What does Twitter's new chatbot mean for social customer service?

Twitter has become the latest social network to offer up chatbot capabilities, with Tuesday’s announcement that it has launched automated welcome messages and quick replies into its Direct Message function.

Among the brands that are already up and running with their own versions of the bots are Pizza Hut, Airbnb and Spotify.

Twitter’s product manager, Ian Cairns, stated that he expected the service to help brands reduce the amount of time it took to resolve a query or assist a customer with a transaction, and that the recent years of social customer service evolution on the platform highlighted the need to move towards more automated service and assistance.

Speaking to AdWeek’s Social Times, Cairns also stated that some brands had found that as much as 80% of inbound customer service requests were now being made through the social network, and that when users tweet at businesses and receive responses, they are willing to spend 3-20% more on average-priced items from those businesses in the future.

Examples of how the latest update is set to be used include helping to order a pizza via the @pizzahut DM service, or tracking a parcel via the @Furni DM:

Get faster and easier help from businesses. Try it now with @EvernoteHelps and @PizzaHut. #CarpeDM pic.twitter.com/T8vHnLESEM

— Twitter Advertising (@TwitterAds) November 1, 2016     

A number of tech companies are already signed up to help brands develop their bots, including Sprinklr and Conversocial, which has been charged with developing Tesco’s new DM tool.

“Our team currently spend 14% of DM replies asking for additional customer information,” says Mike Johnson, improvement manager for product at Tesco. “This new functionality allows us to seamlessly gather more customer context prior to responding, so we can serve them in the most helpful, relevant, and efficient way.”

A beta version is currently in testing, but Conversocial has given a sneak peak of how the service will work:

Conversational commerce

It’s no secret that Twitter is trying to improve the value proposition it offers brands. The social network reported Q3 losses of $103m last week, and has recently closed down its Vine video service – axing hundreds of staff – with the mandate of focusing its attention in areas that prove more attractive to brands, including an improvement of its customer service function.

However, its latest update comes at a time when many businesses are increasingly switching their focus to message apps and more ‘conversational commerce’, with Twitter’s nearest rival, Facebook, very much leading the charge by opening up its Messenger API earlier in the year, for businesses to develop their own chatbots to help serve this need. Whilst Twitter’s update is arguably more primitive, it’s very much a step towards offering a similar foundation for brands to experiment.  

“Most brands have now realised that the way they are interacting with their customers needs to change,” says Simon Robic, social product manager for iAdvize, a conversational commerce platform. “And they first need to address the ubiquity challenge by being available everywhere and at any time for their customers. Social networks are familiar ground for online customers and they expect brands to respond to their questions in real-time on the same channel.

“Twitter is an excellent channel for fostering the interaction. With these new features, brands can now become more efficient and make sure the customer journey is as simple as possible. All social platforms are increasingly focusing on strengthening their ability to provide a better channel for brands to support their customer engagement, one example being Facebook Messenger. This reinforces the need for brands to be available to their customers anywhere, at any time.”

Simon Preece, a social media expert and founder of slp consulting, also foresees the commerce side of Twitter’s update to become as attractive to brands as the customer service side, in the near future:

“Ultimately it’s about getting more people to spend more time within the social networks. Once you get someone using a social network as way to make a purchase/follow-up with customer care etc, you give them a cast-iron reason to come back.

“If I’ve taken the effort to setup an account via Twitter to order from Pizza Hut I’ve got some vested interest to keep using that route. If I have that relationship with 5, 10, 15 brands, then suddenly that social network becomes THE place I do my shopping.”

This is especially prominent on mobile devices, with the combined statistics of 4 out of 5 American consumers now shopping on their smartphones and the fact that messaging apps have now surpassed social networks, in terms of how much time we dedicate to both.

Source: Business Insider

Getting started

So how should interested brands approach Twitter’s latest function, whilst still in its nascent stages? “This is a great opportunity for brands to optimise their customer qualifying process,” says Robic. “You need to define useful scenarios for their customers so that the conversation is helpful for the customer and time-efficient. Deciding what types of questions the customers are most likely to ask in the qualifying phase will help brands offer the best customer support experience.”

Preece adds that there are three further points to consider:

1. You must work carefully when thinking about those welcome message options to ensure they offer options that maximise efficiency. So for example over Black Friday, how might the options change to match what people are most likely to enquire about at that specific time?

2. You need to avoid automation going too far… people will accept automation if it’s efficient, but if they need some human empathy and get an automated response it’s not going to go well.

3. There are many elements that must come together to make this sing, such as having a consistent tone of voice and hooking it up with your CRM database, to ensure customers get a good experience (i.e. so they don’t have to start providing information the business should already know).

It’s also important to set realistic expectations. Some crude tests of the early adopters have proved there are some inconsistencies that still need to be ironed out. Notably the response to our attempt to order pizza via the Pizza Hut chatbot:

Increasingly impatient customers are not often willing to wait that long, especially not when pizza is involved.