Use These Prompts to Analyse a Competitor's Website

Almost every teardown starts by copying pages into a chat window, and the copying throws away the more useful half of the website. Text survives. What pages exist, how many of each kind, which ones the navigation promotes, how far each sits from the front door, and what stands behind a form do not.

The test is easy to run in your head. Take two competitors whose homepages say almost exactly the same thing, which in most categories is most of them. One has forty integration pages, a comparison page for every rival, and nothing written by industry. The other has three integration pages and twelve industry pages. Their copy is interchangeable, and they are not selling the same way at all. No amount of reading the copy gets that back.

So the structure is a separate input, and it goes in first. Not because structure matters more than what a company says. Because the shape of a site was decided by people arguing over scarce space, and a paragraph of copy can be changed by one person in an afternoon. The copy is still worth reading on its own terms, and a positioning teardown quotes it rather than summarising the structure around it.

Two inputs, and the second one is why this page exists. Nearly everybody supplies the first on its own, then asks a question the first cannot answer.

The capture date belongs at the top of each block rather than in your notes. A teardown gets passed around, and six weeks later somebody will disagree with a line in it. If the date is in the document, that disagreement is a useful finding about a page that changed. If it is not, the argument becomes whether your analysis was any good.

Four passes, in the order the evidence supports. The teardown reads the site as a journey. The inventory asks what it is made of. The depth map asks what is being promoted. The gate audit asks what they want in exchange, and that is the least edited statement on any website.

The inventory and the depth map look alike and answer opposite questions. A comparison page that exists, sits three clicks down and is linked from nothing, is a page built for search. The same page in the main navigation is a company that wants the comparison made in front of them. Which of the two you are looking at changes what you do about it, and a list of URLs alone will never tell you. What the pages are built with is a separate read again, and a tech stack analysis answers it from different evidence.

We asked what sits in a real company’s main navigation, then opened the menu and compared.

A coherent answer is harder to doubt than a vague one

Nothing in the first answer reads like a guess. Six product names in one consistent family, arranged around a free tier, with the right top-level sections beside them. That is the problem. A reader looking for signs of invention finds none, because there were none. The suite described was real. It is just not the one on the menu now.

Worth separating, too, is what the run cannot settle. Three names have gone, and three names are there that were not before. Whether those are renames of the same products, or different products entirely, is not something either answer establishes. That is the right output. A teardown should hand you that question with today’s menu attached, rather than settle it from memory in a direction nobody can check.

A navigation slot is the most expensive space on a website. Everything in one beat something else, in a meeting, against people who disagreed. That makes the menu the closest thing a company publishes to a statement of how it has decided to organise selling.

The grounded run above turned that into something concrete. The menu offers to browse by team size, and the split has exactly two options. That is a company naming its own boundary, before a single page has made an argument. It takes an afternoon to work out from case studies and thirty seconds to read off a menu.

Read the labels the menu uses, not the ones you would use

A menu that says Solutions and then splits by use case has decided buyers arrive with a problem. One that splits by role has decided they arrive as a job title. One that splits by industry has decided the product is bought differently in each. Those are three different companies, and you can see the difference before reading a word of the copy underneath.

What a navigation leaves out matters too, and it is the easiest thing on this job to over-read. That is why it belongs to the positioning teardown rather than here. A missing industry page is evidence of money not spent, not of a segment being ignored, and the two get confused constantly. Structure answers what a company has built. It does not answer what they intend. A missing industry page is also a statement about who they sell to, and the packaging usually confirms or contradicts it within one screen.

Every other part of a website is written to persuade. A form is written to qualify. It is the one place where the commercial model underneath shows through undressed. A gate marks the exact point at which a company has decided a visitor is worth a conversation.

Read the whole set rather than any single form, and three things fall out. How early the first gate appears says how confident they are that what comes next is worth the friction. What the form asks for is a routing rule made visible. A field for company size and job title means the submission gets sorted before anybody replies. An email box on its own is a volume play. And what they promise in exchange says what they think an email address is worth.

What is open matters as much as what is closed

The useful cases are the ones that break the category habit in either direction. A vendor publishing something everybody else gates has usually decided that being findable beats being asked. That is a real position, with real consequences for how they get compared. A vendor gating something the category gives away is usually protecting a sales motion that depends on the conversation happening. Worth knowing before you decide to match them.

None of this says anything about results. A gate tells you what a company decided, not whether the decision worked. How many people filled the form in is not on the page, and never will be. What the traffic did is a separate measurement again, and share of search is the closest public proxy for whether any of it worked.

Nothing on a website records what it replaced. A promise gets rewritten, a plan name changes, an industry page appears, a comparison page is quietly retired. None of it produces a date, a note or a notification. The first sign is usually somebody asking why a quote in the teardown does not match the page.

Which makes the re-run interval unanswerable. Monthly is too often for a homepage and far too slow for a page inventory. The only version that works is being told when a specific page changes, then re-running the pass that page belongs to rather than the whole teardown.

Where the output lands decides whether any of this compounds. A dated structural section in a competitor teardown document can be compared against the last one. That is the only way a page inventory ever becomes a trend rather than a snapshot.

Flares watches competitor websites and reports what changed, with the date and the wording it replaced. A rewrite then arrives as an event, rather than as a mismatch somebody notices later. What the change means for your own positioning is still a judgement, and it is the part worth spending the time you get back on. The technologies behind those pages move on their own schedule, and website technology is where a rebuild shows up before the copy does.