A contract fell through. The company I set up for it didn't.
A couple of months ago a consulting contract came up. To take it on properly I set up a limited company, sorted the admin, and wrote a draft of the contract. Then it fell through — the client decided to do the work internally. Which happens.
The thing is, by then the company existed. And the more I worked on it, the clearer it got that the company was the actual project, not the contract.
So: Golden Record Consulting is live, and I'm taking on clients.
I help subscription businesses decide how their revenue should be measured — MRR, ARR, churn, retention — and make sure those numbers hold up, from the board pack back to the ledger.
I know. Revenue metrics. Stay with me.
Ask a finance lead and a growth lead for last month's revenue and there's a decent chance you'll get two different numbers. Usually neither of them is wrong. At VEED, where I worked on the finance data models with the data team, different departments genuinely needed different views of ARR. The problem was never picking the "right" one. It was that nobody had written down which number was for what, and what went into it.
That's the job: decide which metric leads and which ones support it, write the definitions down, and tie everything back to the money that actually arrived.
In data, a golden record is the single, reconciled version of something — a customer, say — built from sources that all disagree with each other. Stripe says one thing, the CRM another, the app store a third. The golden record is the version everyone agrees to trust. That's literally the work.
It's also a nod to the Voyager Golden Record, the disc NASA sent into space in 1977 with the best version of us on it. I studied aerospace engineering, so that was always going to win me over. Clients won't care. I say the name a lot, so I'm allowed to like it.
Three starting points, depending on where a business is:
Every one of them includes a data quality audit, because duplicate customer IDs and messy hierarchies quietly break everything downstream.
On top of that: add-ons for whatever a business actually needs (a database with tests, an ARR snowball, a dashboard, a board pack template, help hiring a data person), hands-on work when something needs building, and monthly advisory for anyone who wants a second pair of eyes before the numbers reach the board. The first few clients get founding terms, in exchange for telling the world how it went.
This being a build-in-public blog: the site is hand-written HTML, CSS and JavaScript, no framework, hosted on Cloudflare. The whole page is one canvas of a few thousand dots that rearrange themselves as you scroll — into a maze of numbers that don't agree, an ARR snowball, a tie-out to the bank, and finally the company mark. Built with Claude Code, with more rounds on the motion than I'd like to admit.
It's also set up to be read by AI assistants, not just people — there's a plain-text version of the site waiting for them. If someone asks a chatbot who can sort out their MRR, I'd like it to have an answer.
Early days. There are a few founders I know who I think will come to me once they hit the stage where I can actually help — and that stage tends to arrive the week before a board meeting.
In the meantime, I'm putting myself out there, baby. Last Friday I went to the MCR Supper Club run by The Female Founder in Manchester, and met some seriously impressive business women. This week it's RevenueCat's App Growth Annual (virtual, so slippers are allowed), and I'm looking at joining the Entreprenista League — call booked for 2 November. In between, a lot of conversations where I don't pitch anything — I just ask people what bothers them about their revenue numbers. The answers are already reshaping the offer. The PE review exists because someone told me exactly what they'd buy.
I'm also sorting out a proper marketing strategy, which for someone who'd rather be building things is its own small adventure. Laura Meyer's Expert Brand Blueprint workshop is next on that list. More on it once it's working (or once it isn't).
The runway from the first post is still counting down. This just gave it a direction.
Next up: finding out whether anyone wants their numbers sorted. If that's you, or you know someone whose board pack has three different ARRs in it — goldenrecordconsulting.com.
// subscribe
Occasional build-in-public updates. Unsubscribe anytime.