GET /v1/conversions
One page of conversion events recorded by traceten.track() or a server SDK, newest first, one row
per conversion. Each row carries two separate answers about where it came from:
- The visit’s source (
source_kind,source_name,session_ai_source): where the visit the conversion happened in came from. This is what the dashboard’s Conversions page shows. - The model’s credit (
ai_source,confidence): the source credited under the attribution model you ask for.
site_id here is the site’s Postgres id, the UUID in your dashboard URL. That is different from
the key-authenticated read endpoints, which take the ttid_ snippet key. The difference follows the
credential: this route only ever sees a dashboard session.
Query parameters
Unknown parameters are rejected with
422 rather than ignored, so a filter this endpoint cannot
apply never returns a 200 answering a different question.
(page - 1) * page_size may not exceed 100000. ClickHouse reads the rows an OFFSET skips, so a
very deep page is the most expensive request this endpoint can serve; past that ceiling it is a
422.
Response
Under
linear and time_decay a conversion is split across the visits that led to it. It still
appears once: ai_source is the source with the largest share and confidence is that share.
Top level:
display_value uses the European Central Bank reference rate for the day of the conversion, or the
most recent one published before it. total_count comes from daily totals, so a conversion removed
with the data deletion API can still be counted after its row is gone.
Errors
401, 403 and 404 all carry { "error": …, "statusCode": … }.
Next
- Goals API for the goal registry and completions over time.
- Sessions API for the sessions these conversions belong to.

