The same answer costs $1 per thousand from one place and $20 per thousand from another. That gap is worth understanding before deciding it is a discount.
Two different cost structures
| WHOIS-based check | Zone-file check | |
|---|---|---|
| Work per row | A network request to a rate-limited service | A lookup in data already held |
| Needs | Proxies, backoff, a retry path | Nothing per row |
| Marginal cost | Real and roughly linear | Close to zero |
| Store price | $0.003 – $0.02 per domain | $0.001 per domain |
Pricing at the marginal cost of serving
The temptation with a cheap-to-serve dataset is to price it just under the competition and keep the difference. Pricing it at what it costs to serve is the better long-run decision, because the gap then has an explanation a customer can check — and an order-of-magnitude difference with no explanation reads as a quality problem rather than a bargain.
What the gap looks like at volume
Screening 5,000 candidate names costs $5.00 at $0.001 each, against roughly $100 at $0.02. At 300 names it is 30 cents against $6. The absolute numbers are small either way at the low end, which is exactly why the structure matters more than the headline: it decides what becomes possible at 50,000 names — a screening pass over a whole shortlist rather than a careful selection of what to check.
Publishing the comparison you lose
Why an empty result costs almost nothing
An empty result consumed almost no resource, so it is charged almost nothing — the start fee alone. That follows from pricing the work rather than the request, and it has a useful consequence: on a monitoring Actor, the outcome a defender wants most often is the one that costs least. Nobody is discouraged from checking. The same rule sends the cost of a blocked page to the operator rather than the customer, which is worked through in pay-per-event pricing.
