Every credit-based enrichment tool has a number they don't put on the pricing page. It's called the match rate: the share of lookups that actually return a verified email. The rest are misses. And in most pricing models, you pay for the misses too.
We call that the bounce tax. Here's how to calculate what it's costing you, and why it gets worse, not better, when AI agents run your prospecting.
The math nobody shows you
Say a tool sells credits at $0.10 per lookup and matches 65% of the time, a respectable rate in this industry. To get 1,000 verified contacts you need ~1,540 lookups:
| Lookups | Cost | Contacts delivered | |
|---|---|---|---|
| Hits | 1,000 | $100 | 1,000 |
| Misses | 540 | $54 | 0 |
| Effective price | 1,540 | $154 | $0.154 per contact |
Your real price per contact is 54% higher than the sticker price, and it scales with every point of match rate the vendor loses. At a 50% match rate, you're paying double.
The second-order cost is worse. Tools that "match" by pattern-guessing (first.last@domain) keep their apparent match rate high by handing you emails that were never verified, and those become bounces. Sustained bounce rates above ~2% degrade your sender reputation with mailbox providers, which quietly lowers deliverability on every email your team sends afterward.
Why agents make the bounce tax explode
A human SDR paces themselves: a few dozen lookups a day, roughly aimed. An AI agent doesn't. Point an agent at a market and it will happily attempt hundreds of enrichments in a session, exploring long-tail companies where match rates are lowest.
That's exactly what makes agents valuable: they exhaust a market instead of skimming it. But under pay-per-attempt pricing, the agent's thoroughness is billed to you at the worst possible rate. The economics of attempt-priced credits and agent-driven prospecting are fundamentally incompatible.
What to demand instead
Start by pinning down what your vendor means by the word. Verified email covers the difference between a checked mailbox and a pattern guess, which is the distinction the whole bill turns on.
Three pricing properties matter when an agent is doing the asking:
- Hit-only billing. A miss costs zero. No verified email means no charge; the vendor eats the attempt cost, which forces them to actually care about their match rate.
- Free retries and cached repeats. Asking again about the same contact shouldn't bill twice. Agents re-ask things constantly; that's how they work.
- Verification before delivery, not after. "Verified" should mean live deliverability checks passed (mailbox exists, domain accepts mail), not a pattern guess with a confidence score.
A vendor's pricing model tells you what they're optimizing. Per-attempt pricing profits from misses. Hit-only pricing only makes money when you do.
How this works under the hood
Fair warning: hit-only pricing isn't something a vendor can just decide to offer. If they pay one expensive upstream provider full freight per attempt, absorbing misses would bankrupt them. It only works with a waterfall: cheap or free discovery sources tried first, expensive deep-search providers invoked only when needed, and aggressive caching so a contact enriched once is never paid for twice.
That architecture is the real product. The pricing model is just the proof it exists.
Run the numbers on your own stack
Pull last quarter's enrichment invoice and divide total spend by delivered, verified contacts, not lookups. That's your true price per contact. If it's meaningfully above the sticker rate, you're paying the bounce tax, and the gap is the discount any hit-only vendor effectively starts with.
If you want to see what hit-only pricing feels like in practice, connect Signl to your agent. Search is free and unlimited, and a contact that bounces is a contact you didn't pay for.