How do I vet an AI-powered lead generation vendor to make sure they're delivering qualified prospects and not just a scraped list?
The short answer
Ask for verification rates (email bounce should be under 2%), the exact firmographic and intent signals used to score leads, and a sample of 20 "qualified" leads before signing. If a vendor can't explain how they define ICP fit beyond industry and headcount, or can't show verification credits or a scoring rubric, you're buying a scraped export with a markup.
Most "AI-powered lead generation" pitches are Apollo or Clay exports with a chatbot wrapper. That's not automatically bad — those tools are legitimate data sources — but the AI framing often hides the fact that no actual qualification happened. Vetting a vendor means separating three things they'll blur together: data sourcing, verification, and scoring. Ask about each separately.
Start with data provenance. Where does the raw list come from — a proprietary crawl, a licensed database like ZoomInfo or Apollo, or public scraping? There's nothing wrong with using a licensed database, but the vendor should say so plainly. If they claim a "proprietary AI engine" with no explanation of underlying data sources, that's a red flag. Ask specifically: "What percentage of your list is sourced from third-party databases versus your own crawling or enrichment?"
Then check verification, not just enrichment. Enrichment adds fields (job title, company size, email). Verification confirms those fields are still accurate and the email is deliverable. Bounce rate is the cleanest proxy. ZeroBounce, an email verification provider, calls an overall bounce rate below 2% healthy, under 1% ideal, and anything over 2% high enough to investigate — so a vendor handing you a verified list should be comfortably under 2%. Ask what verification method they use and how recently the list was checked. Stale enrichment from six months ago on a fast-growing company is often wrong about title, company, or even employment status.
Push on the qualification logic. "AI-qualified" should mean something more specific than "matches your industry filter." A real scoring approach weighs firmographic fit, buying signals (funding, hiring, leadership change, tech stack), and reachability of the actual decision-maker — not just a generic contact at the company. Ask the vendor to walk through their scoring dimensions and roughly how they're weighted. If they can't produce anything beyond "our AI does it," they're describing a keyword filter, not qualification.
What to ask for before signing
Request a sample batch — 15 to 25 leads — scored and delivered exactly as they would be in production, before committing. Check three things on that sample: do the job titles and companies check out on LinkedIn, does the contact still work there, and does the "why this lead" rationale make sense given your ICP. If the vendor resists providing a sample or wants payment first, that alone is disqualifying.
Also ask what happens to bad leads. A vendor confident in their process will offer some form of replacement or credit for leads that bounce or are clearly miscategorized — verification credits, a refresh cycle, or a guarantee threshold. No guarantee at all usually means no accountability for accuracy.
| Vetting question | What a strong answer sounds like | Red flag answer |
|---|---|---|
| Where's your data from? | Names specific sources (Apollo, ZoomInfo, proprietary crawl) and blend | "Our AI sources it all" with no specifics |
| What's your bounce rate? | Under 2%, with a verification method named | "We don't track that" or no number |
| How do you define qualified? | Named scoring dimensions (ICP fit, intent signal, reachability) | "Our algorithm decides" |
| Can I see a sample batch? | Yes, before or with minimal commitment | Sample only after contract signed |
| What happens to bad leads? | Credits, replacement, or refresh cycle | No answer or "that's the cost of doing business" |
Price shouldn't be the first filter, but it's a useful gut check
Per-lead or per-contact pricing that seems too cheap to include any verification step probably doesn't have one. Verification alone — running a list through a checker — costs real money; SalesARC, for reference, sells email verification credits at $49 for a pack of 2,000 checks, which gives you a sense of what verification actually costs at scale. If a vendor's per-lead price is lower than what verification alone would cost them, either they're not verifying, or they're eating margin somewhere you should ask about.
Where SalesARC fits
SalesARC's own prospecting sits in ArcClaw™ Prospecting Agents, which starts at $499 monthly (Basic) and scales to $3,499 monthly (Enterprise), and SalesARC Prospect, which runs $199 to $1,499 monthly depending on tier (Prospect Basic is included with every ArcClaw™ tier). Every address is verified before Prospect sends to it; extra verification credit packs ($49 for 2,000) only top up volume. ArcClaw™ scores each lead for fit and readiness against your playbook's qualification criteria and shows the rubric behind every score, rather than applying a flat industry filter — the same disclosure this article recommends asking any vendor for. This is most relevant for founder-led B2B companies without a dedicated sales leader, which is SalesARC's core fit; larger enterprise buyers with established RevOps teams may find more mature data infrastructure elsewhere. See SalesARC Prospect for details.