Reporting - Everflow API

Four tools cover every reporting question. Pick by the shape of the answer you need, not the subject.

You need	Use	Requires
Totals, trends, or a breakdown by dimension	run_performance_report	Reporting
Headline numbers for a period, optionally vs. the period before	run_network_summary	Reporting
Raw click or conversion records over a window	search_activity	Reporting
The valid dimensions, filters, and metrics	get_report_schema	Reporting
One click, conversion, or full attribution chain by ID	get_entity	Reporting

get_report_schema

Returns every valid dimension key, filter key, and metric name for run_performance_report, plus an authoritative definition for each metric. Call it first when you’re unsure which keys to use — the exact strings it returns are the accepted values. Requires: Reporting (Read Only) Ask for it: “What metrics can I report on, and how is EPC calculated?” Parameters None. Example

get_report_schema()
→ the full list of dimensions, filters and metrics, with a definition for every metric

Returns

Field	Type	Description
`columns`	array	Valid dimension keys for the `dimensions` parameter
`filters`	array	Valid filter keys for the `filters` parameter
`metrics`	array	Valid metric names for `sort_by` — also returned in every report row
`metric_definitions`	array	One entry per metric: `{ name, description, unit, formula, is_custom }`

Gotchas

For built-in metrics, unit is count, currency, percent or ratio, and the human-readable formula lives in description (the formula field may be empty).
For custom metrics (is_custom: true — your network’s own definitions), formula is populated but unit may be empty. These are definitions only; see Custom metrics.
This is the source of truth for “what metrics do I have?” and “how is X calculated?” — don’t infer a formula from the metric name.

run_performance_report

Queries aggregated performance data grouped by one or more dimensions. The primary tool for campaign analysis, partner comparison, and revenue reporting. Requires: Reporting (Read Only) Ask for it: “Show me revenue by offer for the last 7 days, top 10.” Parameters

Parameter	Type	Required	Default	Description
`from`	string	Yes	—	Start date (`YYYY-MM-DD`)
`to`	string	Yes	—	End date (`YYYY-MM-DD`)
`dimensions`	string	Yes	—	Comma-separated dimension keys (see the accordion below). Max 256 chars
`filters`	string	No	—	Comma-separated inclusive `type:value` pairs, e.g. `offer:123,affiliate:456`. Pipe-separate to match several IDs on one key: `offer:1\|2\|5`. Inclusive only — there is no exclusion syntax. Max 4,096 chars
`metric_filters`	string	No	—	Numeric conditions on aggregated metric values — comma-separated, ANDed, max 10. E.g. `clicks>100,cvr<2`. Operators `>`, `>=`, `<`, `<=`. Applied after aggregation, before sorting. Max 512 chars
`sort_by`	string	No	no sort	Metric name to sort by — must match a metric name exactly. An unrecognized name is silently ignored, not rejected. Max 128 chars
`sort_direction`	string	No	`desc`	`asc` or `desc`. Max 8 chars
`timezone`	string	No	Network timezone	IANA timezone name. Max 64 chars
`currency`	string	No	Network base currency	Currency code. Max 8 chars
`page_size`	number	No	50	Rows per page (1–100)
`cursor`	string	No	—	Pagination cursor from a prior response. Max 1,024 chars
`comparison`	string	No	—	`previous_period` — compare `[from, to]` against the equal-length window immediately before it
`compare_from`	string	No	—	Explicit comparison-baseline start (`YYYY-MM-DD`). Use instead of `comparison` for an arbitrary prior window
`compare_to`	string	No	—	Explicit comparison-baseline end. Must be paired with `compare_from`

Example

run_performance_report(
  from="2026-08-01", to="2026-08-31",
  dimensions="offer,affiliate",
  filters="advertiser:99",
  metric_filters="clicks>500",
  sort_by="revenue", page_size=25
)
→ 25 offer × partner rows for advertiser 99 with 500+ clicks, ranked by revenue,
  plus network-wide totals and the resolved query that produced them

Returns

Field	Type	Description
`rows`	array	One row per dimension combination, each carrying every built-in metric
`totals`	object	Every built-in metric aggregated over the full matched set — all rows, before pagination and the display cap
`applied_query`	object	The exact resolved query: `from`, `to`, resolved `timezone` and `currency`, `dimensions`, and any `filters`, `metric_filters`, `sort_by`/`sort_direction`. Empty sections omitted
`metric_glossary`	object	Inline one-line definitions for the easily-misread metrics, so methodology travels with the numbers
`custom_metrics`	object	Per row — your network’s custom metrics, when any are defined. See Custom metrics
`rows_returned`	number	Rows in this page
`total_rows`	number	The true number of matching rows, before any cap
`next_cursor`	string	Pass as `cursor` for the next page. Absent when there are no further pages
`has_more`	boolean	More pages exist — ordinary pagination, not missing data
`result_capped`	boolean	The query matched more than the hard row limit and rows were dropped
`row_limit`	number	Present only when `result_capped` — the ceiling that dropped rows

has_more and result_capped are independent. has_more: true alone just means “page for the rest.” result_capped: true means the result is genuinely incomplete and paging every page will not recover the missing rows — narrow the date range, add filters, or use fewer dimensions.

Gotchas

Use totals, don’t sum rows. Ratio metrics can’t be summed or averaged into an overall figure, and summing one page misses the rest. totals re-derives ratios from the underlying sums (overall cvr = Σtotal_conversions ÷ Σclicks), so it stays correct even when only one page of rows is returned.
filters and metric_filters are different axes. filters narrows by dimension/entity (offer, affiliate, country, sub1…); metric_filters narrows by metric (clicks, revenue, cvr…). A metric name is not a valid filters key, and vice versa. Use conversions, not cv.
Percent metrics are percentages. cvr, ctr and margin return 2.84 for 2.84%, not 0.0284. In metric_filters, cvr<2 means “under 2%”. For “no invalid traffic” use invalid_clicks<=0.
This tool returns total_rows. List tools like list_offers return total_matching. Different fields, different meanings.
Timezone resolution. Everflow supports a fixed set of timezones. A valid IANA zone outside that set resolves to the supported zone with identical day boundaries and DST rules (America/Toronto → America/New_York) — the numbers are unaffected. When that happens, applied_query carries timezone_requested and a timezone_note, so the resolved zone isn’t mistaken for your request being ignored.
filters also accepts a JSON-object form — {"offer":"1|2"} or {"offer":["1","2"]} (array values work too). Don’t mix it with the key:value string form in one call. list_entities, by contrast, accepts only a JSON object.

Period-over-period comparison

To answer “which offers or partners moved the most”, have the report compare two periods rather than running it twice and subtracting. Set comparison=previous_period, or pass an explicit compare_from/compare_to baseline (e.g. the same month last year). Each row then carries, for every metric, an object of { current, prior, delta, pct_change }, all computed server-side. pct_change is null when the prior value was 0 — the change is undefined, not infinite. Rank the movers with a delta sort key: sort_by accepts <metric>_delta and <metric>_pct_change (e.g. epc_delta, revenue_pct_change) alongside the raw metric names.

run_performance_report(
  from="2026-07-01", to="2026-07-14",
  dimensions="offer", filters="affiliate:3296",
  comparison="previous_period", sort_by="epc_delta"
)
→ offers this partner runs, ranked by change in EPC vs 2026-06-17–2026-06-30

Comparison mode has its own rules:

Rule	Detail
Current-anchored	Ranks rows active in the current period against their prior values. An entity with prior-period activity and none in the current period is not returned — so comparison mode cannot answer “what did I lose”. Run an explicit report over the prior window for that
`sort_by` required	Comparison mode will not run without it
No pagination	`cursor` is not supported — there is no paging across a join
No time dimensions	`date`, `hour` and `month` never align across two windows
Bounded at 200 rows per period	If either period matches more than 200 rows, the call errors and names the count rather than truncating. Add `metric_filters` (e.g. `revenue>1000`), add `filters`, or drop a dimension

The prior period is fetched scoped to the current period’s entities, so for single-dimension comparisons narrowing the current period is enough. With multiple dimensions the prior scope is a cross-product of the current IDs, so a very active prior period can still exceed the cap.

Custom metrics

If your network defines custom metrics, every row also carries a custom_metrics object keyed by metric name. It is not selectable — there is no parameter to request or suppress it. Every custom metric defined on the network is computed and returned on every row; if the network defines none, the key is absent entirely.

Behavior	Detail
Values are display strings, not numbers	Computed and formatted by the reporting engine
Formatting follows each metric’s `format` and `decimal_places`	A `money` metric with 2 dp returns `"2895.00"`; a `percentage` with 1 dp returns `"4.3"`; an `integer` returns `"1930"`
No currency symbol or `%` sign is included	Parse the string and apply the unit yourself, reading `format` from the metric’s `get_report_schema` entry
Not usable in `sort_by` or `metric_filters`	The values are formatted strings, so they can’t be sorted or compared. A custom metric name in `sort_by` is silently ignored (you get the default order, not an error) — check `applied_query` to see what actually ran
Rows only	`custom_metrics` never appears in `totals`, so a custom metric has no network-wide aggregate in the response
Absent from the schema’s `metrics` array	They appear only in `metric_definitions` with `is_custom: true`

Custom metrics are distinct from the custom_payout_revenue entity.

Reference

Available dimensions

Pass one or more of these keys as a comma-separated dimensions string. The schema also returns two aliases not listed below: smart_link (alias of campaign) and device (a general device dimension alongside device_type).

Dimensions are served from one of two stores, which affects latency, not correctness.Pre-aggregated summary (fast) — the time dimensions (date, hour, hourly, week, month, day_of_week), the entity dimensions (offer, originating_offer, offer_status, advertiser, affiliate, affiliate_status, creative, campaign/smart_link), the manager dimensions (affiliate_manager, account_manager, sales_manager, account_executive), plus currency, project_id, advertiser_campaign_name, app_identifier, bundle_id and meta_platform.Raw event detail, backed by BigQuery (slower) — every other dimension: geo (country, region, city, dma, postal_code), device (device, device_type, platform, browser, os_version, device_make, device_model), connection (isp, carrier, language, connection_type, is_proxy), sub1–sub10, adv1–adv10, source_id, event_name, transaction_id, attribution_method, order_id, coupon_code, referer, tracking_domain, category, offer_group, and the error-code breakdowns.Grouping by any raw-detail dimension moves the whole query to BigQuery. Totals are computed over whichever store answers the query, so if a summary-backed grouping and a detail-backed grouping of the same window ever disagree, the two stores have drifted — report it rather than averaging them.

Time

Dimension	Description
`date`	Day (YYYY-MM-DD)
`hour`	Per-hour timestamp bucket (e.g. `2026-07-14 15`) — one row per distinct hour in the range
`hourly`	Hour of day (0–23, as `HH:00`) — aggregated across the whole range; use for peak-hour / time-of-day analysis
`week`	ISO week
`month`	Month (YYYY-MM)
`year`	Year
`day_of_week`	Day of week as a name (`Monday`…`Sunday`), aggregated across the range

Entities

Dimension	Description
`offer`	Offer name
`offer_id`	Offer ID
`affiliate`	Affiliate name
`affiliate_id`	Affiliate ID
`advertiser`	Advertiser name
`advertiser_id`	Advertiser ID
`campaign`	Smart link / campaign name
`campaign_id`	Campaign ID
`smart_link`	Alias of `campaign`
`creative`	Creative name
`creative_id`	Creative ID
`offer_group`	Offer group name
`offer_group_id`	Offer group ID
`offer_url`	Offer URL
`offer_status`	Offer status
`affiliate_status`	Affiliate status
`originating_offer`	Originating offer in a redirect chain
`category`	Offer category
`network`	Network identifier
`source_id`	Traffic source ID
`event_name`	Conversion event name
`advertiser_event_name`	Advertiser-side event name
`advertiser_campaign_name`	Advertiser campaign name

Geo

Dimension	Description
`country`	Country name
`country_code`	ISO 2-letter country code
`region`	Region / state
`city`	City
`dma`	Designated market area (US only)
`postal_code`	Postal code

Device & technology

Dimension	Description
`browser`	Browser name
`platform`	OS platform
`device_type`	Device type (desktop, mobile, tablet)
`device`	Alias of `device_type`
`device_make`	Device manufacturer
`device_model`	Device model
`os_version`	OS version string
`language`	Browser language
`connection_type`	Connection type (wifi, cellular, etc.)
`carrier`	Mobile carrier
`isp`	Internet service provider
`is_proxy`	Whether traffic came through a proxy
`meta_platform`	Meta / Facebook platform

Tracking & attribution

Dimension	Description
`tracking_domain`	Custom tracking domain
`transaction_id`	Raw transaction ID
`attribution_method`	Attribution method
`order_id`	Order ID passed on conversion
`coupon_code`	Coupon code
`referer`	Referring URL
`click_error_code`	Click error code
`conversion_error_code`	Conversion error code

People & managers

Dimension	Description
`affiliate_manager`	Affiliate manager name
`account_manager`	Account manager name
`sales_manager`	Sales manager name
`account_executive`	Account executive name
`customer_support_manager`	Customer support manager name
`admin_account_manager`	Admin account manager name

Payout & revenue

Dimension	Description
`payout_type`	Payout model (CPA, CPC, etc.)
`payout_amount`	Payout amount
`revenue_type`	Revenue model
`revenue_amount`	Revenue amount
`currency`	Currency code
`custom_payout_revenue`	Custom payout/revenue rule name
`custom_payout_revenue_id`	Custom payout/revenue rule ID

Sub-parameters

Dimension	Description
`sub1`–`sub10`	Affiliate sub parameters (pass separately: `sub1`, `sub2`, …)
`adv1`–`adv10`	Advertiser sub parameters (pass separately: `adv1`, `adv2`, …)

App

Dimension	Description
`project_id`	App project ID
`app_identifier`	App identifier
`bundle_id`	App bundle ID

Available filters

Pass filters as comma-separated inclusive type:value pairs — e.g. offer:123,affiliate:456,country_code:US. Only inclusive filters are supported — there is no exclusion syntax. status is not a supported filter or dimension on this tool; to filter conversions by status, use search_activity(type="conversion") instead.Multiple values (OR): for any ID filter, separate IDs with a pipe to match any of them in one query — offer:1|2|5 returns offers 1, 2, and 5 together. Repeating a key (offer:1,offer:2) does the same. Supported on the ID filters: offer, offer_group, affiliate, advertiser, creative, campaign, category, network, tracking_domain, channel. Prefer one multi-value query over many single-ID queries. list_entities supports the same pipe form on type="coupon_code" (coupon_code, affiliate_id, offer_id), capped at 100 values per filter.Accepted aliases: the _id form and common synonyms are normalized to the canonical key, so you don’t have to guess — offer_id→offer, affiliate_id/partner→affiliate, advertiser_id/brand→advertiser, offer_group_id→offer_group, campaign_id/smart_link→campaign, and the error-code dimensions click_error_code/conversion_error_code→error_code (so you can filter by the same key you grouped a breakdown on). An unrecognized filter key returns an error with a suggestion rather than being silently ignored. Note: campaign is the smart-link/rotator entity; the individual ad campaign is an offer.Entities

Filter	Value format	Example
`offer`	Numeric offer ID	`offer:123`
`offer_group`	Numeric offer group ID	`offer_group:45`
`affiliate`	Numeric affiliate ID	`affiliate:678`
`advertiser`	Numeric advertiser ID	`advertiser:99`
`creative`	Numeric creative ID	`creative:12`
`campaign`	Numeric campaign ID	`campaign:7`
`category`	Numeric category ID	`category:3`
`network`	Numeric network ID	`network:1`
`source_id`	String	`source_id:google`
`event_name`	String	`event_name:purchase`
`offer_url`	Numeric URL ID	`offer_url:8`
`offer_status`	`active`, `paused`, `pending`, `deleted`	`offer_status:active`
`originating_offer`	Numeric offer ID	`originating_offer:123`
`transaction_id`	32-char hex string	`transaction_id:abc123...`
`error_code`	Numeric error code	`error_code:0`
`coupon_code`	String	`coupon_code:SUMMER20`
`tracking_domain`	Numeric domain ID	`tracking_domain:5`
`channel`	Numeric channel ID	`channel:2`

Geo — region/city take internal numeric IDs; resolve a name to its ID with list_entities (type=region / type=city, and type=country for the country_id that scopes a region lookup).

Filter	Value format	Example
`country`	ISO 2-letter code (resolve names via `list_entities` type=country)	`country:US`
`country_code`	ISO 2-letter code	`country_code:US`
`region`	Numeric region ID (resolve via `list_entities` type=region)	`region:1140`
`city`	Numeric city ID (resolve via `list_entities` type=city, which requires `region_id`)	`city:534`

Device & technology

Filter	Value format	Example
`browser`	Browser name	`browser:Chrome`
`device_type`	`desktop`, `mobile`, `tablet`	`device_type:mobile`
`device_platform`	Platform string	`device_platform:iOS`
`device_make`	Manufacturer name	`device_make:Apple`
`device_model`	Model name	`device_model:iPhone 15`
`carrier`	Carrier name	`carrier:Verizon`
`language`	Browser language code	`language:en-US`
`connection_type`	Connection type string	`connection_type:wifi`

People & managers — numeric employee IDs; resolve a name with list_entities type=employee.

Filter	Value format	Example
`account_manager`	Numeric employee ID	`account_manager:55`
`affiliate_manager`	Numeric employee ID	`affiliate_manager:12`
`sales_manager`	Numeric employee ID	`sales_manager:8`
`account_executive`	Numeric employee ID	`account_executive:3`

Sub-parameters

Filter	Value format	Example
`sub1`–`sub10`	String	`sub1:source_a`
`adv1`–`adv10`	String	`adv1:campaign_x`

Billing & labels

Filter	Value format	Example
`label`	Label name	`label:top_affiliate`
`business_unit`	Business unit name	`business_unit:retail`
`affiliate_tier`	Tier name	`affiliate_tier:gold`
`billing_frequency`	Frequency string	`billing_frequency:monthly`
`advertiser_billing_frequency`	Billing frequency	`advertiser_billing_frequency:weekly`

Available metrics

These are also the valid values for sort_by and metric_filters. Call get_report_schema for the same definitions in machine-readable form (metric_definitions).Volume

Metric	Description
`impressions`	Ad impressions served
`gross_clicks`	All clicks before dedup/validation
`clicks`	Total valid clicks (post-dedup)
`unique_clicks`	Distinct clicks within the dedup window
`duplicate_clicks`	Clicks dropped as duplicates
`invalid_clicks`	Clicks rejected by validation (geo, cap, fraud, …)
`conversions`	Base conversions; excludes scrubbed and view-through
`invalid_cv_scrub`	Conversions rejected by a scrub rule
`view_through_cv`	View-through conversions (impression-attributed, no click)
`total_conversions` / `total_cv`	Conversions + scrubbed + view-through (does not include events)
`events`	Additional conversion events beyond the base conversion

The conversion family reconciles exactly: total_conversions − conversions − invalid_cv_scrub − view_through_cv == 0. Use invalid_cv_scrub to size scrubbing rather than inferring it from the gap.Revenue & cost

Metric	Description
`payout`	Amount paid to partners
`revenue`	Revenue from advertisers, including event revenue
`profit`	`revenue − payout`
`gross_sales`	Sale amount attributed to conversions
`media_buying_cost`	Partner-reported media buying cost

Rates & efficiency

Metric	Description
`ctr`	Click-through rate = `clicks ÷ impressions` (percent)
`cvr`	Conversion rate = `total conversions ÷ clicks` (percent)
`epc`	Earnings per click = `(revenue − payout) ÷ clicks`
`rpc`	Revenue per click = `revenue ÷ clicks`
`cpa`	Cost per acquisition = `payout ÷ conversion`
`cpc`	Cost per click = `payout ÷ clicks`
`cpm`	Cost per thousand impressions = `payout × 1000 ÷ impressions`
`roas`	Return on ad spend = `gross_sales ÷ revenue` (ratio)
`margin`	`(revenue − payout) ÷ revenue` (percent)

cvr, ctr and margin are percentages (2.84 = 2.84%); roas is a ratio. This matters in metric_filters: cvr<2 means “under 2%”.metric_definitions from get_report_schema also lists your network’s custom metrics, marked is_custom: true and carrying their formulas. Those entries are definitions only — not computed by this tool and never present in totals. They are also not usable in sort_by or metric_filters: an unrecognized sort_by name is silently ignored rather than rejected, so a report that looks unsorted usually means the metric name didn’t match. Sort and filter on the built-ins above.

run_network_summary

Returns headline performance totals for a date range — no grouping, just aggregate numbers. Optionally compares against the prior equivalent period, and can be scoped to specific entities. Requires: Reporting (Read Only) Ask for it: “How did the network do last week compared with the week before?” Parameters

Parameter	Type	Required	Default	Description
`from`	string	Yes	—	Start date (`YYYY-MM-DD`)
`to`	string	Yes	—	End date (`YYYY-MM-DD`)
`include`	string	No	—	`comparison` — adds prior-period metrics for delta analysis. Max 256 chars
`filters`	string	No	—	Scope the totals to specific entities — comma-separated `type:value` pairs. Keys: `offer`, `affiliate`, `advertiser`, `creative`, `campaign` (pipe for OR: `offer:1\|2`). Applied to the comparison period too. Max 4,096 chars
`timezone`	string	No	Network timezone	IANA timezone name. Max 64 chars
`currency`	string	No	Network base currency	Currency code. Max 8 chars

Example

run_network_summary(from="2026-08-25", to="2026-08-31",
                    include="comparison", filters="offer:91")
→ offer 91's headline totals for that week, each with its prior-week value and delta

Returns Headline totals for the window, plus applied_query (the exact resolved query — from, to, resolved timezone and currency, and any filters and include; empty sections omitted) and the same metric_glossary object as run_performance_report. Gotchas

Pair filters with include=comparison for a filtered period comparison — “is offer 91 up or down vs last week?”
Only the five entity keys above are supported here. For breakdowns by country, device, sub-parameter or error code, use run_performance_report — those dimensions don’t exist on the summary.
Timezone resolution works exactly as it does for run_performance_report, including timezone_requested and timezone_note.

search_activity

Searches raw event records within a date-time window. Set type to choose the stream. Requires: Reporting (Read Only) Ask for it: “Show me yesterday’s rejected conversions on offer 1234.” Parameters

Parameter	Type	Required	Default	Description
`type`	string	Yes	—	`click` or `conversion` — selects the event stream
`from`	string	Yes	—	Start datetime (`YYYY-MM-DD` or `YYYY-MM-DD HH:MM:SS`), network timezone
`to`	string	Yes	—	End datetime, same format
`offer_id`	string	No	—	Filter by offer ID. Max 64 chars
`affiliate_id`	string	No	—	Filter by affiliate ID. Max 64 chars
`advertiser_id`	string	No	—	Filter by advertiser ID. Max 64 chars
`country`	string	No	—	2-letter ISO country code. Max 64 chars
`sub1`–`sub10`	string	No	—	Filter by sub parameter value. Max 600 chars each
`source_id`	string	No	—	Filter by traffic source ID. Max 600 chars

Type-specific:

Parameter	Applies to	Default	Description
`error_code`	`click`	—	Filter by click error code
`status`	`conversion`	—	`approved`, `pending`, `rejected`, `invalid` (case-insensitive; an unrecognized value errors). Reporting categories, not the raw `conversion_status` — see below
`event_name`	`conversion`	—	Filter to a named post-conversion event (e.g. `Purchase`). Base conversions have no event name
`adv1`–`adv10`	`conversion`	—	Filter by advertiser sub parameter value. Max 600 chars each
`order_id`	`conversion`	—	Filter by the merchant order ID on the conversion
`email`	`conversion`	—	Filter by the consumer email on the conversion. Returned values stay GDPR-masked
`page_size`	`conversion`	50	Records per page (1–100)
`cursor`	`conversion`	—	Pagination cursor from a prior response — reuse the same filters and window

Example

search_activity(type="conversion", from="2026-08-31 00:00:00", to="2026-08-31 23:59:59",
                offer_id="1234", status="rejected", page_size=100)
→ up to 100 scrubbed or geo-blocked conversions on offer 1234 that day, with next_cursor

Returns

Field	Type	Description
`clicks` / `conversions`	array	The records, keyed by the requested `type`
`returned`	number	`conversion` — records in this page
`page_size`	number	`conversion` — page size applied
`has_more`	boolean	`conversion` — more records match; pass `next_cursor` back as `cursor`
`next_cursor`	string	`conversion` — cursor for the next page; empty on the last page
`total`	number	`click` — records returned
`capped_at`	number	`click` — hard ceiling applied (`1000`)
`truncated`	boolean	`click` — more clicks matched than were returned
`total_matching`	number	`click` — the true number of matching clicks when `truncated`
`note`	string	Guidance when more records exist — page with `next_cursor`, narrow the window, or switch to `run_performance_report` for counts

Per-record fields — type="click" transaction_id, timestamp, offer_id, affiliate_id, advertiser_id, campaign_id, error_code, error_message, is_unique, is_view_through, is_test_mode, payout, revenue, currency, country, region, city, browser, platform, device_type, os_version, user_ip, referer, coupon_code, previous_transaction_id, sub1–sub10, source_id. Empty values are omitted. campaign_id is returned only when the click is on a smart link. offer_name, affiliate_name, advertiser_name and has_conversion are not on this stream — each needs a per-record lookup, so use get_entity(type="click") when you need them. Per-record fields — type="conversion" conversion_id, transaction_id, timestamp, click_timestamp, status, error_code, offer_id, offer_name, affiliate_id, affiliate_name, advertiser_id, advertiser_name, payout, revenue, sale_amount, payout_type, revenue_type, currency, event_id, event_name, order_id, country, region, city, platform, device_type, browser, os_version, referer, language, brand, dma, session_user_ip, conversion_user_ip, http_user_agent, isp, carrier, app_id, idfa, google_ad_id, android_id, sub1–sub10, adv1–adv10, source_id. Empty or zero fields are omitted — coupon_code, email, notes, is_scrub, is_view_through and adv1–adv10 appear only when populated. event_name is resolved on this stream, and a non-zero event_id means the row is a post-conversion event. error_message and campaign_id are not on this stream — use get_entity(type="conversion") for those.

status filters on reporting categories, not on the conversion’s own status field. The four values don’t map one-to-one onto the stored conversion_status:

Filter value	Matches
`approved`	`conversion_status = approved`
`pending`	`conversion_status = pending`
`rejected`	Only conversions that were scrubbed or geo-blocked
`invalid`	Conversions that failed for any other reason — including those whose own `conversion_status` reads `rejected`

So a conversion showing "status": "rejected" in its own payload is returned by status="invalid" when it was neither scrubbed nor geo-blocked, and is not returned by status="rejected". This is intentional and matches the REST conversion report. The per-row status is always the raw value, so it will not always equal the filter that matched it — don’t read a mismatch as a bug.Known gap: a conversion whose conversion_status is rejected with error_code 1 and no scrub flag satisfies neither branch — rejected requires the scrub flag, invalid excludes error_code 1 — so no status filter value returns it, although it does appear in an unfiltered query. Don’t treat a set of status-filtered calls as an exhaustive partition of the window.

Gotchas

Pagination differs by type. type="conversion" is paginated (page_size + cursor; page while has_more is true). type="click" is not — it returns a single set capped at 1,000 records, most-recent first.
Window rules differ by type. click has a maximum 14-day window. conversion needs a minimum of one full day in the network timezone (00:00:00 → 23:59:59); shorter windows may return an error.
Passing a filter that doesn’t apply to the chosen type (e.g. status with type="click") returns an error rather than being ignored.
Results are filtered by affiliate visibility. On a limited-scope account, rows for affiliates outside that scope are silently excluded — counts can look lower than expected when you query without an affiliate_id filter.
email is obfuscated for conversions from GDPR countries. See Event response fields.
For counts and breakdowns rather than raw records, use run_performance_report (e.g. dimensions=click_error_code or conversion_error_code).

Single-event and transaction lookup

Single events are served by the generic get_entity tool — type="click", type="conversion", or type="transaction" for the full attribution chain. For response fields and a walkthrough of using them together, see Attribution debugging.

​get_report_schema

​run_performance_report

​Period-over-period comparison

​Custom metrics

​Reference

​run_network_summary

​search_activity

​Single-event and transaction lookup

get_report_schema

run_performance_report

Period-over-period comparison

Custom metrics

Reference

run_network_summary

search_activity

Single-event and transaction lookup