Messaging + Workflow: The reality of chatbots for assisting customer experience

A range of topics including No UI, chatbots or straight up AI based interface are all over the digital press. Can organisations expect to create and embed our very own AI assistant to replace our business model? What does the next generation of apps actually look like over the coming years. We investigate...

Typy-typy: Natural language interfaces

It’s not difficult to see where the growth of the natural language interface is coming from (i.e. typing or voice to control an app) when looking at the rise of messaging platform with billions of people logging in each month. That’s more people connected on real time than on the internet in 2010.

The leaders of messaging; Gmail, Facebook, Whatsapp and the myriad of other instant chat platforms all share the same defensible attributes - people are used to typing into them and they have a huge amount of information on their users.

But what does this mean for all of us creating digital products where we don’t have a billion users? How can we leverage this platform without trying to create our very own AI technology? Be prepared for this will certainly be a new channel of acquisition.

Lets wind back a bit… those that remember the day of Eliza can recall the seemingly magically replies and had a slightly inquisitive personality - it’s always nice when someone asks right?

But we're not looking for a conversation when we're using an online service.

First, self-directed rather than message-only interfaces

Looking at the market place with Cortana, Google Now, Alexa and Siri the approaches are quite different. Cortana and Siri are really search interfaces rather than interactive but how do Google Now and Alexa interact with us as users?

Imagine trying to sell a pair of shoes through describing them. Long winded indeed and not particularly effective. As a result we predict less continuous chat and more personalised interfaces. Take Google Now as a good example of the leading assistant:

Or let’s look at the evolution of GoButler from when they started as an SMS app:

But check out how the experience changes in the app:

So what we’re seeing here is a presentation layer being exposed into an external service. We’re seeing the creation of micro user-interfaces to appear within closed-loop ecosystems. We're seeing better and more deeply embedded apps.

Should we just be building into these closed platforms?

Clearly the big tech players are throwing many $100 million of dollars into their assistant apps. They’ve got all the history of flight bookings, friends contacts and network effects.

It does seem though that there is an opportunity to turbo what we see as “Notifications” and augmenting them with interactive UI. For example adding a Google Now card to represent your service or offering to resume a transaction.

Google offers a hierarchy for their notifications. My best guess is that we’ll start to see the “opportunistic” notifications evolve into assisted conversations via Google Now.

Discoverability with conversational UI

What’s more the messaging platforms are likely to start to want to invoke UI components iFrames in messaging systems?

We’ll hopefully design guidelines will be reinforced and the messaging platforms will provide a way for users to self direct by mixing screens from various apps woven into their narrative.

Discoverability in these UIs change completely and we’re likely to see new ways to design interfaces, perhaps new grid systems and certainly new interactions. has an interesting article on it and does as this chap on Medium (I found this link when looking for platforms embedding applicaitons in messaging!).

Preparation Tip 1) Agility in your systems team and microservices design

Is my organisation ready to provide drive these new interfaces? Fortunately if we look at the IT changes appearing over the last few years we can see that the rise of CRMs, then online services and most recently cloud and mobile has opened up the enterprise IT world to the internet.

As a result particularly of mobile the enterprise is more open and is representing itself as an API to the world. If your organisation does not have an API representing your core service offering then we’d recommend engaging with a partner to help you accelerate your roadmap and create one.

This need not mean rebuilding your enterprise but working with technologies such as Jitterbit or Mulesoft to sit in-front of your existing infrastructure.

Preparation Tip 2) Collect behavioural data, study it and discover scenarios

No wonder that the social standing of a Data Scientist has overtaken a stock trader. It certainly pays like it too. The fact is data science is challenging however taking just the first step into looking at trends on your existing platforms or interviews with your customers will almost certainly come up with some scenarios to optimise.

See our previous article about “measuring metrics” where we look at a number of tools such as Kissmetrics, MixPanel, Google Tag Manager and others to track behaviour of the users.

Recommendations can be made to new users based only on the data that you have collected. A quick tip? Collect and tag every event you can!

From a set of scenarios (let’s call them user stories, see our couple of articles about agile) then let’s estimate the measureable impact we have on the user base.

Preparation Tip 3) Pick a single core interface for redesign

What is the most critical and highly used aspect of your service? Considering that part and run through some scenarios where your users may need reminders or might come to the end of a workflow.

Here are some questions we asked ourselves in a workshop:

  • Can we find a flow which can be embedded in a conversation? - finding a place where you usually refer customers to your website from a conversation. Can this service be embedded in a conversation on one of the messaging platforms?

  • Redesign the retargeting experience - if a customer aborts how can they resume the flow by invoking a flow within a messaging service?

  • Publishing updates to your service - is there a critical part of you service which changes over time. Can you identify a group of people who might care more about this?

  • Test out messaging within your app - This may simply manifest itself as a messaging system within your app initially. From there it may evolve into recommending products and service or providing opportunities for users to confirm suggested actions.

One thing is for sure the messaging platforms will provide a great opportunity to allow users to access your service without the need to switch into it.

What next?

Well Byng are running a Hackathon on the 20th and 21st May to work through examples for our clients. We’re looking at how we can plan scenarios to test with and some of our other clients.

We’re looking at various recommendation technologies to sit alongside analytics packages. Watch this space!