touch LinkedIn)

1. A real, consenting person. Real name, real face, a genuine (even if junior) professional background in or adjacent to the client's category. Signed consent to operate the profile. Never a fabricated identity or a bought/aged account, both get detected and destroy the trust we sell.
2. The advocate's real city (for the proxy geo) and rough timezone (for activity hours).
3. One topic pillar for this advocate (from Part 4 intelligence), the single narrow subject this profile will be known for (e.g. "retail customer-data / loyalty in Malaysia"). Everything the account posts and comments on serves this pillar.
4. Assets: a real headshot, a simple branded banner, and 60–90 minutes of the advocate's time for the photo/voice/About and any identity verification.
5. A payment method for the proxy (~$3–8/account/month) and the GoLogin seat.

A1 · Buy and configure the proxy (the #1 thing operators get wrong)

A1.1, Pick the right type. For an account-tied LinkedIn profile (signed-in, posting, messaging, long-term), the IP must be stable and look like a home/office line.

• Default: static-residential / ISP proxy. A fixed IP that LinkedIn reads as a real consumer connection, held for the life of the account. This is the right choice for steady-state operation.
• Hardest / highest-value accounts and the warm-up window: mobile (4G/5G). Mobile IPs are shared by thousands of real users via carrier CGNAT, so LinkedIn almost never blocks them. More expensive; use for the accounts you most need to protect.
• Never use: datacenter proxies (flagged on sight), free/shared proxies (pre-blacklisted, and they leak your real IP), or rotating proxies for a logged-in account (an IP that changes mid-session looks like account hijacking and triggers checkpoints).

A1.2, One proxy per account, sticky, geo-matched. 1 IP : 1 account, never shared (a shared IP is the clearest signal for LinkedIn to mass-scan and lock the whole roster). Use a sticky session (or static/ISP, which is inherently fixed). Set the proxy's country and city to the advocate's real city and never move a live account's geo.

A1.3, Provider + 2026 cost (any of these work; pick on budget):

• DataImpulse, value pick. Residential ~$1/GB, mobile ~$2/GB, 195 countries, per-account sticky via a sessid in the username, city/ASN targeting. Traffic doesn't expire.
• IPRoyal, longest sticky sessions (up to ~7 days), ~$7/GB.
• SOAX, strong mobile+residential mix, ~$3.60/GB.
• NetNut / Proxy-Seller / Decodo, solid static-ISP for long-term nurturing (Decodo static ISP from ~$0.27/IP).
• Budget ~1–3 GB/account/month for our usage (signed-in browsing + light automation is low-bandwidth).

A1.4, Configure the sticky IP (example, DataImpulse). Host gw.dataimpulse.com, port 823 (HTTP) or 824 (SOCKS5). In the username, pin country + city + a unique session id per account:

username:  YOUR_LOGIN__cr.id;city.jakarta;sessid.advocate07
password:  YOUR_PASSWORD

The unique sessid.advocateNN is what gives that account its own fixed IP. Use a different sessid for every advocate.

A1.5, Verify the IP is clean BEFORE logging in (do not skip).

1. Load the proxy in the GoLogin profile (A2), then open https://ipinfo.io (or similar). Confirm the city and country match the advocate's and the ISP reads as a consumer ISP (not a hosting/datacenter name).
2. Run a WebRTC leak test (browserleaks.com/webrtc). If it shows your real IP or ISP, the proxy isn't covering all traffic, fix it (enable WebRTC masking in GoLogin) before proceeding.
3. Dirty-IP check: a proxy IP abused by someone before you is pre-flagged. If the very first LinkedIn load throws an immediate CAPTCHA/checkpoint on a brand-new clean profile, the IP is likely dirty, request a new sticky IP (new sessid) and retest. Don't try to push through a dirty IP.

A2 · Create and configure the GoLogin profile

The proxy hides the IP; GoLogin isolates the device fingerprint (Canvas, WebGL, fonts, hardware). You need both, they're separate signals, and one without the other still links/flags the account.

Steps (exact):

1. GoLogin → New profile. Name it by convention: LI, [Client], [AdvocateName], [City] (e.g. LI, Loyalytics, Sarah, Jakarta).
2. OS / fingerprint: pick a common real configuration (e.g. Windows + Chrome) and leave it. Do not regenerate or edit the fingerprint after creation, changing it later looks like a new/unstable device.
3. Proxy tab: set type (HTTP or SOCKS5), then paste host, port, username, password from A1.4. Click Check proxy / Test, confirm it connects and shows the right city.
4. Timezone & geolocation: set to match the proxy's city (GoLogin can auto-match to the IP, confirm it does). A mismatch (IP in Jakarta, timezone in London) is a flag.
5. Save. Open the profile, run the A1.5 verification (IP + WebRTC + dirty-IP check).
6. One advocate = one GoLogin profile, forever. Never log two advocates into one profile, and never log one advocate's account into a different profile/proxy.

A3 · First login and profile build (to 100%, on-pillar)

A3.1, Login discipline. Open LinkedIn inside the verified GoLogin profile only. If it's a brand-new account, don't fire any actions for the first day, just complete the profile and browse 5–10 minutes. Acting within minutes of account creation is a classic bot signal.

A3.2, Complete the profile to 100% (this is a reach and trust signal, not vanity).

• Photo: the advocate's real headshot. Banner: simple, on-pillar (not a logo dump).
• Headline: name the pillar in plain keywords the buyer would search/recognize, e.g. "Helping Malaysian retailers turn customer data into revenue · Loyalty & CRM." 360Brew reads this to categorize the account's expertise, so the pillar words matter.
• About: 3–4 short paragraphs in the advocate's real voice, on-pillar, first-person, with the buyer's language (from the Part 4 keyword bank). No corporate filler.
• Experience: at least 3 real entries. Custom URL: set linkedin.com/in/firstname-lastname. Skills: add the pillar's core skills.
• Featured: once you have a lead magnet, pin it here.

A3.3, The pillar lock. Write the 3–4 sub-topics this account will post on and stick to them 80%+. It takes ~90 days of consistent on-topic posting for LinkedIn to recognize the account's expertise (the "semantic fingerprint"), which is why reach is low at the start and why drifting off-topic resets the clock.

A4 · The 30-day warm-up (day by day, do exactly this)

The first ~30 days are LinkedIn's behavioral-modeling window. No automation in weeks 1–2. Ramp like a human; one session at a time; activity only during business hours in the proxy's timezone; take one day off each week. If acceptance drops below ~35% or you see CAPTCHAs/verification, stop increasing and reduce (A6/A8).

Throughout: randomized gaps between actions (not mechanical), no bursts (don't send all requests at 9am), weekends at 30–50% of weekday volume, withdraw any request unaccepted after ~2–3 weeks.

A5 · Session rules (the three-session trap, re-stated because it's the top killer)

One account = one consistent IP = one session at a time.

• The trap: GoLogin profile + a separately-logged-in cloud tool (e.g. Leadspicker on its IP) + the advocate's own phone = three sessions from three IPs = "session jumping" = a primary bot/compromise flag (and it suppresses content reach too).
• Safe: run automation as a browser extension inside the GoLogin profile (one session, one IP). If you must use a cloud tool, give it the same dedicated proxy as this profile and never run it while the GoLogin browser or the phone is active. Close one session before opening another.
• Phone: ideally the advocate doesn't use their phone for the account; if they do, it's from the same city as the proxy (it is, proxy = their city), light, and never concurrent with automation or the desktop session.
• Changing the password does not reset risk, LinkedIn identifies the environment (IP + device + fingerprint), not the credentials.

A6 · Definition of "campaign-ready" (the gate to start outreach)

Do not start real outreach until all are true:

• ☐ Profile 100% complete, pillar-locked headline/About, custom URL.
• ☐ Account aged ≥30 days with the warm-up above.
• ☐ Acceptance rate ≥35–40% over the last 2 weeks.
• ☐ No open CAPTCHAs, verification prompts, or feature restrictions.
• ☐ Posting 3–5×/week on-pillar with a daily engagement habit established.
• ☐ One stable session (in-profile automation, phone controlled).

If any box is unchecked, keep warming, starting outreach on an unready account is how the fleet gets restricted.

A7 · The daily operating routine (per campaign-ready account, ~35–45 min)

Run in this order. One session, business hours, randomized pacing. (For a pod of 8–12 accounts, batch the same block across accounts.)

A7.1, Health check (2 min). Open the profile. Any CAPTCHA / "verify" / restriction banner? Acceptance % still healthy? Pending requests under 500? Log it in the one-row-per-account health sheet. If anything's off → A8.

A7.2, Engagement block (15–20 min, the highest-value thing you do). Before posting anything. Comment substantively on 5–8 on-pillar / ICP / thought-leader posts (find them via: your pillar hashtags, the thought-leader tracking list from Part 4, your ICP feed, and notifications). This primes your own reach and warms target prospects so later connection requests land warm (it's also how you "engage before you post"). See A9.4 for the comment bar.

A7.3, Connection block (10 min). Send 15–20 connection requests, prioritizing active, in-ICP people (posted/engaged in the last 30–60 days, skip dormant profiles), ideally people you just engaged with in A7.2. Light personalized note, no pitch. Withdraw requests older than ~2–3 weeks.

A7.4, Conversation block (10–15 min). Work the warm queue (people who engaged your content / accepted + replied): value-first DMs tied to the exact post they engaged with, follow-ups due (spaced 2–5 days), advance earned-WhatsApp conversations. Max 50 first-contact DMs/day. Log outreach by channel in the Outreach Funnel Log.

A7.5, Posting (only on posting days, ≤1/post/day). Publish the on-pillar post (link in first comment if any), then reply to every comment within ~30 min to build the 3+ person threads that earn amplification. Don't judge the post by hour one.

A8 · Failure modes and the immediate fix (anticipate these, they stop operators cold)

A9 · Reference

A9.1, Proxy decision in one line: static-residential/ISP for steady-state, mobile for warm-up and high-value accounts, sticky, one per account, geo = advocate's city. Never datacenter, free, or rotating.

A9.2, The numbers, locked: 15–20 connection requests/day · ≤50 first-contact DMs/day · 3–5 posts/week (never >1/day, spaced 18–24h) · 15–20 min/day engagement · acceptance ≥35–40% · pending <500 · withdraw at 2–3 weeks · one session at a time.

A9.3, One-page setup checklist:

• ☐ Real consenting advocate, pillar assigned, city + timezone noted
• ☐ Static-residential/ISP (or mobile) sticky proxy bought, geo = advocate city, unique sessid
• ☐ GoLogin profile created, fingerprint set and locked, proxy attached, timezone matched
• ☐ IP verified clean (ipinfo + WebRTC leak test + dirty-IP check)
• ☐ Profile built to 100%, pillar-locked headline/About, custom URL
• ☐ 30-day warm-up run per A4; campaign-ready gate (A6) passed

A9.4, What a "substantive comment" looks like (the bar for A7.2): The point is to add a specific, on-pillar thought that invites a reply, not to be seen. LinkedIn now reads comment length, relevance, and whether you're in the author's graph, and 3+ person back-and-forth threads get the biggest amplification.

• ❌ Anti-pattern (hurts you, pattern-matched low-effort comments are detected): "Great post! Thanks for sharing 🙌"
• ✅ Substantive (adds a specific angle + a question): "The part about loyalty data sitting unused matches what we see in MY grocery, most chains can identify maybe 20% of transactions. Curious whether you'd start with basket-level tagging or with a single-customer view first?"
• ✅ Substantive (respectful contrarian, invites debate): "Agree on the diagnosis, but I'd push back on leading with discounts, in this market the faster win is fixing identification before you spend on offers. Have you seen it work the other way round?"
• ✅ Substantive (adds a local data point): "Worth adding the regional angle: in ID the same problem is worse because cash-heavy baskets never tie to a profile at all. Changes where you'd put the first dollar."

PART 17 · THE DM FLOW (after they reply → booked call)

17.1 The shape of every reply

Every message you send is built from the same pieces. Keep it to two or three lines — never a paragraph.

17.2 What "value" actually is (so you never have to guess)

"Add value" is vague, so here it is spelled out. Value is one of these, tied to what they just said:

Not value (do not send these): "Let me know if you need anything." "We help companies like yours." A calendar link. A feature list. They give the person nothing to reply to.

17.3 The questions that move a chat forward

End every message on one of these. They get the prospect to say the problem themselves, which is what makes the call feel like their idea.

17.4 A whole chat, start to finish

17.5 The segue — turning the chat into a booked call

You earn the ask once they have named a real problem and you have given real value. Not before — never on message one or two. The move ties the call to the specific thing they said:

"Sounds like this is worth 20 minutes properly rather than over DM. I can walk you through what [similar company] did about [the exact thing they raised] — want me to grab a time?"

Offer a call, not a demo. Keep it 20 minutes. If they hesitate, do not push — give one more piece of value and ask again next message.

17.6 When they go quiet

17.7 Objections — stay peer-to-peer

17.8 What each reply means → the Infinity stage

17.9 The numbers, locked

Reply rate 10–25% is healthy · ≤50 first-contact DMs per day per account · follow-ups spaced 2–5 days, varied · one question per message · move to WhatsApp only after a positive reply (never cold) · hand off at the first confirmed meeting.

17.10 Before you send — the checklist

☐ This person came from the warm queue (they engaged, or accepted and replied)
☐ My message is short: one line of acknowledgement + one question
☐ If I added value, it is specific and real (17.2), not "we help companies"
☐ No pitch, price, or link — unless they asked for it
☐ I am only asking for the call because they named a real problem
☐ WhatsApp only after a positive reply · moved them to the right Infinity stage · logged it

PART 18 · THE OFFER & THE FIRST DM (deep SOP)

The one rule that fixes first DMs. Message one has no pitch, no link, no calendar. One or two lines, one easy question, then you stop. If it is a paragraph, it reads as begging. Short is confident.

Step 1 · Design the offer first, one industry at a time

The single biggest reason DMs are slow and generic is spraying across industries, so every message starts from zero. Lock one industry and stay in it (fintech, logistics, manufacturing, e-commerce, healthtech, SaaS). One industry means one vocabulary, one set of pains, one offer — so the operator gets fast and the messages get sharp. Rotate to the next only when the first is running.

The offer worksheet — fill it before any DM

No worksheet, no DM. This is the step that removes the slop, because the message is now assembled from real, specific inputs instead of invented filler.

Step 2 · Know the buyer (CTO / IT Director / Head of Infra)

You sell to technical buyers, and they get more spam than anyone. The only way in is to sound like a peer who understands their actual job. Pick the pain they will admit (not the one they hide), speak their language, offer the useful thing.

Target both mid-management and top management — but lead with mid. The senior titles above (CTO, IT Director, Head of Infra) sign and set direction, but mid-management (platform leads, senior/lead engineers, product and ops managers under the C-level) reply noticeably more and are usually the real champion who feels the pain day to day. Open with them, prove value, then let them pull you up to the economic buyer. Multi-thread: work the champion and the buyer in parallel, not one or the other.

Step 3 · The first-DM rules (non-negotiable)

Step 4 · The five opener archetypes

These are the shapes. Fill the [slots] from the worksheet, then adapt the words to the country in Step 5.

Step 5 · Adapt to the country (do NOT Americanize this)

The same opener lands completely differently across your markets. An Australian CTO and an Indonesian IT Director do not want to be spoken to the same way. Use the style, then copy the example and tweak per person.

The same opener, five ways — real examples

On writing in the local language. A local-language greeting (Halo Pak/Bu; a courteous Hi) shows respect and is safe. A full message in Bahasa, Malay, or Mandarin goes out only if a fluent advocate writes or checks it — a machine-translated one is worse than clean English and reads as exactly the slop you are avoiding. Match the language of the prospect's own profile; when in doubt, courteous English.

A few more ways in (added 2 Aug 2026)

These are extra, legitimate openings that skip the connection request entirely, plus the signals worth targeting. They sit alongside the flow above, they do not replace it.

Timing that helps a little: Tuesday is the strongest outreach day (~6.9% reply), and avoid Saturdays. Keep it a nudge, not a rule — a strong message on a weak day still beats a weak one perfectly timed.

The principles behind every first DM

Cold DMs work because of trust, not clever wording. Prove three things fast — I understand your world, I can be useful in it, and talking to me is low-risk — and the reply is automatic.

• Trust drives replies: reply likelihood rises with relevance × usefulness and falls with risk/effort.
• Recognition first (show you truly understand their world): prove you understand their role and moment. This is why you work one niche at a time.
• Specific, unexpected relevance creates curiosity; generic creates resistance.
• Usefulness, not "value": offer something usable now, not a vague value prop.
• Risk reduction + ask permission, don't pressure: easy out, no obligation, ask before you send or pitch.
• Then stop: silence signals status; chasing signals desperation.
• Solve one real thing for them: lead with the single specific useful thing this person needs right now.
• Give without asking — be useful first (an intro, an insight) with no demand attached.
• Pitch yourself with what you've done and a bit of human range, not client logos.

Worked examples — the full flow, kept simple

No crazy workflows. Pick the niche, ask one short question, and when they reply, offer the useful thing — asking permission first. That's it.

Before you send — the guardrails

Run this checklist on every first DM:

☐ One industry locked (not spraying across industries)
☐ Offer worksheet filled for this industry × market × persona
☐ The persona's admitted pain + insider language identified (Step 2)
☐ Message one: one or two lines, no pitch, no link, no calendar, one easy question
☐ Written in the country's style (Step 5), not Americanized; local language only if a fluent person checked it
☐ Permission-first: the useful thing is offered after a reply, asking first
☐ Then stop — no double-texting
☐ Every message anchored in the five inputs (Part 8, 8.1): the trigger, the angle, the offer, the stage, the voice

This SOP is the front end of Part 17: SOP-C designs the offer and opens the conversation; Part 17 runs the conversation to a booked call. The country notes describe professional-communication tendencies, not rules about individuals — the advocate's own read of the person always wins.