The Theory · Episode 5
The Chain Store Game: Is That Competitor Discount a Fight or a Bluff?
15 September 2026
Why a competitor's brutal discount in one deal isn't about that deal at all — and how to tell a real fight from a bluff.
The Thesis
You lose a deal to a competitor who dropped 45% to win it. The prospect was small. The contract barely covers their cost to serve. It looks irrational.
It isn't. They weren't buying that deal. They were buying a reputation — and you were the intended audience.
This is the Chain Store Game, formalised by Reinhard Selten in 1978, and it's the single most useful piece of game theory for anyone trying to work out what a competitor's pricing is actually for.
The Game
An incumbent operates in twenty towns. In each town, one at a time, a potential entrant decides whether to open up against them. The incumbent can fight — slash prices, take a loss, make entry painful — or accommodate and share the market at healthy margins.
Fighting costs the incumbent money. Accommodating is always more profitable in that town. So by backward induction: in town twenty, the last one, fighting is pointless — no future entrants left to impress, so accommodate. Which means in town nineteen, the entrant knows the incumbent will accommodate in twenty, so fighting in nineteen buys no deterrent value either. Accommodate. Roll that logic all the way back and the maths says: never fight, anywhere, ever.
Selten's point — the paradox — is that almost nobody actually believes that. In practice, incumbents do fight early, entrants are deterred, and the reputation strategy works. The game-theoretic solution and observed behaviour flatly disagree.
| Incumbent fights | Incumbent accommodates | |
|---|---|---|
| Entrant enters | Entrant loses money. Incumbent loses money too. | Both profitable. Market shared. |
| Entrant stays out | Incumbent keeps the market at full margin. | Incumbent keeps the market at full margin. |
The incumbent's best single-deal move is always to accommodate. The incumbent's best campaign move is to fight the first few loudly enough that entrants eight through twenty never bother showing up.
What This Looks Like in SaaS
Strip the towns and it's a competitor's deal-by-deal pricing behaviour:
- Fighting = aggressive discounting, sudden multi-year lock-ins, free migration, a "competitive displacement" SKU that appears only when your name is on the deal.
- Accommodating = list-price discipline, letting the deal go, competing on fit rather than cost.
- The reputation payoff = every future prospect who hears "they'll just get undercut anyway" and never runs the evaluation at all.
The resolution to the paradox, incidentally, is incomplete information. Selten's backward induction only holds if entrants know for certain the incumbent is rational and profit-maximising in each market. Introduce the smallest doubt — maybe this incumbent is the type that fights on principle, maybe they have a war chest, maybe the CEO takes it personally — and fighting early becomes rational again, because it's cheap to feed that doubt.
Which means the entire game is played on the terrain of information. The incumbent's weapon isn't the discount. It's your uncertainty about whether the discount is real, repeatable, and sustainable.
The Three Questions That Decide Your Response
When a competitor discounts hard against you, you are being asked to infer something about their type. Get the inference wrong in either direction and it's expensive: match a bluff and you've torched your own margin for nothing; ignore a real fight and you lose the segment.
- Is it a pattern or a one-off? One savage discount is noise. The same discount shape across six deals in one segment is a deliberate fight, and it tells you exactly which segment they've decided to defend.
- Is it sustainable? A fight funded by a fresh raise behaves very differently from one funded by margin they don't have. Fighting is costly for them too — that's the whole structure of the game. Incumbents who can't afford it stop.
- Is it aimed at the deal or at the market? If the discount is bundled with public repositioning, a new entry-tier SKU, or displacement messaging on their site, they're buying reputation, not revenue. That's a campaign, and it needs a campaign response — not a per-deal concession.
Framework to Steal: The Fight-or-Bluff Read
Before you match a competitor's price, run this:
- Aggregate the evidence. Pull every deal in the last two quarters where this competitor discounted. Look at the shape, not the size.
- Segment it. Which customer profiles do they fight for, and which do they quietly let go? The ones they release are where your margin lives.
- Cost their fight. Estimate what that discount does to their unit economics. Ask how long they can run it.
- Choose your terrain deliberately. Concede the towns they're defending hardest if the economics are bad, and compete openly where they've shown they won't fight. Matching everywhere is how you fund their reputation with your margin.
Where Pro-Position.AI Fits
Every one of those four steps is an intelligence problem, not a pricing problem. You can't tell a pattern from a one-off if your evidence is three reps' memories of deals they lost. You can't cost a rival's fight if you don't know when their pricing page or packaging last moved. And the incumbent's advantage in the Chain Store Game depends on you not being able to tell.
[Pro-Position.AI](
Selten's paradox resolves on incomplete information. Competitive intelligence is simply the work of making the information less incomplete — and taking away the only weapon that makes the bluff work.
First published as an edition of The Theory, Pro-Position.AI's newsletter on competitive intelligence, product strategy and go-to-market.
Keep exploring: