The current LinkedIn outbound conversation looks like this:
"R.I.P. cold outbound. Claude just booked us 12 demos in 5 days."
"Here's the 23-word DM that gets 80% reply rates."
"7 intent signals that predict buying behavior."
All of it is about the message. None of it is about the account that has to send the message.
The story behind half of those viral posts is the part nobody mentions. The operator burned through three LinkedIn accounts to get the case study they're showing you. They warmed for two days instead of two weeks. They sent 600 connection requests in month one. They got their main account flagged in week six and quietly migrated to a backup. The 12 demos happened. The framework worked. But the channel costs nobody is putting in the post are the difference between a system that compounds and a system that breaks every 90 days.
That's what the LinkedIn Outbound Architecture cheat sheet covers. One page, nine sections, free to read, no gate. This post is the walkthrough.
What the cheat sheet is
A complete production blueprint for running LinkedIn outbound systematically. It maps the five-layer tool stack, the account infrastructure that has to be in place before any automation, the warmup protocol, the seven safety rules, the lead sourcing process, the message arc, the multi-channel sequence with email, the benchmarks that tell you whether the engine is working, and a 90-day launch plan.
Same architecture I run for client builds. Tools, ratios, daily limits, copy principles. Read once, bookmark it, come back when you set up the next campaign.
Why this matters now
LinkedIn outbound is the highest-yield channel for any B2B company sub-25M ARR right now. It is also the easiest one to get wrong in ways that take months to recover from.
A banned account is not a setback. It is three months of profile rebuild, two months of warmup, and a clean restart on relationships you already built. Most operators don't realize this until it happens to them. Then they spend the next quarter rebuilding the thing they should have protected from day one.
The viral playbooks tell you to send more, send faster, use AI. The cheat sheet tells you the constraints you have to respect to send anything at all without burning the account that runs your pipeline.
The five-layer stack
Section 01 of the page lays out the stack as five columns. Read top to bottom, this is the order data moves through your system:
Signals. Trigify for intent triggers, job changes, funding events, tool adoption.
Sourcing. LinkedIn Premium plus Sales Navigator with active-user filtering.
Enrichment. Claude Code for orchestration, Exa for company research, LeadMagic for email verification.
Sequencing. HeyReach for LinkedIn sequences, Unipile for custom API-layer sends.
Unified Inbox. MasterInbox for combined LinkedIn and email view, n8n for cross-channel state.
You don't need every tool in every column. Most engines start with two of the five and bolt on the rest as volume grows. What matters is the order. Signal before sourcing means you only spend connection slots on accounts who are actually moving. Enrichment before sequencing means you don't waste a request on someone with a stale title. Unified inbox at the end means a reply on email doesn't get answered with a DM the next day.
The seven safety rules (the part most people skip)
Section 04 of the page is the gold. These are the seven rules that decide whether your account survives 90 days.
Stay under 400 connection requests per month on a non-premium account. Sales Navigator buys you more, but the rule is per account, not per operator.
Target active users only (posted within 30 days). LinkedIn protects active users harder. Inactive accounts flag you faster.
Three to four sentences maximum per message. Wall-of-text DMs are the strongest "this is automation" signal.
Business hours only, Tuesday through Thursday. Send outside this window once and the spam algorithm notices.
Randomize timing between every action. No fixed-interval sends. HeyReach handles this if you let it.
Include a value hook or lead magnet in the second or third message. Loom, post, webinar, newsletter. Not a pitch.
Never pitch in the connection request. The connection request is the most fragile message in the entire sequence. One pitchy request and you train LinkedIn to flag your future sends.
If you only take one section from the page, take this one. Print it. Pin it above your screen.
The warmup nobody runs
Section 03 is the one I get the most pushback on. The cheat sheet says two weeks of patience. Operators routinely cut it to two days. They get away with it for a campaign or two, then the account gets restricted on day 40.
Here's the actual protocol from the page:
Week 1: Manual activity only. 60 interactions per day. Views, likes, comments. No connection requests, no automation.
Week 2: Ramp to 80 interactions per day. Start light connection requests at 5 to 10 per day. Still no sequencer.
Week 3+: Introduce automation carefully. Watch for warnings.
This is the thing that costs you nothing in dollars and everything in patience. The accounts that skip warmup look fine for the first month and then quietly degrade their reach for the next six. The accounts that warm correctly do the same volume with 2x the acceptance rate.
The multi-channel sequence
Section 07 maps the cadence across LinkedIn and email. The architecture matters here because LinkedIn warms, email closes.
Day 1: LinkedIn profile view plus connection request. No message yet.
Day 3: Connection accepted. Welcome message with no pitch, one sentence of context.
Day 5: Email 1. Treat it as a continuation of the LinkedIn conversation, not a cold open.
Day 9: LinkedIn DM. Drop value. A Loom, a post, a checklist.
Day 14: Email 2. Soft CTA.
Day 21: Final touch on both channels. Breakup.
Most operators try to run LinkedIn and email as parallel cold channels. They end up sending to the same prospect twice with two different value propositions and burning the relationship. This sequence runs them as one continuous conversation across two channels with one shared narrative arc.
How to tune this for your own ICP
The cheat sheet is the architecture. The variables you change for your own engine are these:
Section 02 (account infra): Edit the operator-to-account ratio if you have a larger team. Two to four accounts per operator is the soft cap. Past that, the management overhead eats the volume gain.
Section 05 (lead sourcing): Adjust the 60/40 LinkedIn-to-email split based on where your buyers actually live. Engineering buyers skew email. Marketing buyers skew LinkedIn. ERP-partner buyers (the segment I'm building for right now) skew niche forums and almost ignore LinkedIn DMs cold, but accept connections from operators they've seen post.
Section 06 (copy principles): The four-message arc (no-pitch request → context → value → soft CTA) holds across ICPs. The content of the value drop changes by ICP. Engineers want a Loom. Founders want a 1-pager. Operators want a checklist.
Section 08 (benchmarks): The 25/10/2 baseline (acceptance rate / reply rate / meeting rate) is for cold outbound to active-user-filtered prospects. If you're sourcing from intent signals, expect the top of the funnel to land around 35/15/5. If you're targeting buyers who don't live on LinkedIn, calibrate against 15/8/2 and don't compare to the public benchmarks.
How to use the page
Three ways:
As a setup checklist. Open it before you build a new campaign or onboard a new operator. Walk the nine sections in order.
As a triage tool. When a campaign underperforms, find the section that maps to the part that's broken. Acceptance rate low? Section 04 (safety rules) and Section 02 (account infra). Reply rate low? Section 06 (copy). Meeting rate low? Section 07 (sequence) and Section 08 (benchmarks).
As a hand-off doc. When you transfer the engine to a client team, the page is the SOP. Loom-walk it once and they have the architecture without you.
What I'd add if I were starting today
Two things that aren't on the page yet, that I'd build into the workflow if I were starting fresh in 2026:
An intent-signal layer above sourcing. Don't pull a 1,000-row Sales Nav export and connection-request the top 400. Pull the export, score it against three or four trigger signals (job change, funding, tier-up, content engagement), and only spend slots on the top 100. Same volume of meetings, 80% fewer requests, healthier account.
A weekly account-health check. Acceptance rate per account, reply rate per account, warning notifications. Five minutes a week. Catches the account that's about to get restricted before it does.
Both of these are version-2 layers on top of the architecture. The architecture has to be solid first.
LinkedIn outbound is the highest-yield channel a $2-10M ARR B2B company has access to. It is also the channel where one bad month of sends can cost you a quarter of pipeline and an account that takes three months to rebuild.
The viral content is going to keep selling you better messages. The cheat sheet builds you a better channel.
Bookmark the page. Come back to it when you set up the next campaign.
If you want more like this, the rest of the resource library lives at thegtmarchitects.com.
Until next week,
The GTM Architects