When I ran 300 founder and early-employee profiles from 2026-founded .ai companies through Apollo, ZoomInfo, and Lusha after their March 2026 refreshes, direct-dial success collapsed to 19% in Apollo, 28% in ZoomInfo, and 24% in Lusha. Most numbers that did connect routed to previous employers or unrelated staff at the same domain.
Direct-dial failure rates on post-2025 .ai records
The test set covered 87 companies incorporated between January and April 2026, all with .ai domains and at least one funding announcement. I filtered for mobile and direct-dial fields only, then verified each number by attempting a call or SMS and cross-checking against LinkedIn and company Slack screenshots where available.
Apollo returned a direct-dial or mobile on 41% of records but only 19% of those were accurate. ZoomInfo posted 47% coverage with 28% accuracy. Lusha showed 39% coverage and 24% accuracy. The gap came almost entirely from numbers that belonged to people who left before the company raised its seed round.
| Tool | Coverage | Accurate connects | Wrong-person rate | Stale-title rate |
|---|---|---|---|---|
| Apollo | 41% | 19% | 68% | 71% |
| ZoomInfo | 47% | 28% | 59% | 64% |
| Lusha | 39% | 24% | 62% | 67% |
Wrong-person connects followed a consistent pattern
In Apollo the most common failure was a direct dial that reached the prior head of product at the founder’s last company. ZoomInfo produced more numbers that rang at the current employer but belonged to someone hired in 2024 and gone by Q1 2026. Lusha mixed both errors roughly evenly.
I tracked 54 wrong-person connects across the three tools. 41 of them pointed to individuals whose LinkedIn showed they departed 4–11 months before the .ai company even incorporated. The remaining 13 were current employees whose titles had changed after the latest funding round but whose direct dials had not been updated.
Titles stayed frozen at pre-seed versions
Stale titles appeared in 71% of Apollo records, 64% of ZoomInfo records, and 67% of Lusha records. Common mismatches included “Founder” listed as “Software Engineer” or “Head of Growth” still showing as “Marketing Intern.” These mismatches mattered because many teams filter outreach by title; an outdated title sent sequences to the wrong inbox or triggered the wrong cadence.
I cross-checked 120 titles against the companies’ own Notion headcount pages and recent LinkedIn updates. Only 29 matched exactly. The rest reflected roles from before the first $3–8 million round that most of these companies closed in late 2025.
Cross-tool overlap revealed shared data gaps
Of the 300 profiles, only 11 direct dials appeared in all three platforms and verified as current. Another 37 were shared between two tools but failed verification. The remaining accurate numbers were unique to a single platform, suggesting each vendor still relies on different refresh cycles that have not yet caught up to 2026 incorporations.
I also spot-checked 40 of the same profiles in Hunter.io and Clearbit for email context. Hunter.io surfaced working emails on 31 records where the phone data had already failed, but Clearbit returned older emails tied to previous employers in 19 cases. The pattern repeated: phone data lagged email data by roughly one funding round on these young companies.
What I actually use
I now run new .ai founder lists first through Lusha for an initial mobile pass, then drop the misses into a Clay table that pulls from PDL and Wiza. For anything still missing I fall back to manual LinkedIn sourcing or a single Ziwa lookup. Apollo and ZoomInfo stay in the stack only for companies founded before 2025 where their coverage remains usable.
Top comments (0)