How to do a competitor analysis for a startup

Your competitor list is longer and less glamorous than you think. Build it in four groups:

Direct — products that do roughly what you'd do. Usually three to eight, and the easiest to find.

Adjacent — products solving the neighbouring problem that could extend into yours in a quarter if they wanted to. Frequently the greater threat.

Manual — the spreadsheet, the VA, the agency, the intern. A £60/month VA doing this by hand is a competitor with an existing budget line and a human relationship you have to displace.

Nothing — living with the problem. Almost always the market leader by volume, and the one nobody puts on the slide. If most of your prospects currently do nothing, your real fight is against inertia, not against a product, and that changes everything about how you sell.

Find them by searching your problem in the customer's words, reading "best X tools" roundups, checking what competitors bid on in ads, and — most reliably — asking the ten people you interviewed what they use now.

Not a trial you click through. Pay, and use it for something you actually need done. A day of this beats a week of reading their site.

Record as you go:

• Time to first value. From signup to the moment it did something useful. This is the number most products are worst at, and the easiest place to beat them.
• Where you got stuck. Every point of confusion is a place their customers churn.
• What surprised you positively. Be honest here; the thing you didn't expect them to do well is usually their moat, and you need to know it.
• What they clearly don't care about. Ignored surfaces reveal strategy more reliably than roadmaps do.
• The emails they send. Their onboarding sequence is their conversion thinking, written down and handed to you free.

Then cancel, and watch the cancellation flow. What they offer to keep you tells you their margins, their churn anxiety, and sometimes their real price.

G2, Capterra, Trustpilot, the app stores, and the subreddit where their users complain. Sort ascending. Read fifty.

You're not collecting complaints — you're looking for the repeated structural one. Ignore the single furious review about a billing error; find the sentence that appears twenty times in different words.

Sort what you find into two piles:

Fixable — bugs, a bad onboarding flow, slow support. They will fix these eventually. Not a wedge.

Structural — consequences of a deliberate decision. Too complex for solo users because they sold to enterprise. No API because they're protecting a services business. Expensive because their cost base assumes account managers. These they cannot fix without breaking their own business, and that is exactly the gap you can live in.

Read the five-star reviews too, and take them seriously. What people love is what you'd have to match or explicitly decline to match.

A pricing page is a strategy document that has to be public. Read it as one.

For each competitor, note the price, the value metric, what's in the cheapest tier, what forces the upgrade, and who's excluded by the entry price.

The value metric tells you what they believe drives value — a conclusion they've tested with far more customers than you have. Worth taking seriously even when you disagree.

The upgrade trigger tells you where their revenue really comes from, and therefore who they're actually built for regardless of what the homepage says.

The entry price tells you who they've written off. A £200/month floor means every solo user and small team is unserved by choice, not by accident. That population is your market, and their sales model prevents them from chasing it.

Run your own positioning through the free validator below once you've written it. It'll name the risk in your wedge, which after a day inside a competitor's product is exactly the thing you'll be least able to see.

The output of a teardown is not a table. It's one sentence:

"[Specific person] should choose us over [competitor] because [specific structural reason], and should choose them over us if [honest condition]."

Both halves are required. The second half is what makes the first believable — and it's what makes the sentence usable by a salesperson or a landing page without sounding like a claim.

Example: "A solo founder validating their first idea should choose us over an enterprise research platform because we return a verdict in an hour rather than a quarter and cost a fraction of a seat licence; they should choose the enterprise platform if they need procurement-approved vendor status and a named analyst."

That sentence is worth more than a twelve-row feature grid, and it's the thing you can actually put on a page.

Then set a reminder to redo this in six months. Competitors move, and a teardown from last year is confidently wrong in ways you won't notice until a prospect corrects you on a call.