Export an experience report (CSV/PDF)

Generates an export of an experience's performance report in either CSV (aggregated or daily) or PDF format.
A synchronous export returns a downloadable link in data.download_url.
A routed breakdown returns an export id in data.jobId instead.
Useful for offline analysis or sharing with stakeholders.
The Knowledge Base article "Experiment Report" (Export & Download section) describes this.

Optionally, attribution_breakdown splits the export by the discovered values of a single
attribution dimension. The two non-tabular report types repeat one section per value.
The two tabular types, csv_aggregated_tabular and csv_daily_tabular, are flat CSVs with no
section header. They carry the attribution value in a column next to Goal.
Three dimensions are shipped and populated: campaign
(raw utm_campaign), content (raw utm_content, the ad-creative identifier), and
source (Convert's derived five-value traffic-channel class — campaign | search | referral | direct | ai_tool — not the customer's raw utm_source string). Three are
permanently unavailable and refused with 422 if requested: medium and term (captured
client-side but never stored anywhere in the reporting path) and utm_source (captured
client-side but dropped before transmission, so Convert never stores it). See
AttributionDimension for the full six-entry set and AttributionDimensionAvailability
for the shape a dimension picker can use to render all six with their availability.

A breakdown includes every discovered value, up to 5000 per experiment, selected top-by-volume.
The report's live segmentation uses the same cap and ordering.
When the cap cuts the list, the export states in its own content that it kept the top 5000 values and that further values exist.
The reporting store keeps campaign and ad-creative values without a length limit.
The breakdown therefore lists an over-long value like any other value.

A breakdown request takes one of two paths.
The synchronous path renders the export during the request and returns download_url and expire_at.
It never cuts a breakdown by value count or by time.
The request routes to the async path in two cases.
In the first case, the number of discovered values exceeds the current routing threshold.
The server configures that threshold per report type, and it is a routing rule, not a cap.
In the second case, a synchronous run reaches its time deadline, currently 120 seconds and configured server-side.
The async path queues the work, and the report worker renders the breakdown for the same report types.
The worker emails a download link to the requester.
The 200 response then carries the export id in data.jobId, with no download_url and no expire_at.
The async path includes every value up to the 5000 cap.

The reporting store keeps utm_content for every account.
Requesting any breakdown value at all requires the advanced-segmentation plan entitlement.
A request without it is refused with 403, and sudo users bypass this check.
A routed breakdown takes this endpoint's own plan gate.
It does not require the experience_raw_export plan function that export_raw_data requires.
A routed breakdown is throttled the way export_raw_data is, under its own budget.
That budget is separate from the export_raw_data budget, so a routed request can receive 429.

Recent Requests
Log in to see full request history
TimeStatusUser Agent
Retrieving recent requests…
LoadingLoading…
Path Params
integer
required

ID of the account that owns the retrieved/saved data

integer
required
≥ 1

ID of the project to which save/retrieved data is connected

integer
required

ID of the experience

Body Params

Parameters for exporting an experience report. Requires:

  • report_type: 'csv_aggregated' (summary data per variation) or 'csv_daily' (day-by-day data).
  • data_type: (For csv_daily only) 'cumulative' or 'non_cumulative'.
  • goals: (Optional) List of goal IDs to include.
    Also supports start_time, end_time, utc_offset, segments, and stats_engine_processing settings to match the online report.

Base Parameters needed to drill down the reports.

string
enum
Defaults to csv_aggregated

Type of the exported report:

  • csv_aggregated - export in CSV format the number of visitors and metrics values aggregated per variation for the whole reported period
  • csv_aggregated_tabular - export in CSV format same as above but with headings tabular for export
  • csv_daily - export in CSV format the number of visitors and metrics values per day, per variation for the whole reported period
  • csv_daily_tabular - export in CSV format same as above but with headings tabular for export
Allowed:
string | null
enum
Defaults to non_cumulative

Type of data the exported report (csv_daily only)

Allowed:
goals
array of integers | null

A list of goal Id's for which to return the reporting data.

If no ID is provided, the list of goals attached to the experience is being used.

Max 10 goals at one time available for csv_daily type.

goals
string | null
enum

Opt-in breakdown of this export by one attribution dimension. Omitting this field,
or passing null, produces the export byte-identical to today's output — this
protects a live Airtable import that consumes csv_aggregated_tabular's trailing
columns and must not see its shape change.

When set, the export is split by the values of the dimension discovered for this
experiment (see AttributionDimension). The two non-tabular report types repeat one
section per value, each holding the same variation table the unbroken export already
carries for that filter. The two tabular types, csv_aggregated_tabular and
csv_daily_tabular, are flat CSVs with no section header. They carry the attribution
value in a column next to Goal.

Requires the advanced-segmentation plan entitlement. A request without it is
refused with 403, and sudo users bypass this check. medium, term, and
utm_source are never available and are refused with 422. The reporting store
keeps content for every account. See AttributionDimensionAvailability.

The export includes every discovered value, up to 5000 per experiment, selected
top-by-volume. When that cap cuts the list, the export states it in its own content.
The synchronous path never cuts the breakdown by value count or by time.

A large breakdown routes to the async path. The server compares the discovered
value count with a routing threshold that it configures per report type. A
synchronous run that reaches its time deadline routes too. The report worker emails
the download link, and the 200 response carries data.jobId instead of
download_url. A routed breakdown takes the plan gate of export_report, not
experience_raw_export. It is throttled the way export_raw_data is, under its
own budget, and can receive 429.

Allowed:
integer
-43200 to 50400
Defaults to 0

The time offset from Coordinated Universal Time (UTC) in seconds.
For example, UTC-5 (EST) would be -18000. UTC+2 would be 7200.
Used for interpreting or displaying date/time information according to a specific timezone.

number | null
0 to 4294967295

Unix timestamp (seconds since epoch) for the beginning of the reporting period. If null, data from the experience start is typically used. Should be interpreted with utc_offset.

number | null
0 to 4294967295

Unix timestamp (seconds since epoch) for the end of thereporting period. If null, data up to the current time is typically used. Should be interpreted with utc_offset.

segments
object

Conversion custom dimensions are stored only on conversion (hitGoal) and transaction (tr) events.
Each conversion_custom_dimension_N filter is applied when counting those events and is not applied
to visitor or bucketing (viewExp) events.
Website transactions recorded before the tracking script copies these dimensions onto the
transaction carry none, so filtered revenue for that period counts those transactions as zero.
Within one dimension, an event matches when it carries any listed value. Dimensions combine with AND.
Revenue and product counts are those of the whole matching transaction, not of the dimension value alone.
Requires the advanced segmentation plan function, the same as campaigns and custom segments.
Each value is matched exactly and is stored truncated to 32 characters.

stats_engine_processing

Overall settings for the statistical engine processing results for this experience.

Responses

Language
Credentials
URL
LoadingLoading…
Response
Click Try It! to start a request and see the response here! Or choose an example:
application/json