Skip to main content
Context file — Outreach

PERSONAS.md

Messages that speak to what the reader is measured on rather than to a job title. Not a marketing persona — no names, no stock photos, no "Marketing Mary". Each entry records what the person owns, what they are judged on, what they have already tried, and the specific sentence that ends the conversation, which is the field nobody writes and everybody needs.

PERSONAS.md
.md
# Personas — <company>

One block per person in the buying committee. If you have more than three, you have a target list rather than a persona set.

---

## <Finance Director> — the economic buyer

**Owns:** <the monthly close, the finance team's headcount, the software budget under <amount>>

**Measured on:** <close time, audit findings, and not being the reason the board pack is late. Not on software cost.>

**A normal week:** <reporting, board prep, managing a team of six, firefighting whatever the close threw up. LinkedIn is checked between meetings, on a phone.>

**Already tried:** <adding a person to the payroll team; asking the ERP vendor; a spreadsheet the previous controller built and nobody else understands.>

**What they believe that is wrong:** <that the close is long because the team is stretched, rather than because the entity count grew.>

**Opens a message when:** <it references something specific they said, and does not ask for a meeting in the first line.>

**The sentence that loses them:** <"We help finance teams streamline their processes." They have read it forty times this year.>

**What they ask first:** <"How long does implementation take, and what do I need from IT?">

---

## <Payroll Manager> — the user

**Owns:** <running payroll across five entities and the reconciliation that follows>

**Measured on:** <accuracy and the close-week overtime>

**Already tried:** <three spreadsheets and a lot of patience>

**What they believe that is wrong:** <that this is just what the job is>

**Opens a message when:** <it describes their actual week accurately. Specificity about the work is the whole game with the user.>

**The sentence that loses them:** <anything implying their process is bad. They did not choose it.>

**Note:** <the user cannot sign, but they can kill it in one sentence to the FD, and they are the reason a deal that looked dead comes back.>

---

## <Head of IT> — the blocker

**Owns:** <integrations, data access, security review>

**Measured on:** <nothing going wrong>

**Opens a message when:** <almost never. Do not lead here.>

**The sentence that loses them:** <"It's a simple integration.">

**Note:** <engage only once finance is committed, and lead with what you need from them, precisely.>
PERSONAS.md
Version 1.0 — updated 18 September 2026Downloads as PERSONAS.mdReplaces the persona deck nobody opens
Where it goes
  • Claude Project knowledge
  • ChatGPT Project files
  • Agent folders
What you have to fill in
  • One block per buyer. Two or three is right; six means you have not chosen.

Everything in <angle brackets> is example content, not instructions. Replace it. The file ships filled in rather than empty on purpose: an empty template gets abandoned, a filled one gets edited.

Where the content comes from

The reference behind this file

How a buying decision actually gets made

The comparison pages here are written for the same reader this file describes: someone deciding, on a phone, between meetings.

Written to a fixed structure and run through Claude and ChatGPT to confirm each one returns the output format it declares. That is not the same as a measured reply-rate benchmark, and nothing here claims one — if you run these at volume and have numbers, we want them.

Where to next

One file is a start. The folder is the point.

Files reference each other — CLAUDE.md declares the reading order, the prompts expect the files by name. The packs ship them together, already wired up.