The Theory · Episode 6
The War of Attrition: Surviving a Price War Nobody Can Afford
22 September 2026
What happens when the entrant doesn't back down — and both of you are now burning cash to find out who blinks first.
The Thesis
Last episode ended with a bluff. An incumbent fights hard in the first few towns of the Chain Store Game to build a reputation that deters everyone else — a display, aimed at future entrants more than the one in front of them.
But reputations only work if the entrant folds. Sometimes they don't. Sometimes the entrant has their own reasons to hold — a fresh raise, a board that's told them not to blink, a belief that the incumbent's discount isn't sustainable either. Now you have two competitors both discounting, both losing money on every deal, and both still standing.
This is no longer the Chain Store Game. The reputation play is over — it failed. You're in a different game now, one with a much uglier structure: the War of Attrition.
The Game
First modelled by John Maynard Smith in evolutionary biology — two animals contesting a territory, each able to retreat at any moment, escalation costing both of them more the longer it continues — and it maps onto economic competition almost without translation.
The setup: two firms, one contested segment, both fighting on price. Every period the fight continues, both lose money. Whoever quits first gets nothing further from the fight, but stops the bleeding. Whoever stays gets the segment to themselves — eventually — but only after paying for every period both of them refused to fold.
The critical feature, and the one that makes this game genuinely dangerous: there's no dominant strategy, and there's no natural stopping point built into the game itself. Unlike the Chain Store Game, where backward induction gives a clean answer, the War of Attrition's equilibrium is about endurance, not cleverness. The player who can tolerate the longest losses — or who most convincingly appears able to — wins. Rationality alone doesn't get you out of it; running out of runway does, on one side or the other.
| You quit | You hold | |
|---|---|---|
| They quit | Split — both stop losing. | You take the segment; they eat the cost of holding. |
| They hold | They take the segment; you eat the cost of holding. | Both keep losing, every period, until someone breaks. |
There's a reason biologists reach for this model to explain animal contests that end in real injury even though retreat is always available to both sides: once committed, quitting feels like converting your losses so far into total loss, so both sides keep paying past the point a spreadsheet would tell them to stop. Sunk cost isn't a side effect of the War of Attrition. It's the mechanism that keeps it running.
Why This Is Worse Than It Looks
Two things make a price war different from an ordinary competitive skirmish:
It doesn't end when the economics say it should. Both sides typically know, well before either quits, that the fight is unprofitable for both. Knowing that changes nothing, because the cost of quitting now includes writing off every dollar already spent holding the line. That's what turns a rational decision into an endurance contest.
The signal you're sending matters more than the price you're setting. Every period you hold is a public bet that you can outlast them. If that signal isn't true — if your board is nervous, your runway is short, your unit economics can't absorb it — continuing to fight is making a promise you can't keep, and the moment that becomes visible, you lose the fight and the credibility that would have protected you in the next one.
The Two Ways Out
There are, structurally, only two exits from a War of Attrition, and they call for opposite information:
- You have a genuine endurance advantage and can prove it. Better unit economics, more capital, a larger addressable base for the segment. If true, holding is correct — but only if you can also convince them it's true, because their decision to fold depends on believing you'll outlast them, not on you actually being able to.
- You don't, and the fight was never worth entering. In which case the objective isn't to win the war — it's to recognize the mismatch early and exit before the sunk-cost mechanism has locked you in.
Both paths require the same input: an accurate read of the other side's capacity to hold, not just their current behaviour. Behaviour in a War of Attrition is partly signal and partly bluff, and mistaking one for the other is how rational competitors end up burning capital neither of them can spare.
Framework to Steal: The Endurance Audit
Before matching a competitor's price war, or before starting one:
- Cost your own hold, not just theirs. What does another quarter of this fight cost you in margin, and how many quarters can you actually sustain it — a real number, not a confident guess?
- Look for their tell, not their price. Funding stage, hiring pace, other product signals, investor commentary — anything that suggests their capacity to hold is thinner than their pricing behaviour implies.
- Separate the segment's value from the fight's cost. A segment worth winning at a fair price can still be a segment not worth winning at war-of-attrition prices. Those are different numbers; conflating them is how firms win a segment and lose the year.
- Decide your exit condition before you're in it. Set the walk-away point while you can still think clearly about it — not three quarters in, with sunk cost already doing the deciding for you.
Where Pro-Position.AI Fits
Every step of that audit depends on information you don't generate internally — it depends on reading the other side. And the entire mechanism that keeps a War of Attrition running past the point of rationality is uncertainty about the competitor's actual capacity to hold, mistaken for certainty about their intent to.
[Pro-Position.AI](
The War of Attrition doesn't reward the side with the better product. It rewards the side with the better information about who blinks first.
First published as an edition of The Theory, Pro-Position.AI's newsletter on competitive intelligence, product strategy and go-to-market.
Keep exploring: