How to grow the rate limit when handle with large-scale Twitter Ads

[liyouliang]

Hi, I have some questions about the rate limiting :

1, According to dev documentation (https://dev.twitter.com/ads/basics/rate-limiting), there are 1,000 Rate Limit per 15-minute window on the Analytics endpoint. Actually there only 250 Rate Limit per 15-minute window.

Response example:
{‘x-rate-limit-remaining’: ‘0’, ‘x-response-time’: ‘7’, … ‘x-rate-limit-reset’: ‘1473376070’, ‘date’: ‘Thu, 08 Sep 2016 22:56:41 GMT’, ‘x-rate-limit-limit’: ‘250’, ‘content-type’: ‘text/plain;charset=ut
f-8’}

2, The overview page(https://dev.twitter.com/ads/overview) tell us: “Developer-level partners are limited to five OAuth tokens”. But in my Application Management can only generate one Access Token. There are just options for “Regenerate My Access Token and Token Secret”.

3, When I handle with large-scale Twitter Ads, how could I check in real time on the performance of ads, and adjust it according to the performance if I have only 250 rate limit per 15-minute window.

[JBabichJapan]

1. The rate limit is expected to be different between Synchronous and Asynchronous endpoints, can you clarify which you are referring to?

2. The five tokens are access tokens, those tokens for the count of people or individual Ad Accounts who auth to use your app, the buttons there are only about the consumer tokens for the app itself (the ones used to make API calls with).
   See http://stackoverflow.com/questions/28057430/what-is-the-access-token-vs-access-token-secret-and-consumer-key-vs-consumer-s

3. There are some best practice guidelines https://dev.twitter.com/ads/analytics/best-practices which are generally applicable to both fetching synchronous and asynchronous - with the guidance of trying to not fetch redundant data when you don’t need to (fetching the same unchanging data over and over is what will make your bandwidth too limited). For realtime optimization scenarios, we recommend to work on that after making sure that historical data is being synced with a best practices style algorithm via async endpoints so that the synchronous endpoint bandwidth can primarily be directed towards pulling stats for realtime scenarios.