Meet the sprint
The same US advisory firm behind the decision-maker lead engine on this page ran a second, wider sprint in parallel: a Sales Navigator harvest aimed at building LinkedIn advertising audiences. Where the lead engine cut hard to decision-makers only, this sprint went wide, then let email quality do the sorting.
The challenge
The sprint target was 100,000 leads, and the honest result is that the sources would not give them: the main pool exhausted at about 31,500 and the secondary source ran on a free quota of 500 lookups per account per month across 4 accounts. The workbook closed at 33,049 unique leads. The question became what that workbook was worth as advertising audiences, which depends entirely on how LinkedIn matches the emails, and match counts quoted at upload time are estimates, not results.
The solution
Split by quality before uploading anything
The workbook was split by verification state: 3,665 fully valid emails, 11,193 risky or unknown, and 18,191 bulk leads. The split matters because the audiences inherit it: an advertiser who knows which audience is the clean one can spend against it first.
Upload everything, in two clean audiences
The first audience took the valid, risky and unknown emails: 14,861 addresses. The second took all 18,191 bulk emails, with zero overlap between the two files, so every one of the 33,049 workbook leads was uploaded exactly once. Nothing was quietly held back to make the match rate look better.
Read the counts at Ready, not at upload
Both audiences were left to finish building, then read back at their final state on November 3. The first audience matched 14,262 of its 14,861 uploads. The second matched 17,068 of its 18,191. Both sat at Ready, both above 90% matched, and neither was attached to any ad set, so the firm could decide its campaign structure with the real numbers already in hand.
The results
- 33,049 unique leads in the workbook, all uploaded
- 31,330 prospects matched across the two audiences
- 14,262 matched from the first audience, 17,068 from the bulk audience
- Above 90% match rate on both audiences at their final Ready read
- 0 leads held back and 0 overlap between the audiences
The shortfall is part of the record: the sprint aimed at 100,000 leads and the sources capped out at 33,049. We reported that with the ceilings named, the exhausted pool and the free tier quotas, rather than padding the workbook with unverified rows to hit a rounder number. A smaller workbook that is real beats a bigger one that is not, and 31,330 matched prospects is a working advertising asset either way.
The return: the firm went from a raw harvest to two targetable audiences covering more than thirty one thousand matched prospects, with the quality split preserved, ready for campaign spend the moment it chooses to spend.
FAQ
Why split the workbook into two audiences?
Email quality. The valid and checked leads behave differently from bulk harvest rows, and keeping them in separate audiences lets spend start on the clean set while the bulk set is tested separately.
Why read match counts at Ready instead of at upload?
LinkedIn shows an estimate while an audience builds, and the estimate moved during this build. The only number worth reporting is the final matched count once the audience reaches Ready.
Did the sprint hit its 100,000 lead target?
No. It closed at 33,049 because the main source pool exhausted and the secondary source ran on small free quotas. That is stated in the delivery, with the ceilings, rather than padded over.
Were the audiences attached to campaigns?
No. Both were left unattached at Ready, deliberately, so the campaign structure could be decided with final match numbers already known.
