Follower Count exposed on users endpoint


I think there are many devs still waiting for this to be implemented to continue with the migration. There is a TOP suggestion waiting for response since Apr, 2019 on uservoice but nothing has been done yet.

We need to know if this is going to be implemented or not so we can continue with the migration. We already know and accepted a lot stuff has been gutted from v5 in the new API like summary endpoint being removed etc.

Please let us know so We can plan accordingly if We are going to make one request vs 100 if this is not implemented.


Assume it won’t be.
Generally speaking linking two bits of data on the backend is “expensive” and most use cases of the follower endpoint don’t use follows they just grab the user data after login. So Twitch is/was wasting a lot of CPU cycles on data most people don’t consume from the response

In most of my use cases for I don’t need the follow data when loading a user, Just the user name/login

I know and understand that it costs them extra resources to expose follower count on the users endpoint but I am pretty sure it is much better than having people make thousands of calls to follows api instead of hundreds to users endpoint.

In my use case and many other who signed the Uservoice link above, it is clear that We will be making 100 calls to follows endpoint instead of 1 request to users endpoint. This is going to cost Twitch more cycles/resources in the end imo.


Personally I seed a users follow count once per day from the follows endpoint
Then use eventsub to keep up to date with new follows.

Not really, given your name I’m assuming you are keeping stats.

So now you load user once per day to monitor for name changes (or use the user change eventsub)

And then hammer the follows endpoint for follow count update or seed it once per day with the count and use eventsub to count follows up.

Or just don’t bother counting follows since it’s not much of a useful metric generally speaking.

It’s also worth considering that Twitch doesn’t want to support this sort of external stat gathering.

It’s possible to consider that follows is an item of PII that might be protected via GDPR and other privary laws, and if you are collecting users follower counts without permission you might be in violation of these laws.

And if you do have permission then the eventsub topics for the users you have permission to collect are “free”/zero cost. And don’t count to your limit.

Not really.

100 calls to follows is 100 calls to 1 database table.

1 call to users for 100 users in one go (if it contained follow count), is 100 database table joins and then spit back the response. And if the only data you are after is the follow count then it’s all burnt.

Depending on how this is architectured internally at Twitch it might not be that at all and it’s buried in multiple services. (Which might be the case when we consider Twitch is working on anti follow bot measures, or external Services such as Root’s Followbot Removal tool essentially DDoS Twitch Daily with follow removals)

Thus requiring extra cycly times to get data when joining two data types.

That is likely why Twitch doesn’t have follows on the users endpoint as it allows doing 100 users at once on the uses endpoint.

TLDR: I really don’t think we will see the follow count on the new users endpoint. Since the new users endpoint allows 100 users at once.

We’d be more likely to see a GQL type solution. And we all know what a mess that seems to be

All possible solutions, along some other stuff.

Changes to the new api clearly shows this is the case tho I would like Twitch to say it out load.

Although, it is not the most important metric, it still gives idea about a popularity, longevity and brand of a streamer.

Aggravated statistical data is not considered Personal Information.

Obviously, I agree with you but would like a clarity.

Also, Pls teach Cohh how to drive in Halo, He sucks ;).

1 Like