Four companies already do roughly what you are planning, all positioned in more or less the same place. You read that as proof the market exists.
It is proof of something narrower. They clustered because clustering was individually rational, and none of them chose the middle because it was underserved.
Why these three models
The decision is where to position on entry, and whether to enter at all. The features that fire are competitors choosing locations in a space of customer preferences, a return distribution with no typical case, and a market position that accumulates rather than being announced.
Three lenses. Spatial entry produces an equilibrium answer about why the middle is crowded and what your arrival does to it. The return distribution produces a random answer about what winning is actually worth. Stock and flow produces a complex answer about the gap between the position you claim and the one you hold. The first two disagree usefully: the first says the middle is where the customers are, the second says the middle is where the returns are not.
1. Why everyone is standing in the same place
The classical result is unintuitive the first time and explains a great deal of what you observe.
Model customers as spread along a line of preferences and each choosing the nearest supplier. Two competitors choosing where to stand both move toward the centre, because each step inward takes customers from the other and loses none from the edge, where those customers have nowhere better to go. The stable outcome is both of them at the middle, which is a poor result for customers and a rational one for each firm.1
Two extensions matter more to a founder than the base case.
The first is what happens with more than two players, where clustering becomes unstable and positions start to matter again. The second is entry itself. Brams treats the entry problem directly in the political version, and the important insight transfers: a new entrant does not simply take a share of the existing distribution, they change where it is optimal for the incumbents to stand.2 Your arrival is not an addition to a static picture. It is a move that the others respond to, and their response is part of the outcome you are choosing.
So the crowded middle tells you where rational competitors ended up given each other. It does not tell you where the unserved customers are, and it certainly does not tell you the middle has room.
Spatial entry, the positioning lens
- Assumes: customer preferences can be arranged in a space and buyers choose the nearest acceptable option.
- Fits because: the existing competitors are differentiated along dimensions you could also move on.
- Breaks when: quality dominates position, so that buyers choose the best option rather than the nearest one, which collapses the whole model.
- Evidence: grade B plus. Classical and well understood, with the strong assumption that preferences are single-peaked along a usable dimension.
- Counteracts: reading a crowded category as validated demand.
- May reinforce: differentiating on a dimension nobody actually chooses by.
2. What winning is worth, which is not the average
The second lens changes what the entry decision is about.
Returns in most categories do not distribute evenly and do not have a typical case. A small number of positions absorb most of the value and the rest occupy a long thin tail. That is a different shape from the one an average implies, and planning against the average produces a plan for an outcome almost nobody gets.
Two consequences follow, and the second is the useful one.
A small share of a large category is usually worth much less than it sounds. Being the fifth player in a big market is a tail position, and the tail is thin precisely because the leading positions absorbed the value.
And the leading position in a narrow category can be worth more than the fifth position in a broad one, even though the second sounds more ambitious. That is the argument for entering somewhere specific and defensible rather than somewhere large, and it is an argument about the shape of the distribution rather than about focus as a virtue.
One honest caveat belongs in the body. Concentration is observable and real. The stronger claim, that such distributions follow a fitted power law with a specific exponent, is much weaker than its popular version: testing across nearly a thousand real networks found scale-free structure to be rare rather than universal.3 Use the shape, never a parameter.
Return distribution, the payoff-shape lens
- Assumes: value concentrates, so the mean is a poor guide to the typical outcome.
- Fits because: you are deciding whether a share of a large category is worth having.
- Breaks when: the market is genuinely fragmented and locally decided, where concentration never forms and the fifth position is fine.
- Evidence: grade B plus for concentration as a shape. Barred for any fitted exponent.
- Counteracts: planning against an average in a domain with no typical case.
- May reinforce: lottery-shaped strategy presented as focus.
3. The position you hold, not the one you announce
The third lens is the one that decides whether any of the above survives contact with execution.
A market position is a stock. It accumulates through customers who chose you for a reason they could name, through references, through the specific competence your team built, and it drains through churn, through drift, and through everyone else’s marketing. Like any stock it cannot be set by announcement.
That is why repositioning statements do so little. Changing the words on your site changes an input to a stock whose level was determined by years of prior flows. The position your buyers hold in their heads responds slowly, with a lag, and largely to what you actually deliver.
The operational version is that entering a differentiated position is not a decision, it is a construction project with a duration. Ask which flows would have to run, for how long, before a buyer would describe you in the words you have chosen. If the construction takes longer than your runway, you have chosen a position you cannot reach. The correct move is a nearer one.
Expect this to be underweighted. Highly educated adults routinely fail to infer the behaviour of simple stock and flow systems, and the failure is not attributable to graph literacy, contextual knowledge, motivation or cognitive capacity.4
Stock and flow, the accumulation lens
- Assumes: a market position accumulates through identifiable flows and cannot be set directly.
- Fits because: you are choosing a position you would then have to build.
- Breaks when: the category is new enough that no position exists yet, where announcement genuinely can create one.
- Evidence: grade A. Structural, and the difficulty people have reasoning about it is well replicated.
- Counteracts: treating positioning as a messaging decision.
- May reinforce: refusing to move on a position that is genuinely wrong.
The levers, cheapest first
- Name the dimension buyers actually choose on. Not the one you would like them to. If you cannot name it, the spatial model has nothing to work with and neither do you.
- Ask what the incumbents do when you arrive. Your entry moves their optimal position. A plan that assumes they hold still is not a plan.
- Price the tail honestly. Write down what the fifth position in this category actually earns, not what the category is worth.
- Compare against the narrow alternative. Leading position in a small category against fifth in a large one, both as real numbers.
- Count the flows to the position. How many reference customers, over how long, before a buyer describes you the way you intend?
- Check that against runway. If the construction takes longer than the money, choose a nearer position.
What to do before committing
Do now, sized at one afternoon, effect immediate. Write the dimension buyers choose on, place the existing competitors on it, and place yourself. Reversible, free, and dominant across every scenario about whether the category has room.
Hedge, where the premium is the whole loss. Test the narrow position on one segment before committing the company to it. If the broad position was right you have spent a quarter on one segment, and that is the entire downside.
Defer and trigger, size fixed now. Do not commit the whole roadmap to a position. Pre-commit the trigger: if buyers in the test segment do not use your chosen words unprompted within a stated number of months, the position is not being built and the plan changes. Fix that number now, because judged later it will always feel like it needs one more quarter.
Note the arrivals. The map lands today. A position accumulates over quarters and is invisible in weeks, which is exactly the interval in which most repositioning gets abandoned.
What usually happens next
Run the break test first. Has a rule changed, has an actor entered or left, has a measurement become a target? An incumbent exiting changes the optimal position for everyone remaining, and a map drawn last year describes a different game.
If nothing broke, the pattern is consistent and expensive. The entrant positions at the edge, discovers that the edge is thin, drifts toward the middle over two years for entirely sensible reasons deal by deal, and arrives as the fifth undifferentiated player with a longer cost base than the incumbents. Nobody decides that. It is the accumulation of individually rational moves toward where the customers are.
There is a second regularity, and it is about who moves first. The incumbent nearest your chosen position is the one whose optimal location changes most when you arrive, so they are the one who responds, and they respond by moving toward you. A differentiated position is therefore not a permanent address. It is a lead measured in quarters, and the question is what you accumulate before it closes.
Subtract the counterfactual before crediting a position. If you won deals after repositioning, ask whether you won them on the position or on the reference customer you happened to land the same quarter. The test is whether buyers use your words back to you.
What this ensemble cannot see
All three lenses assume preferences can be laid out in a space you can see. In new categories they cannot.
Where buyers do not yet know what the category is, there is no dimension to position along, no incumbent map, and the spatial model has nothing to say. That is precisely the situation many founders are actually in, and this framework will encourage them to draw a map of a space that does not exist yet and then optimise a position within a fiction.
There is also a strong assumption in the central model that is worth resisting. Spatial competition assumes buyers choose the nearest acceptable option. Where quality genuinely dominates, buyers choose the best rather than the nearest, clustering does not occur, and the whole first section is inapplicable. Distinguishing those two is a judgement made by someone who wants to enter.
And one property none of these models contains: the incumbents have information you do not. Four companies clustered in the middle may have each independently discovered that the edges do not pay. Reading their clustering as a failure of imagination is comfortable, and it is sometimes wrong.
The one action that survives the ignorance: before committing, write down what the fifth position in this category earns in real money. If you cannot find that number, you do not know the shape of the distribution you are entering, and the confident part of your plan is the part with no evidence under it.
Who has to move
The person who needs this is whoever is about to commit the roadmap, usually with a competitive landscape slide that shows four logos and an empty quadrant. The cheapest first test is placing everyone on the dimension buyers actually choose by, rather than on the two axes that make your quadrant empty. It costs an afternoon and it frequently shows the quadrant is empty because nobody buys there.
Sources and notes
- Jeffrey Carpenter and Andrea Robbett, Game Theory and Behavior, MIT Press. Spatial voting, single-peaked preferences, candidate competition games and the Hotelling model of two competitors converging on the centre are chapter 21, along with the treatment of what happens with more competitors and the empirical observation that behaviour is less chaotic than the theory permits.
- Steven J. Brams, Rational Politics: Decisions, Games, and Strategy. Chapter 3 develops spatial models of candidate positioning, the median and the mean of a distribution of voter attitudes, and the entry problem, in which a new entrant alters the optimal positions of those already in the contest rather than simply dividing the existing support. The treatment of the entry problem draws on Brams and Straffin. Used here for the claim that entry changes the incumbents’ optimal positions.
- Anna D. Broido and Aaron Clauset, Scale-free networks are rare, Nature Communications 10, 1017, 2019. Preprint: https://arxiv.org/pdf/1801.03400. The abstract reports that across domains scale-free networks are rare, with only 4% exhibiting the strongest-possible evidence of scale-free structure and 52% the weakest. Cited here to justify using concentration as a shape and refusing any fitted exponent.
- Matthew A. Cronin, Cleotilde Gonzalez and John D. Sterman, Why don’t well-educated adults understand accumulation? A challenge to researchers, educators, and citizens, Organizational Behavior and Human Decision Processes 108(1), 2009, pages 116 to 130. Author copy: https://www.mit.edu/~jsterman/CroninGonzalezSterman061210.pdf. The abstract states that highly educated people are often unable to infer the behaviour of simple stock-flow systems, and that persistent poor performance is not attributable to an inability to interpret graphs, contextual knowledge, motivation, or cognitive capacity.
A note on the slide this argues against. The two-by-two competitive landscape with an empty quadrant is the standard artefact here, and it is constructed by choosing axes until a gap appears. The spatial model is the same idea done honestly: the axis has to be one buyers actually choose on, and it has to be drawn before you know where it puts you. Almost every empty quadrant survives only the first of those two conditions.
Joshua Agonya Pi’Rwot, Founder.