You can see their ads, not their results

Fourteen tabs, all of them somebody else's. Their homepage, their pricing page, their ad library, three of their landing pages, a screenshot folder filling up with headlines. It is one of the most satisfying afternoons in marketing, and most of us have had it more than once. You close the laptop with a plan — and every decision in that plan was made by another company, for reasons you never saw. The one thing you actually needed to know, whether any of it worked for them, was not on any of those screens.

Start with what a competitor's ad genuinely tells you. It tells you the ad exists. It tells you somebody approved it, and if it has been running a while, that it probably is not an outright catastrophe. That is the entire list. It does not tell you what a signup costs them, whether the campaign is quietly bleeding money inside a budget large enough not to notice, or whether it has simply been running untouched since the person who set it up left. Public ad archives show you what is live, never what it returns. You are reading the visible half of a decision and inferring the invisible half, which is the definition of guessing with extra steps.

Then there is the sampling problem, which is worse than it first appears. You do not study a random competitor — you study the one you keep seeing, and being seen is what a large budget buys. So the company you learn from is systematically the one whose economics least resemble yours. Hypothetically: if a customer is worth four thousand dollars to them over three years and sixty dollars to you, they can pay a hundred dollars for a click that would ruin your month, and they should. Their landing page is built for a sales call it hands off to. Their patient, expensive brand campaign makes sense on a five-year horizon. Copy the tactic, inherit none of the conditions that made it rational.

their headline was written for their economics, not yours.

The copy trap is the specific version of this that catches everyone. A good headline is the output of things you cannot see from outside: who they interviewed, what their support inbox told them, four rewrites, one strong opinion from a person who has since moved on. When you paste that sentence onto your page with the nouns swapped, you import the conclusion and leave the reasoning behind. Worse, it may not have been a conclusion at all — plenty of live headlines are somebody's untested first draft that nobody got around to changing.

There is a second cost that shows up only when you look at the category as a whole. Once three products in a space all promise the fastest way to do the same thing, that sentence has stopped carrying information. It reads as the category, not as you. A visitor comparing four tabs of near-identical promises does not pick the best one; they get tired and pick the cheapest, or they leave. Sounding like everyone else is not a neutral choice — it hands the decision to price.

Drafting tools make this easier to do accidentally, ours included. Ask any model to write your landing page in the style of the leading products in your space and you will get something fluent, competent and almost exactly average, because averaging the visible examples is precisely what it is built to do. That output is still useful — read it as a map of the conventions, the things everyone says, the shape a reader expects. Then decide, deliberately, which convention you are going to break, and check every number in it before you approve it, because a confident draft with an invented statistic is the failure worth catching every single time.

Competitor research is not worthless. It is just good at different things than people use it for. Three of them are real. The first is vocabulary — the words customers actually use, which you find in their reviews and support threads, not on their homepage. A homepage tells you what a marketing team wishes people said. The second is the shape of the category: what everyone promises tells you what is table stakes, and anything you have to explain from scratch is either a genuine differentiator or a mistake. The third is gaps. The complaint that shows up in every one-star review of the incumbent is a positioning statement someone else wrote for you for free.

Notice that all three come from what customers say, not from what competitors publish. That is the whole distinction. Their reviews are evidence; their landing page is a claim. And the evidence you can get about your own product beats either one, because you can see both halves of it — the thing you shipped and what it did. Five conversations with people who nearly bought and did not will beat a quarter of competitive analysis, and your own worst-performing page teaches you more than their best one, since you know what it was supposed to do. When you cannot run a clean experiment — and at small volumes [you usually cannot](/blog/too-small-to-a-b-test) — first-hand evidence is what is left, and it is a lot stronger than a screenshot.

So give the research a box. An hour a quarter, a written page of what you learned, then close the tabs. And keep one check for the moments in between: before you change something because a competitor does it, finish the sentence "we are doing this because ___". If the only thing you can put in the blank is "they are", stop — you are about to spend real time on a hunch you borrowed. Hold onto [the one number you steer by](/blog/one-goal-one-number) instead, because it is measuring you.

Here is the question worth sitting with, then. If you could not see any of them — no ad library, no screenshots, no pricing page open in tab nine — what would you build your marketing out of? Whatever you just answered is the part that is actually yours, and it is probably where the next real improvement is hiding.