What it actually reads
A homepage sells a feeling. The pages that say who a company sells to are the customer list, the case studies, the industry pages and the pricing page. So the tool starts at the homepage, works out which of those exist, and reads up to three of them as well, in parallel. It reports which pages it read and how much text it found on each, so you can see whether the answer rests on nine thousand words or on eleven.
From that it pulls the industries they name, the company sizes their language implies, the markets they list, the buying roles they address, the compliance and integration claims they lead with, the customers their logo walls name, any published prices, and the social profiles they link to.
How to read the competitor profiles
They are inferences from marketing copy, not research, and the tool says so on the panel. Three things to keep in mind. Marketing copy overstates reach, so a competitor claiming every industry probably wins in two of them. A site that renders in JavaScript gives almost nothing to read, which is why confidence reflects how much text was actually readable and will say "thin" rather than dress up a guess. And some sites block automated requests outright, in which case you can fill the profile in by hand and carry on.
Every field is editable. The fastest way to use it is to let the tool draft the profile, then correct it from what you already know.
The market they are chasing
Once the profile is on screen, the tool sizes it: how many companies match what that competitor appears to target, across North America, Europe, LATAM, India, the Middle East, SEA and ANZ. It is a modelled estimate built from published business counts, not market research, and it moves as you correct the profile. Every base count is sourced in the panel, and the three regions with no comparable public source are marked so you know which numbers to lean on.
Two competitors sized side by side is usually the more interesting read. A rival chasing ten times your addressable market is not really your rival, and one chasing a tenth of it is probably beating you on focus.
What to do with it
The point is not to copy their targeting. It is to see where they are strong so you can decide where to fight and where not to. If their profile is the same as yours in every respect, you are competing on execution and price, which is the worst place to be. Somewhere in the segments they do not name is a position you can hold.
When you are ready to turn that into a ranked account list, that is the work we do: see building a target account list sales will not ignore and the ICP that builds your account list.
How to use the sizing
The number the tool ends on is not the interesting part. The shape of the narrowing is. If your list barely shrinks between the industry filter and the qualifier filter, your qualifiers are descriptions rather than qualifiers, and the list you hand sales will be exactly as unworkable as the one they are ignoring today.
If Tier 1 comes out larger than your reps can genuinely cover, cut it. One rep sustains roughly 100 to 150 worked accounts, and a target account list sized past that is a list nobody works. We wrote about that failure mode in detail in building a target account list sales will not ignore, and about the scoring model underneath it in the ICP that builds your account list.
What this deliberately does not do
It does not tell you which companies those accounts are. Sizing a market and naming the accounts in it are different jobs, and the second one needs real data rather than a model. It also does not decide whether you need software to run the resulting programme, which is usually the next question. The honest answer for most teams is no, and we set out when that changes in do you need ABM software and the 2026 platform buyer's guide.
Questions
Read what they say in public. A company's homepage, industry pages, customer list and pricing page tell you which segments they are built to win: the industries they name, the company sizes their language implies, the regions they list, the buying roles they address, and the integrations and compliance claims they lead with. This tool fetches those pages, matches the language to a fixed taxonomy, and shows you the phrase behind every conclusion so you can judge it yourself.
An ideal customer profile is a description of the type of company most likely to buy, stay and expand, defined by firmographics such as industry, employee count and region, plus the qualifying conditions that separate a genuine fit from a company that merely looks right. It describes accounts, not individual buyers, which is what makes it usable for account-based marketing.
Partly, and only as a starting point. Marketing copy overstates reach, and a homepage that renders in JavaScript gives almost nothing to read. That is why this tool reports its confidence honestly, shows the exact phrase behind every inference, lists which pages it managed to read, and lets you correct any field. Treat it as a first draft that saves you an hour, not as research.
The homepage first, then up to three of the pages most likely to state who a company sells to: the customer list, case studies, industry or solutions pages, and pricing. A homepage sells a feeling; those pages name names. The tool reports which ones it read and how much text it found on each.
Each teardown makes several requests to the competitor's site on our server, so the tool is gated to stop anonymous abuse. One work email unlocks it for the session and every teardown after the first is instant. Personal and disposable addresses are not accepted.
No. Sites that render entirely in JavaScript return very little readable text, and some block automated requests outright. When that happens the tool says so plainly rather than inventing a profile, and you can still correct or complete the fields by hand.
The tool estimates it bottom up: businesses in the markets and size bands that competitor targets, multiplied by the share of the business population in the industries they name. It is a modelled estimate from published business counts, not market research, and every base count is sourced in the methodology panel so you can challenge it.
Yes. It costs one work email per session and nothing else. There is no trial, no card and no usage charge.