#1

Hello,

We’re hitting the v2 /tweets endpoint to retrieve public_metrics for multiple Tweets. In our testing, we’ve noticed that after a Tweet is liked, there is sometimes latency of up to several hours before the new Like count is returned from the API. For example, the other day we liked 6 Tweets around 1:30pm EDT and 5 out of the 6 Likes did not appear in the API until around 8pm EDT. Is this expected behavior? Are public_metrics counts not considered to be real-time? Has anyone else experienced this?

Thanks for any insight anyone can provide!

#2

Hey @CRSDevs, I’m glad I’m not the only one! However, likes update immediately for me whilst profile_clicks, url_clicks, and impressions do not.

It’d be great to get some insight from the Twitter team on this, there could well be some caveats but I’ve not been able to find them.

Edit: if you’re looking to count non_public_metrics from the API, it helps if the tweet actually has some to begin with - my bad!

#3

Pretty much our entire infrastructure is built around eventual consistency (barring the “firehose” or new v2 filtered stream), so I’m not surprised by these analytics findings. What’s your specific need and your expectation? That might be a better starting point for an understanding of what you’d prefer.

#4

Hey Andy, for me it was to have access to metrics within a ~30-minute window, this seems perfectly possible from what I’ve experienced with the API.

#5

Hello, thanks for the update. We’re tracking public_metrics deltas over time for Tweets we’ve retrieved via v2 recent search. We always assumed that a change in a metric that’s visible on the Twitter front end would be reflected in the API within at most a few minutes. Is that reasonable in most cases? Thanks!

#6

“In most cases” that seems sensible. In extremely high volume cases based on account or Tweet popularity I don’t think our team would be able to provide more granular timed infomation.