What is a 100-day AI GTM plan for a new portfolio company?

A 100-day AI GTM plan switches on eight workflows in four phases: diagnose (days 1 to 14), foundation (15 to 45), stop the leaks (46 to 70) and create demand (71 to 100). Each one runs on data the company already holds: call recordings, the CRM and closed-won deals. Outbound comes last, after the offer and list are fixed.

PE firms hand a new portfolio company a 100-day plan. This is the go-to-market section written as a build plan: eight workflows, in the order to switch them on, each reading data the company already has. Hg and Clay published a three-level framework for the same job, and it is cited below where it supports a step.

Why do the first 14 days go to diagnosis?

Because the plan should be chosen by the company's own evidence, and most of that evidence sits unread. Two workflows run in parallel. The lost-deal reader takes every sales call recording and ranks the reasons deals were lost and the questions buyers ask most. The GTM audit reads the CRM, the website and the tool stack and shows where the pipeline leaks: unanswered inbound, slow hand-offs, empty fields, stages nobody uses.

Data: call recordings, the CRM, the website, the tool list. Measure: a scorecard out of 24 and a ranked list of leaks. Hg's first-level example, "Deal Inspections", is the same move: analyse call transcripts for insights, risks and next steps.

What does the foundation phase fix, days 15 to 45?

The data, before anything reads it. Hg's blueprint says failures at the start "come from poor data foundations rather than AI tool limitations", and Ideagen's CRO told Hg's Growth Summit that AI amplifies broken processes and bad data rather than fixing them.

The CRM fixer dedupes and enriches every record and stamps each field with its source and date. The TAM mapper reads the closed-won deals and finds every account that looks like them, with a tier on each. Add the suppression list (customers, open deals, partners, the fund's relationships) and buy sending domains now, so they warm while the rest is built.

Data: the CRM, closed-won deals, the customer list. Measure: duplicate rate, share of accounts with a tier, share of fields with a source stamp.

How do you stop the leaks, days 46 to 70?

Work the demand the company already has. The speed-to-lead router enriches and scores every inbound form and puts it in a rep's Slack within 60 seconds, with the context to call. The dead-deal winback reads closed-lost and stalled deals and writes a reason to reopen each one: a new executive, a funding round, a competitor gone quiet.

This is where the call-recording findings change the sales process. Objections that lost deals get answered earlier. Hg lists "Lead Enrichment Workflows" for every inbound prospect among its second-level, standardised uses, and Clay's PE playbook puts inbound qualification and routing in its production-ready tier.

Data: form fills, the lead queue, closed-lost and stalled opportunities. Measure: minutes from form fill to first touch, share of inbound worked, deals reopened.

Why does outbound come last, days 71 to 100?

Because the offer and the list come out of the first 45 days. The signal watcher turns funding rounds, hiring and new leaders at target accounts into a weekly call list. The lookalike trigger starts a campaign to the companies most like each new customer on the day the deal closes. Both send through the domains warmed since day 15, and a person approves every message. Clay's playbook draws the same line: personalised messaging is "rep-in-the-loop", and fully autonomous outbound sits in its "not ready" tier.

Data: the TAM map, the signal feeds, the new-customer list. Measure: replies and meetings per 100 sends, by signal and by segment.

What does the plan look like on one page?

Who runs it?

A GTM engineer inside the company, reporting to the CEO and the operating partner. Hg and Clay describe the role as "a dedicated GTM engineering function that sits between traditional RevOps and the broader engineering organization". Their framework runs over longer windows than this plan, six, twelve and eighteen months across three levels, and the 100 days above cover the GTM items from the first two. Simon Walker of Kubrick Group gives the board cadence: two or three use cases, a 100-day plan for each, measured monthly.

Build the eight workflows once and run them in every portfolio company. Each company keeps its own data. The operating partner gets the same weekly pipeline report from all of them.