How to build a competitor set you will actually learn from

The most consequential decision in a competitive audit is made before any research happens, and it is usually made carelessly.

Who is in the set?

Get it wrong and everything downstream is technically accurate and strategically useless. You will have a rigorous, evidence-backed, carefully scored analysis of a market you do not actually compete in.

The three wrong rosters

The companies you admire

This is the most common failure and the most understandable one. Somebody names the impressive players in the category: the national brand, the well-funded one, the one with the good website everybody references.

They may be genuinely excellent and they may be entirely irrelevant to your commercial reality. If a buyer choosing between you and them is a conversation that has never happened, studying them tells you what good looks like in the abstract and nothing about why you lost last quarter.

There is a use for this. It is called aspiration, and it belongs in a different document.

The companies a database says are similar

Pull a competitor list from a keyword tool or a firmographic database and you get companies that resemble you on the dimensions the tool can measure. Similar size, overlapping keywords, same SIC code.

Overlapping keywords is the seductive one, because it feels like market evidence. It is not. It tells you who is bidding on similar terms, which includes lead aggregators, national franchises running broad-match campaigns, adjacent categories, and companies who target your terms without ever reaching your actual buyers.

A lookalike roster produces a lookalike finding. It will be internally consistent and it will not explain anything you care about.

The companies your sales team names from memory

Closer, and still biased. Sales teams remember the losses that stung: the dramatic one, the incumbent, the one where the buyer was rude about it. Recall is weighted by emotion rather than frequency.

Ask a sales team who you compete with and you will get four names. Look at the actual pipeline and there will be nine, and two of the four will not appear at all.

The right roster

The five to ten companies you actually lose deals to.

Not admire. Not resemble. Lose to. The companies that were in the room when a buyer chose somebody else.

That definition is narrow on purpose. It is also the only definition that makes the resulting findings actionable, because it is the only one anchored to a real decision a real buyer made.

How to build it properly

Start from lost deals, not from the market

Go through the last four quarters of losses. For each one, who won, or who else was being considered? If your CRM captures competitor on close, use it. If it does not, this is the single strongest argument for starting.

Count frequency. Not deal value, not memorability. How often did each name appear.

Weight by where you were actually beaten

There is a difference between losing a deal and losing it to somebody. Deals lost to no decision, to budget, or to an internal build are not competitive losses and should not put a company in your set.

Ask specifically: was there another vendor, and who?

Add the ones your buyers name unprompted

Some competitors never show up in a loss because you never get into the deal. They are chosen before you are contacted.

These surface in a different place: in what buyers say during discovery, in the alternatives they mention, in the language they use that they clearly picked up somewhere else. Those references are worth tracking, and they frequently reveal a competitor the pipeline data cannot see.

Include the uncomfortable one

Almost every roster has a company somebody does not want on it. Usually because they are winning and nobody enjoys studying that, or because they are considered beneath you and it is embarrassing that they take deals.

Both belong in the set, and the second one especially. A cheaper, less sophisticated competitor who is taking volume from you is telling you something specific about a segment of your market that you are mispricing or mis-serving. Excluding them because they are not a peer is how you find out about it two years later.

Cap it at ten

Beyond ten, the depth per company drops below what is useful, and depth is what makes a competitive audit worth anything.

This is not an arbitrary preference. If you have an evidence floor in your scoring, and you should, then thin research structurally caps a company’s achievable score. A roster too large to research properly does not produce a broader picture. It produces a picture that is systematically wrong about whoever you ran out of time on.

Five well-researched competitors beat fifteen skimmed ones. Every time.

The depth problem, which is where most audits actually fail

Once the roster is set, the failure mode moves. It stops being about who is in the set and becomes about how evenly you researched them.

This one is nearly invisible from the outside, and it produces confident, well-evidenced, completely misleading output.

Here is how it happens. You research your client thoroughly, because you have access, context and time. You research each competitor from the outside, more quickly. Your scoring standard requires several strong pieces of proof to award a top tier.

Now the client can reach a top tier and the competitors structurally cannot, because the evidence needed to award it was never gathered. The grid does not measure the market. It measures how hard you looked.

The output is a client leading every category, which looks like a triumph and is actually a bug.

We know the shape of that failure because we produced it. The fix was to measure evidence depth across the cohort automatically and refuse to publish a report as final until the field is researched to comparable depth. Researched properly, the client dropped from first to second.

That gate matters more than the roster, and the roster is what makes the gate possible. You cannot research ten companies to equal depth if you picked twenty-five.

The question to ask before you start

One question, asked of whoever will act on the audit:

If this report says we are behind, which companies would we accept that from?

The names that come back are your roster. Companies your organisation will dismiss the moment the finding is unwelcome do not belong in it, because a finding nobody accepts is not a finding.

That is a slightly political test rather than an analytical one, and it is the difference between an audit that changes something and an audit that gets filed.

More on how the cohort is scored once it is set: competitive intelligence.