What does "pay only for successful scrapes" mean?
"Pay only for successful scrapes" means the web scraping service charges you when it returns valid content and eats the cost of failures (timeouts, blocks, 429s, 5xx errors, empty pages). It's the pricing model developers actually want, because it makes the vendor bear the reliability risk: their proxy pool, retry logic, and anti-block strategy have to work, or they don't get paid. The alternative (per-request billing that charges regardless of outcome) puts you on the hook for the target site's mood on any given day, which quietly inflates costs on hard sites and makes budgeting for large scrapes unpredictable.
| Billing model | What you pay for | Who carries the reliability risk |
|---|---|---|
| Success-based (per success) | Only responses with valid content | Vendor |
| Per request | Every request, including failures | Customer |
| Tiered by page complexity | Base + surcharge for JS, proxies, CAPTCHAs | Split, complex to predict |
| Flat monthly, capped requests | Fixed fee up to a quota | Vendor within quota |
Success-based billing is the model to prefer when you're scraping across many domains, when target sites vary in difficulty, or when the workload is bursty (a launch, a competitive event, an agent that goes wide). Per-request billing can still win on a single well-behaved target you own, where failure rates are low enough that the accounting overhead of tracking successes stops mattering.