Verification under enterprise procurement
Enterprise buyers rarely lose a verification deal on price. They lose it on the questionnaire: where the data goes, how long it stays, who can delete it, and what happens when a regulator asks.
Being new cuts both ways here. There is no long incident history to explain, and there is also no long track record to point at. Both facts belong on the table before procurement finds them.
The questions that stall an approval
Retention is usually first. Our default window is thirty days and the floor is zero, which means results are deleted once you have downloaded them. It is a per-key setting, so a stricter configuration on production and a longer one on a test key is a normal arrangement rather than an exception.
Erasure is next. A POST to the erasure endpoint removes an address from every batch on the account and returns a hash of it as proof, which is what a privacy log needs. We keep the hash rather than the address, because holding a plaintext list of people who asked to be forgotten would be a strange way to honor the request.
Then the question nobody writes down but everybody wants answered: what does this product not do. Catch-all domains cannot be resolved by anyone. Pristine spam traps cannot be detected. We publish no accuracy percentage because we have not measured one. A vendor whose answer to all three is a number is telling you something about how they will answer the next question too.
How the work gets done
The order that matters for enterprise, rather than a generic checklist.
Start with a paid sample, not a trial pool
Run twenty thousand addresses from a real list. The thing to compare across vendors is not the valid count, it is how many came back unresolved, because that is the number each vendor has an incentive to hide.
Pin retention before the pilot, not after
Set retention_days on the key you pilot with. Changing it later does not reach batches that have already run, since results carry the configuration they were produced under.
Separate keys per environment
Sandbox keys return scripted verdicts, open no SMTP connections and spend no credits, so CI can run against the API on every commit without a budget conversation.
Wire erasure into the existing DSAR process
The endpoint takes your own case reference and echoes it back, so the response drops into the privacy log you already keep rather than creating a second one.
What verification will not do here
What it costs
Volume pricing runs to $949 for a million addresses, at $0.00095 each. Credits never expire, which removes the year-end problem of a budgeted balance being written off unused.
Every tier works the same way. Unknown results and duplicates are never billed, and credits do not expire.
Questions we get asked
Do you sign a data processing agreement?
Yes. You are the controller for addresses you upload and we process them on your instruction, which is what the agreement sets out.
Can we run with zero retention?
Yes. Results live only as long as the batch and the uploaded file goes with them. Nothing is re-downloadable afterwards, which is the trade.
What is your uptime commitment?
None published yet. We will publish a rolling ninety-day figure measured by an external prober once there is enough production traffic to measure, and an SLA can only sit on a real measurement.
How do you compare on accuracy?
We publish no figure, and neither should anyone else without its denominator. Hunter's own forty-thousand-verification benchmark put the whole category at 63% to 70%, while vendors publish 99%. Both are computed honestly and they measure different things.
Try it on your own list first
100 free checks a month, no card. Run a sample and read the unresolved count before you decide anything.