Proactively avoiding app tripping spam filter

video
api

#1

Hello! Not convinced a public forum is the right place for this question, but I was referred here by the twitter dev account.

My app was unfortunately caught by the spam filters a few weeks ago when I was testing it. It makes sense as I probably posted several tweets with the same message in a row. However, we also got two unfortunate warnings: (1) saying that repeat offenders will be subject to permanent suspension and (2) saying that “Always and/or only including a video will trigger spam filters.”

Our app always includes a video in each tweet by design. Every interaction in the app, though, is genuine and there is no unfair automation: you have to put in several minutes of work app-side in order to compose what will eventually turn into a tweet on one or several twitter accounts. We are putting a lot of effort into ensuring tweets that are produced in our app are high-quality and organic. We are even keeping a closed process for onboarding users onto our app for now.

Which brings me to the point of this message: we are going to be bringing on some important users (relative to us, of course) in the near future, and Twitter represents one of the three important platforms available on our app. Because of that, we would like to pre-emptively avoid any possibility of tripping the spam filter.

Is there any way to:
(1) discuss the general anatomy of tweets produced in our app as they might relate to the spam filter?
(2) discuss proper testing strategy? (I’m almost positive that the previous time we tripped the spam filter was entirely because of our testing).
(3) Undergo a review process to make the spam filter more “reluctant” to catch our app?

If not, I think a separate app review process would be a good way to start being able to identify apps that contribute positively to the twitter ecosystem. It’d be pretty unfortunate if an app were to get permanently suspended just because part of the app is posting a video with your tweet.


#2

Thanks for checking!

I general for a question like this around policy there’s an option on the Policy Support forms to contact us.

We try to treat apps equally, so this can be a difficult area for us to navigate when we’re also trying to keep everyone happy!

One thing we obviously can’t easily do is to publish the way the antispam system works in detail (although we have a blog post and a talk from Flight about how it works in general terms). We don’t want to enable people to “game the system”… also, the system itself has ML which enables it to react to different situations.

Definitely take onboard your suggestion about potential improvements to how we could perform app review, thanks for the ideas.


#3

Andy, thanks for taking the review suggestion onboard. If you need someone with a Twitter app to try out a new policy/spam review system, don’t hesitate to reach out!

Ultimately, I understand that you can’t talk at length about how exactly the antispam system itself and that’s why I suggested the review option as an alternative: from my point of view, it means people who put the work into subject their product to review to proactively ensure it fits Twitter guidelines can reap the rewards. Even though you continue to apply the same policy and rules to every app, just the presence of a review confirming the app conformed to the Twitter policy X weeks ago might be enough to sway a 50/50 judgment call, and that could prove invaluable to the developers.

Thanks for your time.