This request looks like it might be automated

[PelinNomad]

I am testing using the new Account API and working with Direct Messages . ( replying to Direct Messages)

I am getting these errors to Direct messages.

'“code”":226,"“message”":"“This request looks like it might be automated. To protect our users from spam and other malicious activity, we can’t complete this action right now. Please try again later”

It’s only natural that with a customer service solution that the same text is used often… and this error worries me. The fact that Twitter has Customer Cards ( to rate customer interaction) as a future feature made me pretty certain Twitter is open to having Direct Messages used as a customer service channel with large companies.

As an example of a repeated message , my solution is telling the customer
“You are now chatting to X” where X is the name of the agent.

Even in my low level testing I am getting rejected , saying it’s spam.

With customer service there is always lots of repeated statements , which make sense to be automated. Of course quick replies are always automated as well. ( I hope menus don’t get blocked!) I even provide information to customers while they are waiting to chat to someone. ( estimate of when they will be chatting to someone or opening hours etc)

Not sure what I am asking for here… but I guess I am looking for some reassurance that there is some way around getting these rejected messages for my solution. ( which is far from spam, in my case they are only ever responses to incoming requests from a customer)

Thanks

Rob

[bobber205]

I did alot of digging into this and this seems to be the only way around it.
By around it I mean this is what twitter seems to WANT you to do.

You should design your bot to function entirely around quick reply cards for user input.
I’ve been told this is their intention for bots and for the vast majority of user input to come in this way.

Since I changed our bot from accepting text input and instead started to use quick reply cards I have yet to see that spam message.

[PelinNomad]

Thanks… but I don’t think that will work for me.
In my case most of the Direct Messages are actually replies typed by a person. ( a Contact Centre agent). Regardless if the response is completed automated ( “Hi I am XXXX, how can I help you?”) or typed by the agent every time , the content will be the same and is likely to be seen as spam… Agents are actually told to be consistent in how they type things and have tools to select standard replies rather than type them all out every time… ( e.g " Is there anything else we can help with you today?"). I assume this message being sent in 500 Direct messages in 8 hours from the same Account could trigger that message as being spam?

Ideally I would offer as much automated messages as possible to reduce agent handling time. However regardless of automated or not automated, it’s still the same text being sent to a customer via the API.
Perhaps my testing was unusual as I was sending the messages faster to the same account than would normally happen and this was the cause. Fingers crossed…

Rob

[bobber205]

The message BACK to a user can be very similar or the same every time – it’s the user’s input to the bot that matters for the quick replies.