A book of B2B accounts on long cycles — $10.4K typical contract value, 9-month median, LinkedIn-led. Three views of one engine, drilling down: the whole book ranked by what needs a person, then a single account up close, then the portal your client actually sees — with your hands on everything in between.
Churn almost never shows up in campaign performance first. It shows up in behaviour — who joins a call, how fast approvals come back, how long the replies are, what stopped being asked. That lives in your calendar, your project tool and your call recordings, not in an ads platform.
ClickUp Fathom calendar
email metadata HubSpot — none of it from an ads APIIt will not message a client about a health flag, ever. A retention signal is a judgement about people, and the cost of getting it wrong in front of the client is far higher than the cost of missing it. Everything on this screen is internal, and every intervention is a person choosing to act.
What it will do is put the right draft in front of the right person at the right time, with the evidence attached — and then learn from what you actually did instead.
A new account starts high — the engine has no history with it, so nearly everything is held and every correction you make becomes training data. As the golden data set fills out, drop the dial and it handles routine work unattended. Move it back up for one sensitive account without touching the others.
LinkedIn Ads GA4 Search Console Peak HubSpot · tone matched to 14 prior sends from this account leadfinance-in-thread, which caps auto-send at 95% regardless of the dialMeta Ads LinkedIn Ads · matched against 41 prior fatigue events across the book| Channel | Weighted contribution | Influenced | First touch | Closing touch |
|---|---|---|---|---|
| LinkedIn AdsFacilities Director + IT Ops | $754K | 61% | 12% | |
| Organic searchcomparison + pricing intent | $442K | 19% | 17% | |
| Google Adsbrand defence + high-intent non-brand | $331K | 7% | 44% | |
| GEO / AI searchLLM citations · Reddit, comparison pages | $202K | 11% | 3% | |
| Email nurturemid-cycle re-engagement | $111K | 2% | 24% |
Three signals moved together: the CFO joined the last two calls having never attended before, creative approval latency went from 2 days to 11, and a Series B described as "closing in Q3" back in March still hasn't been announced. The pattern matches four prior accounts, three of which cut spend at renewal.
Suggested move: bring a cost-per-opportunity story to the next QBR rather than a spend-growth story.
Fathom · 3 calls ClickUp approval timestamps HubSpot · funding signal unverified$1,840 per opportunity against $4,210. Added in April as a test, now quietly the strongest thing in the account. Budget shift already drafted and sitting in the queue above.
LinkedIn Ads HubSpot · 142 opportunitiesReplies have got shorter and slower since the account changed hands, and two requests were repeated having already been answered. Low confidence and a judgement call about people, so it is surfaced to you and never drafted to the client.
email sentiment Fathom · classified internal-onlyLinkedIn Ads GA4 HubSpot PeakLinkedIn Ads Google Ads HubSpotMeta Ads LinkedIn AdsPeak Search Console · flagged because it names competitor movementThe client portal is the part of the system your clients judge you on, so it is the part you should have the most control over. Every generated block here can be approved, rewritten, or pulled entirely — and you can add a block the engine could never have produced, because it happened in a room. Your edits are stored as training pairs against this account, so the next draft starts closer to how your team actually writes.