Every 'Not Worth Building' Your Business Filed Is Now Quietly Wrong

My CRM had 4,223 duplicate company records rotting in it. Nobody had ever proposed cleaning it up properly, because “hire someone to manually review four thousand records” was never going to win a budget conversation, so it sat there year after year, correctly filed as not worth the money. One afternoon and a script later, it was done — and nobody had to make the case for it at all. The records hadn’t got any less duplicated. What changed was the price of fixing them.

The bar was never about the problem

Every backlog has a graveyard in it: the ideas that got a nod and then quietly died. Not because they were bad. Because someone ran the sum: a developer’s day rate against ten minutes saved a week. The sum said no. That “no” wasn’t wrong. It was correct, given what building used to cost. The mistake is assuming it’s still correct now that the other side of the sum has moved.

It isn’t speed. It’s a different backlog

I used to think getting faster meant clearing the same list quicker. It doesn’t work like that. Collapse the build time from a fortnight to an afternoon and the list itself changes shape: a whole tier of ideas that never cleared the old bar clears the new one without anyone having to argue for it. The duplicate records are the bit that surprised me, because they were never on a list of things I’d meant to get round to. They were never on a list at all. The change doesn’t just speed up the work you’d already agreed to. It quietly readmits the work you’d never have agreed to in the first place.

The graveyard reopens whether you look at it or not

This is the part I didn’t expect. An old “no” doesn’t sit quietly waiting to be reconsidered. It rots. The moment the cost assumption behind it stops being true, that “no” is wrong, and it stays wrong, unnoticed, until somebody goes back and rechecks it. Every list of things a business decided weren’t worth the effort was scored against a build cost that’s since fallen through the floor. Nobody goes back and rescopes a rejected idea from three years ago. It just sits there, correctly filed as “not worth it” under economics that no longer apply.

This isn’t just my backlog

I don’t think this is a personal quirk of having six months and a capable model. Any business sitting on a pile of small internal fixes is sitting on the same stale arithmetic: the report someone compiles by hand every month, the data nobody got round to cleaning, the one process with an annoying manual step in the middle. The fixes were always technically possible. They just never cleared the cost bar at the old price. At the new price, a meaningful slice of that pile clears it easily, and most of it has been sitting there so long nobody thinks to check any more.

The catch is deciding, not building

None of this means build everything that’s now cheap enough. Some of what’s in the graveyard was correctly killed on the merits, not the cost, and cheap building doesn’t fix a bad idea. But that’s a different, harder problem than the one I’m describing here. It’s the one about who actually decides, and it deserves its own essay rather than a bolt-on paragraph at the end of this one. What I’m pointing at is narrower: a huge amount of genuinely good, genuinely wanted work has been sitting unbuilt for years for no reason except that the sum used to say no. That sum has changed. Almost nobody has gone back to redo it.

So go back through whatever list your business keeps of the fixes that were never worth the time, and rerun the numbers at this year’s build cost, not last year’s. Some of it will still be correctly dead. I’d bet more of it than you think isn’t dead at all. It was just priced out, and nobody’s checked the price since.

If you’ve got a pile like that and no real appetite for digging through it yourself, get in touch. Going back through the small internal fixes a business wrote off as not worth building, working out which were killed on cost rather than on merit, and rerunning the arithmetic at what building actually costs now — that’s the work I most want to be doing at the moment.

Companion essay to I didn’t build products. I built the thing that builds products — the AI Sabbatical, post 3.