Four stages. One rule.
Every signal is about an organisation, never a person, and shows the public page and date it came from. Here is what happens between a list of organisations and the file you receive.
From a list to a file
Nothing in it sends an email or contacts anyone. The outreach, if any, is yours.
- 01
List
We build the list of organisations in a category from open, public sources, or start from a list of company domains you give us.
- 02
Read
We read what each organisation publishes about itself, from sources allowed one by one. Each finding becomes one signal, with the page, the sentence and the date.
- 03
Re-check
Open signals are re-read on a schedule. A signal stays while its evidence still says what it said, and expires when the evidence is gone.
- 04
Deliver
You get every open signal as CSV and JSONL, best first. An organisation that has asked us to stop is never in it.
The page, the sentence, the date, and a copy
A signal cannot be stored without the public page it came from, and the quoted sentence must appear on that page word for word. Behind every signal we keep a saved copy of what we read, so the claim can be checked after the page changes or disappears.
Each signal carries a confidence (high, medium or low) with the reason for it, so you know how much weight to give it.
When a signal expires
Only on positive proof that its evidence is gone. Expired signals are kept as history, never silently deleted.
What expires a signal
- The record it cites no longer exists, or no longer says what was quoted.
- The page loads and the quoted sentence is no longer on it.
- The job posting has been taken down.
- The date that made it urgent has passed.
What leaves it alone, marked unverified
- A timeout, a server error, or a site that does not answer.
- A block, a bot challenge, or a login wall.
- A robots.txt rule that now disallows the page.
A page we could not read is not a page that stopped saying something.
An organisation that asks us to stop is never read again
One email, no reason needed. The domain and its subdomains are never requested again, the pages we saved from it are deleted, and nothing about it is delivered to anyone. How to ask: our crawler’s page.
What Signalkeen does not do
Said here so nobody has to ask twice.
- It does not find, store or infer a person. Finding whom to write to is yours, under your own lawful basis.
- It does not read LinkedIn, in any form.
- It does not log in, solve a challenge, disguise itself or route around a block. A refusal is recorded as a refusal.
- It does not promise that every signal is right. It promises that every signal shows its evidence, and says how confident it is.
- It does not see everything. Honouring refusals means a vendor that ignores them will always see more sites. We sell evidence, not breadth.
Want to see a real sample?
Tell us your category and we will show you what the signals would look like for it.