You only know once every assumption about your strongest point has a source of evidence that someone else can verify. As long as "what we win on" only lives in the head of management or the sales leadership, it's an opinion with a stamp on it, and opinions diverge the moment you put them side by side.
Ask around in a company of 50 to 300 employees and you get three different answers: sales says price, marketing says brand, management says service. All three can be right about part of the market and all three wrong about the whole, because none of the three has systematically set the deals of the last twelve months against the peer group. The answer that remains is usually the answer of the loudest department, not the answer that comes from the deals themselves.
A dimension is an axis on which customers in your market actually choose: delivery time, technical depth, price, aftercare, brand awareness, flexibility in customization. A company can differ from the rest on a dimension without that difference ever deciding a deal, and that distinction between "different" and "winning" is exactly where most internal discussions get stuck. How you test that distinction is worked out in how do you test whether your distinctive capability really exists; only after that test do you know whether a strong point is also a winning point.
A score on a dimension without a source is an opinion in table form. An evidence matrix links each score to the place it comes from: a published SLA, a rates page, a job posting that says something about capacity, a customer review, a lost or won deal with a concrete reason attached. At a technical company with 120 employees, this can reveal that the presumed win on "speed of implementation" in the matrix comes down to two anecdotes from three years ago, while the source for "technical support" consists of ten independent reviews and a consistent pattern in deals lost to the competitor. That is the moment an assumption becomes a fact or collapses.
The matrix is only reliable once it's placed next to the right opponents. A comparison with the market leader says little if your real competition in the tender is three regional players of comparable size; how you define that group is described in what is a peer group and how do you assemble it. The problem after that is that many of those players don't publish figures on turnaround time, margin or customer satisfaction. There are methods to still obtain usable signals, worked out in how do I get information about competitors who don't publish figures; think of job postings, tender documents, reviews and the way a competitor describes its own services.
Once the matrix is complete, a second question often arises: if this is what we win on, then where are the gaps that no one in the peer group covers well? That is the domain of what is a white space analysis, and those two questions — what do we win on, where does no one win well yet — belong together. A company that knows its winning dimension without seeing the white space defends a position without knowing whether there's a bigger one to conquer.
Before a full matrix is in place, there's a faster entry point: the free eight-question loss-on-price check shows which dimension is leaking the most right now, so where lost deals are likely coming from. That is no substitute for the matrix, but a first direction that is often enough to see whether the problem lies with price, with another dimension, or with missing evidence.
Closing a gap that the matrix exposes is execution work: it touches processes, people and systems, and changing that costs time and money. What that change roughly requires and which part of it can be carried by AI is shown by the work scan on ftetoai.com, as a next step once you know what you really win on.