White-label engine

Add "did it work?" to your product.

Marketing analytics, retail dashboards, BI tools, agencies' reporting stacks: your users keep asking whether a change caused a result. Counterfact answers with one API call: the effect, an honest range, robustness checks, and a plain-language explanation you can show under your own brand.

One call, no model fitting

Send the series and the change date. The pretrained model returns in seconds, with no per-customer tuning, no MCMC and no data-science team needed.

Honest by design

On 1,200 real test windows: 90% ranges held 90% of the time, with about 1 false alarm in 10, against about 4 in 10 for a typical CausalImpact-style setup.

Verified receipts

Every API result can carry a receipt in our Bitcoin-anchored public record: proof for your customers that results weren't edited after the fact. Only a data fingerprint is recorded.

White-label output

Plain JSON: verdict, effect, range, counterfactual path, checks and explanation text. Render it in your UI, your PDF, your colours.

The call

POST https://counterfact.p.rapidapi.com/v2/analyze
{
  "series": {"orders": [...], "region_b": [...], "region_c": [...]},
  "index": ["2026-01-05", "2026-01-12", ...],
  "metric": "orders",
  "change_at": "2026-09-07"
}
→ {"summary": {"verdict": "worked", "effect_per_period": 41.2, "likely_range_per_period": [18.5, 63.9], ...},
   "counterfactual": {...}, "explanation": "...", "receipt": {"id": "r…", "verify_url": "..."}}

Try it on RapidAPI Python client

Volume and white-label plans

Self-serve plans on RapidAPI go up to 100,000 calls a month. For embedding in your product (higher volume, an SLA, a dedicated endpoint, your own model calibration, or on-premise deployment), tell us about your use.