A field note on the work AI used to be told it couldn’t do — and why most of that list is now a 60-day build.

The field-note version of our essay on price — where the cheapest bid actually comes from, what it quietly costs, and why we don’t run the race to the bottom.

This is the field-note version of the essay “Why you keep losing on price.”

ON PRICE

If you run a small shop, you already know the feeling. You put real care into a proposal, you price it honestly, and then you lose the job to someone who came in thousands of dollars under you. Maybe it’s a prospect who went straight for the cheapest name on the list; maybe it’s work you were genuinely the best fit for. Either way it stings, and the first instinct is almost always to wonder what’s wrong with your number.

Here’s the part worth sitting with: usually, nothing is wrong with your number. To understand why this keeps happening, it helps to look at how the other quote got so low in the first place — because once you see the mechanics, the whole problem starts to look different.

A bid, stripped all the way down, is just a stack of costs with a margin on top. When someone comes in far below you, they didn’t discover a clever way to bend the laws of math. They took something out of the stack. And there are really only a few places it can come from.

The first is wages. It is a lot cheaper to staff a project with junior or rotating contract labor than with experienced people who expect to be paid what they’re worth. The second is location — shipping the work overseas to whoever will bill the fewest dollars per hour, trading day-to-day oversight and accountability for a smaller line item. The third is margin itself: plenty of shops quote at a net margin of just 4 to 6%, thin enough that one slow quarter turns into an emergency, and they make the difference back later through change orders you never saw coming.

None of that shows up on the proposal. What shows up is a smaller number. But the cost didn’t actually vanish — it just moved. It moved onto the worker who’s underpaid, onto the client who discovers the quality gaps three months in, onto the deadline that slips because nobody senior was really watching. Cheap is rarely cheap; most of the time it’s simply deferred.

So when a competitor undercuts you by a wide margin, that gap isn’t evidence that you’re overpriced. It’s a readout of what they decided to remove. A shop that pays a livable wage, keeps the work close to home, and puts senior people on it needs a little more margin to stay healthy — realistically closer to 8% where the underbidders are scraping by on 4 to 6. That isn’t greed. It’s the plain arithmetic of doing the work properly and still being around next year to stand behind it.

Which points to what you should actually do, and it isn’t chasing the bottom. You lose that race even when you win it, because the only way to match the lowest bid is to start removing the same things they removed — and then you’ve just become the vendor you were competing against. The stronger move is to change what’s being compared. Shift the conversation from sticker price to total cost: the rework nobody has to pay for, the institutional memory you don’t lose to turnover, the deadline that gets hit because someone experienced owns it end to end.

That gap is exactly what we built Anchor to close. We keep wages livable and the work onshore, we cut the overhead that quietly inflates everyone’s rate — there’s no trophy headquarters folded into your invoice — and we run AI and human effort in tandem, so you get the speed of automation without paying to feed a swollen org chart. We replace the work, not the worker. The goal was never to be the cheapest name on the list; it’s to make the little bit extra buy you noticeably more — work that’s owned by your team, done end to end, and built to outlast the first invoice.

So if it feels like you can’t win on price, here’s the reframe: you’re right, and that’s actually good news. Price was never the game worth winning. The work is.

→ Read how we price. → Run your real number. → Talk to us for 20 minutes.