First 90 Days as PMM on an Agentic Finance Product

The first 90 days are for making one finance workflow sellable: a control path a Controller can defend, a baseline the board can check, and a residual owner with a name.

Outcome of the 90 days

Sales and product can tell one control-first story on one beachhead workflow, with a baseline, a named residual owner, and enablement that survives a CFO objection. Homepage, demo, and first-call narrative match.
 
The job is to make the story the company already sells defensible. It is not to add a thought-leadership calendar.

Days 1 - 30: Diagnose and lock the beachhead

Week 1

  • Sit in 5 recent sales calls and 3 lost-deal writeups. Extract the exact objections. Map them to the seven-objection set: accountability, false approval, messy data, headcount theater, existing automation, unmeasured ROI, and governance. Add anything missing.
  • Interview Sales, Product, CS, and one friendly customer (Controller or close/AP lead).
  • Pull the live homepage, deck, and demo script. Mark every line that leads with autonomy, headcount, or “the agent does the job.”

Week 2

  • Choose one beachhead workflow. Criteria: exception-heavy, rules exist, measurable, politically landable. Default is not forecast and not “transform finance.”
  • Write the control path on one page: agent proposes → policy check → named human owns residual decision → audit trail.
  • Name the expensive failure mode (false approval, bad journal, wrong recognition). Positioning is built around that miss, not average accuracy.

Week 3

  • Baseline what you can get without a new data project: exception volume, cycle time, manual touches, rework, time-to-approve, escape rate. If the company never measured it, say that in writing. That gap is now a product-and-PMM problem.
  • Draft before → after messaging for that workflow only (hero, primary value, buyer, demo order). Same shape as the AP and close cases.

Week 4

  • Internal review with Product and Sales. Kill claims the demo cannot show.
  • Decision memo: beachhead, expensive miss, control path, baseline metrics, claims the company will stop making.
  • Staffing note: who inside this company (Product, CS, or solutions) is the Accounting Engineer equivalent (domain + workflow + guardrails). If nobody, flag it as a product risk, not a content idea.

Days 1-30 output

  • Objection log from real calls
  • One-page control path
  • Before → after narrative for one workflow
  • Baseline scoreboard (or an explicit “no baseline” finding)
  • List of claims to remove from site and deck

Days 31 - 60: Rewrite the motion and enable the field

Week 5

  • Ship the messaging hierarchy: category line, problem, control-path value, proof, CTA. Trust and ownership before autonomy and hours saved.
  • Rewrite homepage hero, first two deck slides, and demo opener. Demo order: miss → control path → one exception → then throughput.

Week 6

  • Sales kit: talk track, 1-pager, objection card (the seven plus whatever showed up in Week 1), competitive alternatives page in the buyer’s language (ERP module, RPA, shared services, status quo).
  • Required demo path for SE/AE. The open is not “watch the agent finish the close”.

Week 7

  • Enablement: one 45-minute session with Sales. Role-play “who owns the miss” and “show me ROI.” Record it. Put the card in the deal room.
  • Align CS on the same residual-owner language so post-sale does not reopen autonomy theater.

Week 8

  • Measurement with Product: put the Week 3 baseline in-product or in CS QBRs. Estimate vs actual on the beachhead. No new vanity dashboard.
  • Soft launch the narrative on the next 10 opportunities. Track which objections arrive softer.

Days 31-60 output

  • Live messaging + homepage / deck / demo changes
  • Objection card and alternatives framing
  • Sales session shipped
  • Measurement hooked to the beachhead
  • 10-opportunity test started

Days 61 - 90: Prove it and lock the operating cadence

Week 9

  • Review the 10 opportunities. What did buyers repeat? What did Sales still improvise? Patch the card and demo. Do not add a second workflow yet.

Week 10

  • Customer proof: one named or anonymized beachhead story with before / after process metrics. If a logo is not available, get a Controller quote on ownership and auditability.

Week 11

  • Founder / product review: what the company will not say in Q4. Written. Autonomy stays inside the control path.
  • Next-beachhead decision: stay, deepen, or only then expand (AR, close, rev rec). Expansion requires a measured win, not a bigger story for the board.

Week 12

  • 90-day readout: official story, claims retired, objection frequency before vs after (even if directional), baseline existence and numbers, named residual owner in product and GTM, recommended next hire (PMM specialist vs Accounting Engineer vs sales engineer).
  • Propose the next 90: deepen enablement, second workflow only if the scoreboard exists, hiring plan. 

Days 61-90 output

  • Evidence from live deals
  • One proof story
  • Official “do not say” list
  • 90-day readout a CEO can send to the board

Out of scope for the first 90 days

  • Repositioning the company around “CFO leads AI” — crowded claim, not a demo
  • A thought-leadership or podcast calendar as the PMM job — does not fix the first call
  • A finance-wide transformation story — splits the beachhead before anything is measured
  • Headcount takeout as the lead value — triggers the team-risk objection on contact
  • Adding three personas and four workflows before one path is measured — looks like a plan, produces no baseline

How to score this hire at day 90

  • Can a new AE open the demo on the control path without PMM in the room?
  • Can Product point to the expensive miss and the named residual owner?
  • Does ROI language start from a baseline on the beachhead workflow?
  • Did autonomy come off the first screen of the site and the deck?
This is the same plan I would run in that seat — founding PMM, VP / Head of PMM, or a product-line PMM.
If you are hiring for that role: