Google Ads data deviation

  • Is this your first time deploying Airbyte?: No
  • OS Version / Instance: Ubuntu
  • Deployment: Docker
  • Airbyte Version: What version are you using now?
  • Source name/version: google-ads v0.1.40
  • Destination name/version: Postgres
  • Step: The issue is happening when reading the data from the connector
  • Description: The sum of clicks and impressions for a specific client fetched from Aibyte differs from the values we see in the UI.

Hi @iberchid, thanks for the post.

Could you upgrade the Google Ads connector to the latest version? It’s currently at v.0.1.44. After which, can you rerun the sync and report back with the logs? Let me know if you need help or additional clarification!

Hello @sajarin , thanks for your reply. I upgraded the connector to the latest version, and still see the discrepancies.

Airbyte Google Ads UI
campaign.id Sum - metrics.clicks Sum - metrics.impressions Sum - Clicks Sum - Impr.
1750830383 54 214 127 456
1750830386 127 1516 2850 1308
1750830389 16 451 180 4698
1750830392 41 845 334 11146
1750830395 2 352 380 9899
2055644298 12 205 142 3605
2055644301 1 177 881 8734
2055645720 72 921 114 1559
2055645723 0 9 0 10
6478660293 85 1139 9358
6517485353 2 217 219 3796
12570602353 15 188 58 739
12571858005 5 354 4972
12600612400 68 861 211 9033
12984726867 1 38 682 14318
12986499089 3 267 519 7735
12986499092 15 226 123 6059
12986499095 2 600 78 4970
12986635859 35 1161 3459 19095
12986707427 1 162 171 3777
12986707430 6 228 1131 17733
12986707433 1 179 1128 11491
12986707565 0 111 1066 11168
13602429568 8 351 719 11806
14456891208 1 76 106 2180
14872460564 10 118 291 2915
15225232797 18 203 2173 6205
15426279029 1 131 388 7824
15590103962 1 73 873 6525
17328526511 2555 1849445 15536
Total Result 3158 1860818 48269 188784

Hey @iberchid,

Thanks for getting back to me, sorry for the delay here. Could you describe what your configuration looks like for this sync? Just to check all of the trivial boxes here: do both tables represent the same time period? I believe the Google Ads connector has an option for start_date and end_date, did you specify these values? It looks the data deviation is consistent across most of the campaigns which is why I’m asking about such a trivial detail.

If that’s not the issue, could we try maybe running a much smaller sync and see if that syncs properly without any deviation. Furthermore, what sync mode are you using for this stream? Let me know and thanks for being patient, I definitely want to get this working for you.

Hi @sajarin , no problem, thanks for your support, here are the requested info:
of course I am aware of the start and end dates in the config file, so here is what it looks like

{
  "credentials": {
	  "developer_token": "my_developer_token",
	  "client_id": "my_client_id",
	  "client_secret": "my_client_secret",
	  "refresh_token": "my_refresh_token"
  },
  "start_date": "2022-05-31",
  "end_date": "2022-06-30",
  "customer_id": "7198268996",
  "login_customer_id": "7732297609",
  "conversion_window_days": 0
}

Furthermore, I am filtering on the dates after normalization. Both tables represent data from June 2022. I also compared day by day data for the same time period and they were different. Here are the examples for a few days:

Airbyte UI
customer.id campaign.id segments.date metrics.impressions metrics.clicks Impr. Clicks
7198268996 12986707427 2022-06-27 7 0 169 7
7198268996 17328526511 2022-06-12 94150 143 504
7198268996 12986707565 2022-06-29 5 0 404 37
7198268996 12570602353 2022-06-11 6 1 10 1
7198268996 13602429568 2022-06-30 12 0 561 26
7198268996 1750830392 2022-06-07 56 5 475 17
7198268996 12986499095 2022-06-12 16 0 175 5
7198268996 12986499095 2022-06-15 27 0 289 4
7198268996 2055644298 2022-06-25 6 0 70 0
7198268996 15426279029 2022-06-08 3 0 332 15
7198268996 1750830392 2022-06-05 39 1 518 11
7198268996 12570602353 2022-06-20 2 0 39 4
7198268996 12600612400 2022-06-23 21 2 224 7
7198268996 1750830383 2022-06-06 1 0 16 5
7198268996 1750830386 2022-06-21 1 0 52
7198268996 2055644298 2022-06-29 5 0 189 7
7198268996 2055645720 2022-06-08 51 5 51 5
7198268996 12571858005 2022-06-20 7 0 169
7198268996 1750830386 2022-06-17 5 0 205
7198268996 14456891208 2022-06-16 1 0 84 4
7198268996 15426279029 2022-06-11 4 0 186 13
7198268996 12986499092 2022-06-10 5 0 192 1
7198268996 6478660293 2022-06-06 44 1 257
7198268996 15590103962 2022-06-03 6 0 297 36
7198268996 12571858005 2022-06-16 16 0 222
7198268996 2055644301 2022-06-04 1 0 136 8
7198268996 1750830383 2022-06-27 7 1 10 1
7198268996 1750830389 2022-06-03 2 0 195 2
7198268996 17328526511 2022-06-27 33098 45 493
7198268996 6478660293 2022-06-21 27 1 318
7198268996 15590103962 2022-06-11 2 0 107 21
7198268996 15225232797 2022-06-05 27 7 846 54
7198268996 15426279029 2022-06-28 2 0 112 9
7198268996 1750830389 2022-06-02 3 0 255 12
7198268996 13602429568 2022-06-22 2 0 520 33
7198268996 15590103962 2022-06-22 4 0 306 37
7198268996 6517485353 2022-06-09 13 0 212 14
7198268996 2055645720 2022-06-28 11 0 11 0
7198268996 12986499089 2022-06-09 28 0 315 17
7198268996 15590103962 2022-06-16 2 0 292 37
7198268996 1750830386 2022-06-05 3 0 55
7198268996 12986499089 2022-06-30 13 0 182 14
7198268996 12600612400 2022-06-07 21 2 189 6
7198268996 2055644298 2022-06-12 1 0 44 2
7198268996 13602429568 2022-06-20 21 0 398 24
7198268996 2055645720 2022-06-13 1 0 71 6
7198268996 1750830395 2022-06-06 14 1 295 11
7198268996 12986707427 2022-06-14 11 0 150 9
7198268996 14456891208 2022-06-28 6 0 121 6

The syn mode we are using is full_refresh

      "sync_mode": "full_refresh",
      "destination_sync_mode": "append_dedup"

Hello, did you check your database to validate there is any left data from previous syncs?
Do you mind create a new connection and sync using full refresh and share the values?

Hey @marcosmarxm , the results above were actually got with a new connection and a separate table.

Hey @iberchid,

Thanks for following up here. My recommendation here is to run a fresh instance of Airbyte and run a full refresh overwrite sync. It’s possible that there may be some _tmp tables or _scd tables from a prior sync that messed up your previous syncs.

Finally consider upgrading the connector as there is a now a new version. This may solve your problem but it may not.