Audience creation: regional impact on email processing?


We have a client who is seeing a discrepancy between the number of matched contacts from an audience uploaded via our Social Audiences application (uses Advertiser Apis) and an audience created directly using Twitter Audience Manager. We used the same data with our internal account, creating an audience using our app and then in Audience Manager, and the number of matched contacts matched between the two audiences (although significantly less matches than the same data uploaded by the client). We are located in the US, our application is not internationalized/localized, and the client is in Japan. So, we wanted to find out if there could be a chance that the data is being processed differently (international characters or whatever) based on the locale from which it was uploaded?

silverpop_ea account:

client account:

Any assistance or information you can offer will be greatly appreciated.

–Frank Lacour


Hey @silverpop_ea

Can we get some additional details around the issue, specifically:

  1. The Ads account_id for both yours and the client account
  2. The Audience ids of the Audiences exhibiting this behavior
  3. Timestamps (if possible) for when both of the Audiences were uploaded
  4. The API request and response bodies

However, it is important to note that your audience match rates are impacted by two additional factors:

  • The method in which users log on to Twitter
  • User activity

For example, users accessing Twitter via a 3rd party such as HootSuite are not included in the matched audience. Additionally, users who have not logged on to the platform in the last 30 days are also not included.

Additionally, please note that this criteria is universal, and international audiences aren’t processed differently.


client: 18ce546xe5i
ours: 18ce53z8vuo

Audience ids:
2257h (created/upload from our Social Audiences app on 2017-04-10 11:38:18)
2257r (don’t have more than a date listed in audience manager)
22fkb (created/upload from our Social Audiences app on 2017-04-17 17:34:26)
22fkc (don’t have more than a date listed in audience manager)

We do not store the request and response bodies, so those are, unfortunately, not available. I understand there are several factors that can impact match rates, and am trying to find out what could cause the match rate to be different between audiences that are based on the same data/contacts.

–Frank Lacour


@imit8me, were you able to find anything based on the provided account/audience ids?


Can someone please provide a status update? Why would these audiences based on the same data return different number of matches on Twitter platform?


Hey @silverpop_ea

I just wanted to let you know that we’re currently trying to reproduce this issue, internally and we will get back to you once we’ve got more details around what may be going on.



Ok, much appreciated. Please let us know what you are able to find. Thanks!


Hi @silverpop_ea!

We ran an internal test using an audience of mobile IDs already hashed.

The following took place through the API:

EX ----

twurl -X POST -H "/1/accounts/18ce5475gij/tailored_audiences?name=api_prel_22385_test2&list_type=DEVICE_ID" | jq
  "data_type": "tailored_audience",
  "data": {
    "targetable": false,
    "name": "api_prel_22385_test2",
    "targetable_types": [
    "audience_type": "CRM",
    "permission_level": "READ_WRITE",
    "is_owner": true,
    "id": "23f0s",
    "reasons_not_targetable": [
    "list_type": "DEVICE_ID",
    "created_at": "2017-05-04T16:42:44Z",
    "updated_at": "2017-05-05T20:03:49Z",
    "partner_source": "OTHER",
    "deleted": false,
    "audience_size": null
  "request": {
    "params": {
      "account_id": "18ce5475gij",
      "name": "api_prel_22385_test2",
      "list_type": "DEVICE_ID"

We then uploaded the same audience through the Ads Dashboard and selecting mobile advertising IDs with pre hashed selected. The results of the audience sizes were exactly the same.

Our recommendation to alleviate differences would be to make sure that the audience file is normalized first before each party uses the data. By normalized, I mean pre-hashed and then uploading through either endpoint.