5 things you need to know about Facebook’s chatbots and the future of service

14th Apr 2016

This week’s Facebook F8 conference offered the latest evidence that chatbots will play a significant part in the future of customer service. 

Following the social network’s announcement that it will be giving businesses access to its API in order to develop chatbots for the Facebook Messenger app, many people have questioned what purpose these bots are set to serve, how they might aid the relationship between business and consumer, and why customer service professionals are being tipped as most likely to benefit.

Here’s a few answers:

What are they?

Wikipedia’s explanation of bots is arguably the most concise, defining them as “a computer program which conducts a conversation via auditory or textual methods”.

The first bot was created in 1966 (named Eliza), and bots have subsequently gone onto fulfil a number of guises, both positive and negative. They’re a key component in the customer service chain given the rising popularity of their use within Live Chat and virtual assistant software, and some are regularly passing the Turing Test, suggesting their intelligence has reached a level humans can now interact with.

Facebook’s overall goal in opening its API is to encourage businesses to develop chat bots that allow its 900m+ Messenger users to simply ask a question or raise an issue that can be answered or solved within the app itself, thus negating a need to ever exit the app to perform other tasks. It’s seen by some as a danger to the mobile app industry, but the reality is it’s not the first chat app to do this (WeChat users are already au fait with the benefits), just arguably the biggest.    

“Automated chat has been around for at least a decade or more and bots aren’t necessarily new as a technology concept, however the bots we know today are much more sophisticated than they were just a couple of years ago,” says Evan Wray, co-founder and VP of Swyft Media.

“With the use of mobile messaging applications continuing to surge, there’s a huge opportunity for brands to engage with consumers in new ways, and in the place they spend the most time. Bots allow brands to become a part of the consumer conversation – rather than an interruption, and reach customers in a way that is one-to-one, highly scalable and with a personalised approach.”

Can they eradicate human involvement in an interaction?

All our dystopian fears are set to be realised with the rise of artificial intelligence, and chatbots are simply the latest iteration. Or not. Depending on who you talk to, chatbots are either a descendant of human life as we know it, or a way of automating certain processes for the benefit of both ends of a transactions. The latter seems the more likely, in the short-term at least.   

"Given that businesses now face a huge challenge in handling large volumes of customer conversations, bots allow companies to automate a certain amount of low value tasks, accelerating their customer care services significantly,” says Julien Hervouët, CEO of iAdvize.

“In using these robots in this way, firms will instead be able to focus on customer interactions and conversations which have a high value and actually affect their bottom lines. Most importantly, these bots won't replace humans, especially now that customer conversations have become so important, especially evidenced by the recent controversy surrounding the abusive AI bot, Tay, which had to be turned off within 16 hours. Instead, bots and human agents are complementary when it comes to offering the best customer experience possible."

AI bot, Tay - got abusive, pretty quickly

From a customer service perspective, it seems a dangerous proposition to suggest a bot might ever be able to negate the need for human involvement in a support query.

“It’s still very early days, but any organisation thinking of deploying chatbots should choose very carefully how they do so,” says Maurice Vink, principal consultant, PeopleTECH. “Issues with natural language have already caused a number of businesses to abandon natural language call routing and deflection front-ends in their contact centres because of negative customer feedback. Chat bot developers should target the less intimidating use cases first before they get carried away.”

According to Jay Sorrels, senior social media strategist at Ogilvy, this will mean companies honing in on developing bots to answer “functional questions like ‘Is my flight delayed?’ ‘Do you have that item in stock in a specific store?’ ‘Can you make the wi-fi work in my train carriage?’ and ‘Is there any parking available?’ 

“A good chat bot means customers can get what they need from a brand or business faster and more easily than before,” Sorrels adds.

What’s the benefit beyond customer service?

Beyond customer service is exactly where Mark Zuckerberg is looking, with retail expected to benefit most from Facebook’s Messenger API.  “There is the potential to add a more transactional element to the mix,” says Stephen Morgan, co-Founder of digital transformation business Squiz.

“For instance, if someone uses Messenger to ask William Hill for the odds on tonight’s big game, the chatbots could provide an answer and offer to place a bet within the app. Equally, if a person uses Messenger to ask Topshop if they have a certain pair of shoes in stock at a local store, the app would be able to say yes or no – and could offer the ability to reserve a pair of shoes for collection or order them for delivery.”

But isn’t that dangerous?

Customers have to opt-in to engage with any company’s Messenger bot and, for now, companies aren’t supposed to use it simply as an advertising tool. However, despite this, much like with the advent of location-based marketing, the value proposition in the early iterations could determine whether consumers get on board with the Facebook bots or not.

“The scale of this consumer shift is huge so the main challenge for Facebook will be around building and protecting consumer trust,” says Hannah Giles, head of marketing at Zensend.

“Facebook has done a great job of doing this within its social network, but it will need to ensure that users of its messaging products are also given total control over how and when any brand engagement takes place.”

Maurice Vink adds that beyond having control, the quality of early Messenger chatbots will ultimately determine whether people build up any level of trust in them.  

“Zuckerberg stated that with AI and natural language processing combined with human help, we will be able to talk to chatbots just as we do with friends. However, it starts to get more problematic where there is a lot of duplication and variety of spelling, such as in names of places, hotels, restaurants and even flowers, to use some of Zuckerberg’s examples.

“Most of us struggle with the spellings and pronunciations of foreign words and place names, so disambiguation will be very difficult. How many times will we put up with “Did you say..?” and “Did you mean..?” before we give up in disgust, leaving some poor contact centre agent to try to calm down a frustrated customer?”

Indeed, the Guardian’s recent tests on certain chatbots reveals that they have some way to go to be able to cope with the rigmarole of colloquialism, while the BBC found one bot hell-bent on declaring that the population of Leeds was seven billion, suggesting hiccups are inevitable.

Which companies are already doing them?

As with any seismic tech announcement, a plethora of vendors have announced their integration with Facebook Messenger in the aftermath of the chat bot API opening.

Salesforce has announced Salesforce for Messenger, which will allow retailers to embed a Messenger plugin on the checkout workflow on its website so a customer can ask any final questions before making a purchase.

And Spring, the digital shopping platform available on mobile and web, is already using Zendesk to integrate human chat into a Messenger bot experience that allows users to shop, discover or browse Spring’s collections. The general consensus appears to be - expect many more to follow, and including the feasibility of developing your own.

Spring's chat bot has already sprung

“Chatbots take a small part of AI’s wider potential and apply it only to two-way text based communication,” adds Jay Sorrel. 

“This technology has lurked around the ‘Live Chat’ feature already present on so many sites, often as the first point of contact before a human takes over, but it’s clear they are now taking this to a new level by adding much more back-end intelligence and more of a front end social face.”

Replies (3)

Please login or register to join the discussion.

By EJohn Morris
18th Apr 2016 09:08

Thanks for sharing Chris, great article and whatever we want, prefer or complain about it will happen. I shared this on LinkedIn and Twitter with the caption "when you don't know you're interacting with one you won't care that you are, that's when their ready." Which i believe will be the case and with many call centres using scripts who will be able to tell anyway?

Thanks (1)
Replying to EJohn Morris:
Chris Ward
By Chris Ward
19th Apr 2016 12:07

John - your point raises a very interesting ethical debate: at what point are customer service bots sophisticated enough that brands need to confirm up-front that they're a bot? If bots are passing the Turing test this seems like the most ethical approach, however I imagine consumers would be sceptical if this were to be the case.

Thanks (0)
By JasonAtUBot
23rd Aug 2016 19:08

Appreciate this article. IMO, the really revolutionary bots will be so much better than this - and we'll probably stop referring to these kinds of chat bots as bots at all. This is like the Web 1.0 version of bots. Every article about them is mostly about hiccups and issues! Honestly, at UBot Studio we've been wondering what are bots going to progress to for a long time. Customers of ours have been building much more complicated "bots" than this for 6+ years now, and most don't converse! Hopefully soon we'll move away from the idea that a bot must be queryable and have this limited sort of possibility of responses, and move towards bots serving functions for people instead of functioning based on the limitations of our 'bot-building' imaginations.


Thanks (1)