guide
Consent-first talent data, and why sourcing needs an audit trail
Most candidate databases are assembled by scraping, and the people in them never agreed to be there, cannot see what is held about them, and cannot get it corrected or removed. That is a compliance exposure and an accuracy problem at once — stale scraped records are wrong records. The alternative is not less data; it is data with consent, exposure control and an audit trail attached.
updated
A public page is not a consent record
"It was publicly visible" describes how data was obtained, not what someone agreed to. A person who published a profile on one platform for one purpose has not agreed to sit in a third party's permanent database, and once copied they lose the two things that made publication acceptable: the ability to correct it and the ability to take it down.
This is why scraped datasets decay in exactly the way that hurts most. The record persists while the person moves on, and the recruiter contacts them about a job they left two years ago.
Exposure belongs to the person, field by field
Consent expressed as a single yes-or-no checkbox is too coarse to be meaningful, because people are not equally willing to publish every fact about themselves. On Humetric every field carries its own exposure setting, and the profile as a whole can be fully private, reachable only through the authenticated recruiter API, or fully public — with changes taking effect on every surface at once.
Contact details are the deliberate exception to any publicity setting: email and phone are hidden by default and no setting can expose them. What a profile can publish instead is a contact-willingness signal — open to offers, passive, or not looking — which tells a recruiter whether to approach without handing over an address.
Logging is what makes consent checkable
Without a log, "we have consent" is an assertion nobody can verify — including the person who supposedly gave it. Humetric writes every public profile read and every API result to a consent log recording who asked, which fields were returned, and for what purpose.
That record is what turns a policy into something auditable. It also sets the boundary for agents: read the stated facts, do not infer fields a person has hidden, do not re-identify an anonymous profile, and do not contact someone who has marked themselves not looking.
Revocation has to reach every surface, including the caches
Consent that cannot be withdrawn is not consent. Withdrawing it on Humetric removes the profile from the public directory, from the sitemap and from the API — but the genuinely hard part is everything downstream that already holds a copy.
Cache lifetime is the honest measure here, because a cached projection is a withdrawal being ignored for as long as it lives. Humetric's public profile API is cached for sixty seconds with no stale-while-revalidate window, which bounds the worst case at one minute rather than the hours a conventional CDN policy would allow. Any team publishing people data should be able to state that number.
Key takeaways
- A publicly visible page is not a consent record.
- Make exposure a per-field setting, and keep contact details hidden by default.
- Log every read: who asked, which fields, for what purpose.
- Make withdrawal propagate everywhere, and know your cache window.
Questions about this guide
What is a consent log?
A consent log is an append-only record of every access to someone's data: who requested it, which fields were returned, and for what stated purpose. It is what makes a consent claim auditable after the fact rather than a policy statement nobody can check.
Is scraping public profiles allowed?
It depends on the jurisdiction, the platform's terms and what you then do with the data, so it is a question for your counsel rather than a settled yes or no. The design answer is different: consented, revocable data avoids the question and stays accurate, which scraped data does not.
Can someone remove their data from Humetric?
Yes. Consent is revocable, and withdrawing it removes the profile from the public directory, the sitemap and the API. The public profile API caches for sixty seconds, so that is the upper bound on how long a withdrawn projection can still be served.
Keep reading
Publish a profile machines can actually read.
Submitting a profile is free. You choose what is public field by field, contact details stay hidden, and every read is logged.