Disrepancy when getting data from Reports API endpoint

[developerf43676]

Hi,

I have a question regarding the https://analytics.adobe.io/api, specifically the /reports endpoint. We’re seeing a potential data discrepancy reported by one of our clients. They observe one set of values in the Adobe Analytics UI and another in the data fetched via the API and stored in their destination system, which is handled by Fivetran.

The discrepancy is partial - out of 9 metrics being pulled and stored, 6 match the UI values exactly. However, the following 3 metrics show mismatched values:

• visits

• unique_visitors

• mkz_qualified_visits

We don’t perform any transformations on these values - we request them from the API and write them to the destination as-is. I would appreciate any guidance or context around why such discrepancies might occur for these specific metrics, and whether there are any known limitations or caveats with the /reports endpoint that could explain this behavior.

This is an example of the request performed by Fivetran:

curl -X POST --data '{"rsid":"investorsgrmackenzieprod","globalFilters":[{"type":"dateRange","dateRange":"2025-01-01T00:00:00.000/2025-01-02T23:59:59.999"},{"type":"segment","segmentId":"s300005393_5f9c2f0d0d9f0c5e4e8e0d3d"}],"metricContainer":{"metrics":[{"columnId":"0","id":"metrics/downloadlinkinstances","filters":["0","1"],"sort":"desc"},{"columnId":"1","id":"metrics/itemtimespent","filters":["0","1"],"sort":"desc"},{"columnId":"2","id":"metrics/visitors","filters":["0","1"],"sort":"desc"},{"columnId":"3","id":"metrics/pageviews","filters":["0","1"],"sort":"desc"},{"columnId":"4","id":"metrics/visits","filters":["0","1"],"sort":"desc"},{"columnId":"5","id":"metrics/event7","filters":["0","1"],"sort":"desc"},{"columnId":"6","id":"metrics/bounces","filters":["0","1"],"sort":"desc"},{"columnId":"7","id":"cm300005393_6729221be248b00df166f48e","filters":["0","1"],"sort":"desc"},{"columnId":"8","id":"cm300005393_673ca150e248b00df16719ea","filters":["0","1"],"sort":"desc"}],"metricFilters":[{"id":"0","type":"breakdown","dimension":"variables/daterangeday","itemId":"1250001"},{"id":"1","type":"breakdown","dimension":"variables/marketingchannel","itemId":"8"}]},"dimension":"variables/campaign","settings":{"limit":1000,"page":0}}' -H 'Accept: application/json' -H 'Authorization: Bearer xxx' -H 'x-proxy-global-company-id: investb' -H 'x-api-key: eabc8d8f5fc84d8ebea69e2b8eeddb55' -H 'Content-Type: application/json' -A 'Jersey/2.45' 'https://analytics.adobe.io/api/investb/reports'

The same query pattern is used to fetch data for the entire month of January. For comparison purposes, I’ve highlighted the columns where discrepancies were observed, along with the page_views column, which matches the UI perfectly. A screenshot of the output file is attached below to illustrate this.

Could someone take a look at this and help us determine whether there’s something off in the way we’re fetching data from the API, or if there’s indeed an ongoing discrepancy between the API and what’s shown in the UI?

Kind regards,
Fivetran Developers

[AmitVishwakarma]

Hi @developerf43676 ,

There are three primary causes why values from Adobe's /reports API may differ from the Adobe Analytics UI:

1. Segment Scope or Configuration Mismatch

You’re applying this filter:

{"type":"segment","segmentId":"s300005393_5f9c2f0d0d9f0c5e4e8e0d3d"}

If the UI segment used to view the data isn’t the same segment ID (or includes different definitions: visit vs. hit-level), the values will differ.

Solution: Double-check the segment in the Adobe UI against the one used in the API by opening the Segment Builder in Adobe Analytics and ensuring it's the same segment ID and scoping (visit/hit/visitor level).

2. Filters or Breakdowns in API Affecting the Totals

You're applying multiple filters via:

"metricFilters":[{"id":"0","type":"breakdown","dimension":"variables/daterangeday","itemId":"1250001"},{"id":"1","type":"breakdown","dimension":"variables/marketingchannel","itemId":"8"}]

These breakdown filters can limit or slice data differently than the UI’s total metrics view (especially if UI shows summary, not breakdowns).

Solution:

  - Try removing metricFilters entirely to fetch pure totals, unless your intent is to only get e.g., “Direct” marketing channel (itemId: 8).

  - Alternatively, explicitly reproduce the exact UI breakdowns (like channel filtering or specific campaign filtering) in your API query.

3. Data Latency / Processing Timing

Data in Adobe Analytics can lag or be reprocessed. Depending on when Fivetran pulls the data vs. when UI loads it, slight differences can emerge.

Especially relevant for day-level metrics near the end of the day/month.

Solution:

Use the API parameter "useCache": false or pull data with a delay (e.g., pull January’s data only after February 5).

Confirm data freshness settings and report suite latency policies with Adobe.

Here's how to ensure you match the Adobe UI exactly:

Step 1: Simplify API Query to Match UI Summary

Use a query like this to fetch totals by day without breakdown filters:

{
  "rsid": "investorsgrmackenzieprod",
  "globalFilters": [
    {
      "type": "dateRange",
      "dateRange": "2025-01-01T00:00:00.000/2025-01-31T23:59:59.999"
    },
    {
      "type": "segment",
      "segmentId": "s300005393_5f9c2f0d0d9f0c5e4e8e0d3d"
    }
  ],
  "metricContainer": {
    "metrics": [
      {"id": "metrics/visits"},
      {"id": "metrics/visitors"},
      {"id": "cm300005393_6729221be248b00df166f48e"}  // mkz_qualified_visits
    ]
  },
  "dimension": "variables/daterangeday",
  "settings": {
    "limit": 1000,
    "page": 0
  }
}

Key improvements:
Removed unnecessary breakdown filters (e.g., marketingchannel).
Focused on daily granularity via daterangeday.
Removed filters on metrics – these can change meaning in unexpected ways.

Step 2: Align Segment Definitions

Open the segment used in the API (s300005393_5f9c2f0d0d9f0c5e4e8e0d3d) and compare its logic with the Adobe UI segment:

Same dimension filters?

Same container scope (visitor vs. visit vs. hit)?

No exclusions or rules applied in UI but not in API?

tep 3: Re-fetch and Compare Fresh Data

After ensuring the above:

Clear cache ("useCache": false in settings if supported).

Wait 48–72 hours after month-end for data stabilization.

Re-pull January data using the updated query.

Regards,
Amit