Historic daily Ads data metric unstable

[drsagitn]

Hello,

I made some segmented async requests for Engagement metric over Campaign entity. Below is the results
CampId===Placement=========Impression==Data Timestamp=======Request timestamp
bak6j=====ALL_ON_TWITTER===13421======2018-09-25 00:00:00====2018-09-25 20:14:00
bak6j=====ALL_ON_TWITTER===17702======2018-09-25 00:00:00====2018-09-26 18:20:30
bak6j=====ALL_ON_TWITTER===17760======2018-09-25 00:00:00====2018-09-27 20:36:34

It’s expected that daily data will be finalized one day after the data timestamp.
But we can see that data value (impression) of 2018-09-25 timestamp was changed every day even 2 days after the data timestamp (2018-09-27)
Could someone please give explanation for this change? And what is the maximum time frame we can expected the daily data to be stable?

Regards,
Thinh

[drsagitn]

Hello,

Anyone please let me know about this issue?

Thank you

[imit8me]

Thank you for raising this issue! Can you confirm a few details like the Ads account_id and the exact request/response bodies for the API call?

Ideally there should be no changes to analytics stats 24 hours after the campaign has ended, so we’d like to confirm the exact end_date for the campaign to ensure that the campaign was not serving when the requests for stats were made. I’d also recommend taking a look at our Analytics best practices for more details around stats, as well.

Hope that helps!

Thanks!

[drsagitn]

Thanks for replying.
The account_id is 18ce54qwn2r
The campaign end date is 2018-09-26

The 1st request to async API:
twurl -X POST -H ads-api.twitter.com “/4/stats/jobs/accounts/18ce54qwn2r?start_time=2018-09-23&end_time=2018-09-26&segmentation_type=PLATFORMS&granularity=DAY&metric_groups=ENGAGEMENT&placement=ALL_ON_TWITTER&entity=CAMPAIGN&entity_ids=bak6j”

Response:
{“request”:{“params”:{“start_time”:“2018-09-22T15:00:00Z”,“segmentation_type”:“PLATFORMS”,“entity_ids”:[“bak6j”],“end_time”:“2018-09-25T15:00:00Z”,“placement”:“ALL_ON_TWITTER”,“granularity”:“DAY”,“entity”:“CAMPAIGN”,“metric_groups”:[“ENGAGEMENT”]}},“data”:{“start_time”:“2018-09-22T15:00:00Z”,“segmentation_type”:“PLATFORMS”,“url”:null,“id_str”:“1047427356065841153”,“entity_ids”:[“bak6j”],“end_time”:“2018-09-25T15:00:00Z”,“country”:null,“placement”:“ALL_ON_TWITTER”,“id”:1047427356065841153,“expires_at”:null,“account_id”:“18ce54qwn2r”,“status”:“PROCESSING”,“granularity”:“DAY”,“entity”:“CAMPAIGN”,“created_at”:“2018-10-03T10:05:16Z”,“platform”:null,“updated_at”:“2018-10-03T10:05:16Z”,“metric_groups”:[“ENGAGEMENT”]}}

The 2nd and 3rd requests were performed days later with same parameters except ‘end_time’ updated to 2018-09-27 and 2018-09-28

Regards

[drsagitn]

Any update please?

[drsagitn]

Please help…

[drsagitn]

2 months later, still silent

[jrsyo]

Hi @drsagitn ,

Sorry for not responding this earlier. You mentioned:

The 2nd and 3rd requests were performed days later with same parameters except ‘end_time’ updated to 2018-09-27 and 2018-09-28

after this job completion, you download the JSON file and parse it as needed, right?
In my understanding, it has id_data array in its main data array, and it contains multiple objects per segment type like iOS, Desktop and etc… and the metrics have each data depending on time_series_length (e.g., if "time_series_length": 3 then "impressions": [XXXX, XXXX, XXXX]), so you are referencing a particular index data of this impressions array and it’s unstable, correct?

Is this happening only with impression or other metrics as well? Also, do you see the same result if you initiate a job with the exact same parameters but on a different day?
like, 1st call => 2018-11-26, 2nd call => 2018-11-27 with same start_time and end_time

Best,
Shohei

[drsagitn]

Hi,

after this job completion, you download the JSON file and parse it as needed, right?

Yes

so you are referencing a particular index data of this impressions array and it’s unstable, correct?

Yes

Is this happening only with impression or other metrics as well?

Other metrics like ‘retweets, likes, engagements, clicks’ have the same unstable problem.

Also, do you see the same result if you initiate a job with the exact same parameters but on a different day?

After 2018-09-27, all requests return same result, this means the data has been finalized since 2018-09-27. But it is expected to be finalized on 2018-09-26.