How we measure accuracy, and how you can check it.

The claim

We report near-100% accuracy on LeadQC output. This page sets out what that means, what is being counted, and how you can test it on your own data rather than taking ours.

What is being measured

Accuracy is the proportion of records where the verdict LeadQC returned matched the true state of the contact. It is not a bounce rate, not a deliverability score and not a match rate — three numbers frequently reported as if they were the same thing. A record counts as correct only when both the verdict and the reason hold up.

What counts as a miss

• A false removal — we removed a record that was in fact a live, correctly-employed contact.
• A false keep — we passed a record that should have been removed.
• A wrong reason — the verdict was right but the stated reason was not. This still counts against us, because the reason is what you act on.

Records we decline to judge are not counted as correct. They are returned flagged for a person to decide, and reported separately — because a validation system can post a very high accuracy figure simply by refusing to rule on anything difficult.

How the comparison is made

Each record in your file is compared against the prospect's live professional profile at the moment of the run, not against a stored copy. Employment state, employer, title, seniority and tenure come from that live read. Email deliverability is established by live checks on the domain and mailbox. Addresses in APAC and EMEA are verified against Google at run time. Phone numbers are checked live rather than pattern-matched.

Because the comparison is against live data, the same file run twice weeks apart can legitimately produce different verdicts. That is the point — a list verified last month is not a list verified today.

Why we would rather measure it on your file

An accuracy figure quoted from a vendor's own sample tells you very little about your segment. Your list has its own geography, seniority mix and source quality, and those are what determine how a validation system performs on it.

So the number that matters is the one produced on your data. The free 5,000-record assessment returns your accuracy, your removal rate, your flagged rate and the reason for every decision — inside 24 hours, before you spend anything.

The record-level working

The detail behind our own measurement — sample composition, how ground truth was established, and the per-category breakdown — is available under NDA to customers and prospects in evaluation. Ask and we will send it.