"GTM engineering" has become the name for a job most B2B teams already need: wiring billing, product usage, CRM and marketing data together so that sales, success and marketing act on the same facts at the right moment. Most of that job isn't building pipelines. It's asking the same questions every week — who is ready to buy, who is about to churn, what actually sourced last quarter's revenue — and getting an answer someone can act on.

Cogny already connects to the systems those answers live in. Today we're adding nine playbooks that ask those questions on a schedule. Each one is a scheduled report: it runs on its own, reads your connected data, writes a report, and files tickets for the actions it finds.

The nine playbooks

1. Buying Intent Signals — twice a week

Finds the accounts showing buying intent right now: repeat pricing and demo page visits, form fills, product sign-ups from a target-account domain, LinkedIn ad engagement by company, brand mentions. It matches each one to your CRM by domain, routes it to the deal or account owner, and says why now, with dates. An ICP account on the pricing page on three different days in one week is always flagged.

For the hottest accounts it researches recent company news and drafts a two-to-three-sentence opener that cites a real, linked fact. The owner edits it and sends it themselves: Cogny never sends anything.

Needs: HubSpot, Pipedrive or Upsales. Gets sharper with Upsales web tracking, HubSpot Forms, LinkedIn Ads and product analytics.

2. Product-Qualified Leads — weekly

Works out the "aha moment" from your own data instead of a generic benchmark. It tests 10–20 value actions (connect an integration, invite a teammate, run a report) at different counts and time windows, and picks the threshold that best separates accounts that went on to pay. Then it scores this week's free and trial accounts and lists the ones that just crossed it, the ones hitting plan limits, and the ones that stalled.

Needs: PostHog, Mixpanel or Segment, plus billing status from Stripe or a plan property.

3. Pricing & Monetization Signals — monthly

Plan mix, ARPA, upgrade and downgrade paths, revenue lost to discounts that never expire, and customers still on old prices. Then the strategic part: it compares each account's usage with what it pays, flags the accounts that have outgrown their plan and the ones paying for far more than they use, and simulates a revenue-neutral usage-based price so you can see who would pay more and who would pay less.

Needs: Stripe (or Stripe tables in BigQuery) and a usage measure.

4. Closed-Won Source Attribution — monthly

Starts from closed-won revenue, not platform-reported conversions. For every deal won or lost in the last 180 days, it collects every touch before the close (ad reach and clicks by company, web sessions, form fills, webinars, meetings, outbound) and shows first-touch, last-touch-before-opportunity, position-based and linear attribution side by side. It also shows which combinations of channels go with the best win rates, and cost per won deal for each paid channel.

Needs: HubSpot, Pipedrive or Upsales, plus whichever ad platforms you run.

5. Objection Handling Battlecard — monthly

Reads the last 90 days of lost-deal reasons, sales notes, recorded sales calls, pre-sales chats and cancellation reasons. It groups the objections, ranks them by lost revenue, and finds the answers that worked in deals you won. It keeps one living battlecard in your context tree at sales/battlecard, with verbatim quotes and every answer marked proven or untested. Your team reads it there, and so does every Cogny agent working in the workspace.

Needs: HubSpot, Pipedrive, Upsales or Crisp.

6. NRR & Expansion Signals — weekly

Net and gross revenue retention, with the formula worked out on your own numbers. Then the forward-looking part: accounts renewing in the next 45 days whose usage or active seats are dropping, each with the specific change behind the flag, and paying accounts at or over their plan limits, with the estimated uplift of the next plan. Each run checks whether last week's at-risk accounts were saved.

Needs: Stripe and a product usage source.

Measures the demand attribution can't see: branded search, direct visits that land on deep pages, mentions in AI answers, X and LinkedIn mentions, community threads. If your forms ask "How did you hear about us?", it puts what people said next to what your analytics credited for the same sign-ups. If they don't ask, adding that question is its first recommendation.

Needs: Google Search Console. Uses GA4, AI Visibility, X, LinkedIn Organic and Discord when they're connected.

8. Competitor Pricing Teardown — monthly

Reads each competitor's pricing, plan-comparison and getting-started pages and records a structured snapshot: tiers, value metric, free plan or trial, credit allowances and top-ups, feature gates, enterprise threshold. Each month it shows exactly what changed since the last snapshot, with sources, and where your own pricing sits next to theirs. Snapshots are saved in your context tree under competitors/pricing.

Needs: nothing beyond your competitor context.

9. Billing ↔ CRM Reconciliation — weekly

Checks that your CRM tells the same revenue story as billing. It finds paying customers the CRM doesn't know about, customers who churned but are still marked as customers, closed-won deals with no subscription, and MRR or plan values that don't match. It lists every correction (record, field, current value, correct value) and writes a short sync spec so the drift stops coming back.

Needs: HubSpot, Pipedrive or Upsales, plus Stripe.

What they have in common

Read-only. None of these playbooks sends an email, posts, or edits your CRM, billing or ad accounts. Their only writes are reports, tickets, and the battlecard and pricing snapshots in your own context tree.
Real numbers or nothing. Every figure comes from a query or tool result in that run. When a source isn't connected, the report says what's missing instead of estimating.
Correlation is labelled. Attribution, PQL lift and dark-funnel footprints show what moves together. Where it matters, the report suggests the holdout or experiment that would prove cause.
Tickets, not just reading. Each playbook turns its findings into tickets assigned to the right person: accounts to contact this week, renewals at risk, CRM records to fix.

Getting started

Open the Reports page in your workspace. Playbooks whose data you've connected show up as recommended; enable one, keep or change its schedule, and add custom instructions if you want to steer it (your ICP, a named account list, the competitors to track). Runs count against your plan the same way as any other scheduled report.

Agents connected to your workspace over MCP see the same reports and tickets, so your own coding or sales agent can pick up where a playbook leaves off.