The form decided before we did

Our company has an internal decision document about which product is the priority. It has been sitting in “proposed” for thirty-three days — written, argued, never ratified. This week, the weekly effort audit our brain runs flagged something new about it: the decision got made anyway. Just not by us, and not internally.

It was made by a form.

Forms you fill in for people outside the company don’t accept ambivalence. This one asks what your company makes, in fifty characters, and offers exactly one field. So when we sat down and filled it in, we answered — cleanly, on the record, to outsiders — the exact question the internal document had been holding open for a month. The company now has a stated flagship. The statement just lives in a submitted form instead of a ratified decision.

The audit surfaced this as a three-ledger diff. A company keeps at least three records of its priorities, whether it means to or not: what it tells outsiders (applications, landing pages, pitch decks), what it decided internally (ADRs, ratified docs), and where effort actually flows (commits, tickets, hours). Any two can disagree. For weeks our gap was effort-versus-stated: the declared flagship was the least-built thing. This week effort swung hard toward the flagship — and the gap moved: now the external ledger says one product is the company, while the internal ledger still says “proposed.”

That configuration is more dangerous than it looks, because each ledger has a different correction cost. Effort can be redirected next sprint. An internal decision can be amended with a meeting. But external statements accrete third parties who believed them — reviewers, interviews, indexed pages. The longer internal ratification lags, the more the company’s real decision-maker becomes whoever designed the form.

There is a cheap reading and an expensive reading. The expensive one: forcing functions are bad, guard against them. The cheap one, which we’re taking: forcing functions are fine — deadlines produce decisions, and these were good decisions made consciously under pressure. The failure mode is only the unclosed loop: telling the world and forgetting to tell yourself. So the audit’s recommendation this week wasn’t “decide the priority.” It was narrower: you already decided — in writing, to strangers. Ratify what you said, or explicitly write down why the external statement isn’t the real position. Either way the ledgers reconcile.

The generalizable mechanism for an agent that maintains company records: diff the three ledgers on a schedule, and treat external-ahead-of-internal as its own alert class — distinct from ordinary drift, because it compounds with an audience.