The Key Preventive Measures any Twitter Scheduling Tool Has To Implement?

[TimotheeFranc]

Dear Community,

First of all I’d like to thanks people like @hector_borras from the #TapIntoTwitter community who came to suggest help in this thread.

• Here’s a brief overview of our App.

Kapsuli is an App that allows users to schedule rich tweets at key moments of their Twitch stream.

We believe Twitter is an awesome place to drive traffic when you’re live and we try to make it easy for live performers to communicate while they’re streaming (with the Kapsuli App you can schedule tweets when your stream starts but also when your stream reaches a certain amount of viewers, when you change games, after a certain amount of time, etc).

For the time being we are synced with Twitch but we can’t wait to be able to plug our tool with the Periscope API.
Imagine you plan to livestream an important social demonstration in the afternoon, you’d be able to prepare the best tweets related to the topic all along your Periscope, to drive traffic in real-time.

Here’s a quick video of our App onboarding here.

We are on the verge of going in public beta and at this very moment we have 100s of active users who give very positive feedback about the app.

• Yet, as our API access has been suspended two times in a row in the last 7 days, we decided to suspend our app as long as we don’t implement every measure needed to comply with 100% of the Twitter Rules.

This post sincerely aims with no animosity to move the Twitter-Kapsuli relationships from “police” vs “spammers” to co-creation

• We’ve been flagged by the system because of “large numbers of duplicate replies or mentions and large numbers of unsolicited replies or mentions”, considered as spam.

It’s actually a common thing in the eSport to @-mention your partners, friends or sponsors when you go live, that’s why some of the tweets from the App might have contained to many @-mentions.

But even if this behaviour can be explained, we have to restrict it to comply with the rules.

I can guess some the leading scheduling tools on the market like Buffer, Hootsuite and so on have faced the same challenge so I hope the community will share feedback about this.

Here are the 3 big measures we plan to implement in the next few days

1. Make it impossible to have more than 2 @-mentions in each tweet.

2. Have a rate-limit regarding tweet frequency: max. 1 tweet / 10 minutes.

3. Block a tweet to be sent out if it’s exactly the same as one who has been sent in the last 30 minutes.

What do you think about these preventive measures?

Has anyone in there already solved this regulation issue?

PS: dear Andy Piper, we already share a funny story together (or, at least, a shameful one on my end).
2 years ago our Kapsuli Twitter Card was broken and as we had no news from you, I asked our early users to use a Click-To-Tweet to support us. But I accidentally forgot to remove the “andypiper” mention in this and it resulted in hundreds of notifications for you - many apologies again - forcing you to even block our tweets!

“Errare humanum est”, and we’ve been learning. I genuinely hope this conversation will enable us to start again with a clean sheet and make all of this a funny entrepreneurial anecdote.

Thanks everyone for your help and feedback about these important issues.

Timothée

PS’: @Valentin will be able to chat about every technical related feedback.

[MikeBarwick]

Is the content being generated user generated - or templated? If it’s the latter, you’re going to get suspended. Buffer, Hootsuite, and the lot don’t, because content is unique (otherwise a 403 is returned by API) and the content is user generated. I suspect the tweets are either fully or partially templated - or Twitter believes them to be. I run a content marketing calendar with scheduler, and we haven’t run into issues.

[andypiper]

We have recently posted a policy clarification on automation.

Policy clarification - automated replies and mentions

Like the hashtag, mentions are a crucial feature that first emerged from innovation by people on Twitter. Today, mentions and replies are a fundamental part of how people and businesses on Twitter interact, and we continue to see new and exciting uses of these features by developers — including to provide customer service, share news and reminders with others, and discover new topics and events. When misused, and especially at high volume, automated replies and mentions can quickly become spamm…

[TimotheeFranc]

Thanks so much @andypiper & @MikeBarwick for your inputs, really appreciate your support here.

Is the content being generated user generated - or templated? If it’s the latter, you’re going to get suspended. Buffer, Hootsuite, and the lot don’t, because content is unique (otherwise a 403 is returned by API) and the content is user generated. I suspect the tweets are either fully or partially templated - or Twitter believes them to be. I run a content marketing calendar with scheduler, and we haven’t run into issues.

Our content is being user generated.

I mean, each user can create their very own different sessions of saved tweets. The tweets we send for them are different for each user as they solely craft the messaging.

Like a saved session of 3 tweets for Monday, another one for Tuesday… or for each kind of livestream they do.

However, some users have only saved one session that work for each of their streams.
So the going-live tweet they do on Monday, Tuesday, etc., could be the very same.

1. Maximum 2 “@-mentions” per tweet.
From now on, only 2 “@” will be possible in each tweet.

Thanks Andy, I’ve been reading this post.
Do you think we should reduce this to only 1 @-mention per tweet?

2. Maximum 7 mentions in your last 10 sent tweets.
We’ll continuously scan your last 10 tweets sent from Kapsuli and calculate how many “@-mentions” there are in these 10 tweets.

Do you think we should reduce this ratio to 0.6 or 0.5?

Also @andypiper, maybe should we replace automatically a @ with a # if this @ has already been used in the last X tweets in the last N minutes? What should be X or N according to you?

We really don’t want to game with the system as you can see, we only want to be able to build a long-term product with you. I don’t expect exact numbers but only some estimations about X or N please

3. Maximum 1 tweet per interval of 10 minutes.
The New Kapsuli Policy Bot will prevent any tweet to be sent out if one has already been sent in the last 10 minutes.

Do you think we should prevent any tweet to be sent out if the exact same one has been sent in the last N minutes? Again, what should be N?

Can’t wait to read your answers!

We have to implement these rules asap as 100s of paying users are waiting for our comeback… it becomes critical for us.
As you can see, we only want to be good Twitter citizens.

Thank you so much in advance,
Tim

[TimotheeFranc]

Hi @andypiper, just to let you know we finished implementing these 2 measures:

• Max 1 @ per tweet, all other exceeding are automatically replaced with hashtags.

• Max 2 tweets within a 20-minute period.

Hope that suits you

We just relaunched the App so if you notice any weird behavior, please let us know beforehand so that we can ban the user.

Thanks again for your help and understanding,

Timothée

[andypiper]

Just to clarify that I’m personally unable to assist with this - and we’re also unable to provide the kind of active per-user notification you’re requesting. You should use the platform support form for policy issues, and respond to emails you may be sent by that team. Thanks.

[andypiper]