What Qravitas sees on company websites, free for anyone to use
Every reading Qravitas makes of a public company home page: the technologies it loads, whether it has analytics, whether campaign tags survive a redirect, what machine-readable markup it carries. No key, no sign-up, no paywall. Because every company is read the same way, the readings can be compared, checked and built on. Scores and findings stay in the paid product; the observations do not.
Get it
| What | Address |
|---|---|
| Everything, one company per line | /data/v1/companies.ndjson |
| Everything, for spreadsheets | /data/v1/companies.csv |
| Look up one company, and check it | /data/company.html |
| One company, as JSON | /data/v1/c/{slug}.json |
| What changed between readings (0 changes since 2026-11-02) | /data/v1/changes.ndjson |
| Field dictionary and limits | /data/v1/schema.json |
| Version, counts, refresh | /data/v1/manifest.json |
| Find a company's slug | /lookup.json |
| Aggregate statistics | /insights/qravitas-data.json |
Static files, served from a CDN. Cross-origin requests are allowed, so a browser app can read them directly.
What a record looks like
A real record: ITC Ltd. A technology we detect is something the page loaded, not proof of how it is used.
What is not here, and why
- Scores, indices, verdicts and findings. These are the judgement Qravitas sells. The observations they rest on are open.
- Competitor and peer links. We sampled the links we hold and too many are wrong to publish as evidence. Tell us a company's real competitors when you check a record, and that is how this layer gets fixed.
- Sources whose redistribution terms are not cleared. Real-visitor speed data, ad-library items, news and social are left out until each source's terms are confirmed to allow republishing. Access to a public source is not the same as the right to republish it.
- Companies whose website we cannot show is their own. Reading the wrong site and attributing it to a company is the worst error this dataset could make, so those companies are listed with observations withheld, and the reason:
| Why observations are withheld | Companies |
|---|---|
| we could not show the website is the company's own | 1,016 |
| the website did not resolve when checked | 372 |
| the home page was not read successfully | 136 |
| the company's website changed after the reading | 12 |
How this differs from the Insights figures
The statistics on Qravitas Data apply stricter rules (a minimum sample, only recent readings, only high-confidence websites), so they can differ by a point or two from what you compute here. Use the Insights figures when you quote a statistic. Use this dataset to check them, to look up one company, or to run your own cut.
Cite it, correct it
Licensed CC BY 4.0: use it, change it, sell what you build, with attribution and a link. Each record has its own observed_at; a reading is a moment in time. If a record is wrong, write to hello@qravitas.com with the company and what you saw. Confirmed errors are corrected in the next weekly build, and an entry can be removed at the company's request.
Check it against your own data
The point of an open evidence base is that it can be tested. Look up any company, say whether the reading matches what you know, and name its real competitors. No account needed. Each check is reviewed before it changes a record, and anything published from checks is anonymised and aggregated.
Comparing your own first-party numbers against these readings, including what you pay for media, is planned and not available yet. It will be opt-in and anonymised, with minimum group sizes. Please do not send confidential figures through the check form.
Observations, not assessments. Not legal, financial or investment advice. Method: Methodology.