Bad Lead Teardown · Case 01

We reviewed 10 LLM-generated leads. Zero were ready for outreach.

The list appeared complete enough for outreach. A row-by-row review found that none of it was ready for us to look for contact details.

Can an LLM build a qualified lead list? It can suggest names and companies, but those details still need checking. In this audit, none of the 10 leads was ready for us to look for contact details.

The list looked complete, but 9 people could not be verified at the companies listed. The last company was not the kind of company requested. That is the risk with LLM-generated lead lists: convincing output can still be wrong.

We checked the person and company before looking for contact details.

We treated every row as a suggestion. A lead moved forward only when every earlier check passed.

  1. 01

    Person and company match

    Could we verify that the person worked at the company in the role listed?

  2. 02

    Right kind of company

    Was this the kind of company the client asked for, with no clear reason to exclude it?

  3. 03

    Right person

    Was this person likely to care about the offer, rather than simply holding a senior title?

  4. 04

    Public proof

    Did reliable public sources agree?

  5. 05

    Contact check

    Only after the first four checks passed would we look for contact details.

All 10 rows stopped before we looked for contact details.

A neat-looking row was not enough. We first checked whether the person, company, and role could be verified and matched the request.

Results for the submitted 10-row list
Review outcomeRowsWhat we found
Held9We could not verify that the person worked at the company listed.
Rejected1We could not verify the claimed leader, and the company was not the kind requested.
Moved to contact search0Contact checks were intentionally skipped for all 10 rows because earlier checks did not pass.

“Held” means we could not verify the match using public sources. It does not prove that the person or job does not exist.

The replacement list held up better.

We also checked a separately sourced list of 10 leads using the same steps. It is useful context, but it was not a controlled test.

Results for the separately sourced replacement list
CheckResultWhat we did
Person and company match9 clear matches; 1 close role matchWe kept the close match marked as close instead of calling it exact.
Work email check9 verified work emails; 1 not usedWe did not use one because the email domain did not match the company website.
Reason to contact now6 current; 2 with no current reason found; 1 stale; 1 not checkedWe did not say there was a current reason to reach out when the proof was missing or old.

Already have a list?

Upload a list and check each row before looking for contact details.

Bring a list for review