You only pay for the data. Everything the agent does to find and shape that data is free.
- Chatting and planning with the agent
- Building tables, columns and labels
- Qualifying from data already in your table
- Testing a source before it runs
- A database lookup that finds nothing
- Sourcing rows: companies, people, posts, creators, jobs, local businesses
- Enriching rows: emails, phones, funding
- Web searches and page scraping
Credits are spent only on the real data Retriever pulls from providers and the open web. The intelligence layer that decides what to fetch, and turns it into a clean list, never costs you anything. Every action shows its price before you run it, and enrichment is billed only on the rows actually enriched, so your estimate is always a safe upper bound.
The real cost is a qualified lead, not a row
A single sourced row is cheap. What matters is the price of a lead that passes your criteria, and that depends on how many rows Retriever has to source to find one. One relationship ties it together, shown here on an example search:
The price per row barely moves. The qualification rate is the lever: the higher the share of rows that pass, the fewer Retriever sources per qualified lead, and the lower your cost. Narrow, rare targets lower that rate, so the cost per lead climbs. Both numbers are shown on every source before you commit.
Qualification rate
The share of sourced rows that pass your criteria. Retriever works to keep it high by picking the source that best matches your target, which is what pulls your cost per qualified lead down.
Price per qualified lead
The all-in cost of one qualified lead. It already accounts for every row Retriever had to source and reject to find one that fits.
What each action costs
Credits per row. Where a range is shown, Retriever tries the cheapest source first and only falls back to a pricier one if the data is not there.