Voice. Six Lessons Beyond the Hype – Part 2

In this two-part series Erik Rave (Technology Director)and Jeroen Thissen (Creative Director) from Creative Digital Agency CODE D’AZUR look back on three years of developing voice applications for the likes of KLM and LeasePlan. In this second part, they will touch on their final three learnings: Intent over context; The people and the brand; and Keep it simple.

Intent over context

We have arrived at our fourth out of six lessons about developing for voice. This is an important one, which also reflects on the previous learning about your architecture. Why is it important to put ‘intent’ over ‘context’?

‘Context’ is the structure of the conversation. At CODE D’AZUR we start the design process of each voice application with a Conversation Tree (much like a ‘choose your own adventure game’) in which we design the expected conversation a user will have with our service. A ‘happy flow’ of our dialogue. We quickly learned the obvious:

People don’t follow trees.


Image Source: Chatbots Life

When people talk to a chatbot they will curate their answer. You ask: ‘When would you like to depart?’, people think for a bit and answer: ‘next week Tuesday’. Alternatively, when using a voice service, people speak before they think. You ask: ‘When would you like to depart?’, people answer: ‘Uuuuhm next week Wednesday 18. Oh no 17. Tuesday that is, please’. They will use different wording every time – even when saying the same thing. They will say random things at random moments, move back to things they mentioned before or jump ahead to parts of the information retrieval process you didn’t expect. People are people. And so, your conversation can’t be linear. It’s not a webform.

That’s why it’s important to put intent over context. In ‘Intent Based Conversational Design’ we focus on what the user wants to get out of the service we provide, first. Then we make sure the conversation is flexible enough to cater to the intent of the user at any moment. The result? We can jump backwards and forwards and out of the conversation at any time and we still know what’s going on and how to proceed.

By default, DialogFlow is set up to keep all intents open in any part of the conversation, but it can be tempting to limit intents per interaction to make sure the user walks through the conversation in a linear fashion, and to save back end development time. This is where you start to make choices based on your expected behavior of the user. In practice, you’ll see that it is very hard to predict behaviour and, UX wise, it is much smarter to keep all options open for the user and have your system figure out where we are in the conversation.

This is a lot more work though. DialogFlow does not allow for any exchange of data between these intents. It lacks the functionality to make sure that one part of the conversation still knows what has been said in other parts of the conversation we built ourselves. For now, at least. This is another good reason not to rely fully on DialogFlow for the design of your action.

The people and the brand

The past couple editions we’ve talked mainly about the need to stay flexible and control your own environment. We’ve mainly looked at HOW to structure your service. In this part, we want to start looking at WHAT your service might entail. In this sense there are two stakeholders that define a successful service: the people (primary) and the brand (secondary).

People fit: Why do people need this?
This really goes for anything a brand does. You’ve got to make sure you add real value for people. People will have to want to use your service. Other than a TV ad or a banner you can’t buy your way in. In that sense, voice is really transparent and clear. Build a good service and people will come back. Build a bad service and people won’t.

As mentioned before, you can only really validate your service after your first release. But this doesn’t mean you can’t do research up-front to improve your chances. At CODE D’AZUR we combine contextual interviews, desk research and video diaries to explore service opportunities and determine pains (or gains) in the lives of people we can leverage upon. After ideation and idea selection we quickly build a simple prototype to do a quick validation of the idea. This is how you vastly improve your chances in creating a relevant and useful service. A good place to start looking is the customer journey: what steps do your customers make throughout the entire process around using your product?

Brand fit: Why is the brand doing this?


Image Source: Shorty Awards

Once you feel you have an idea that can work for people, it is obviously also imperative that it works for your brand. It needs to have ‘brand fit’. If there is no connection between the solution you are offering and the vision of your brand people will simply not understand what you are doing. Even if the solution itself is great, you are still diluting the brand value. It would make very little sense for Tesla to start a cooking manual or for Coca-Cola to start a train departure time table.

Keep in mind that, ideally, the service also is unique to your brand. Look for ways to use your own data (that’s not available anywhere else, also not for Google) to make a relevant service. In KLM’s voice services we use data from the Schiphol API, Amsterdam’s local airport, to advise people on when to leave their homes for a flight. This makes the service unique for KLM.

This unique offering can be anything from special price offers you have to services in which you use other specific customer data. Because if the data is already public you’re not offering anything new or unique. If you’re a mayonnaise brand, you could create an action to give me information about your product, or even recipes that involve mayonnaise. But I can already ask Google this information. And it’s probably better and faster. So why would I take the extra step and go to your brand? There’s no value for me as a user. That’s why Disney does not just give Star Wars trivia, but turns it into a challenge. Or why Netflix does not just give you information about their show Stranger Things, but turns it into a game.

Keep it simple

A.I. is not smart enough yet to really set up a good conversational interface. Although voice recognition has been almost 100% accurate for some time now, the way humans converse is still hard to automate. Google and Amazon can understand WHAT we say, but struggle to understand WHY we say it. We know this is essential to have a real conversation.

In 2017 we launched Pack Assistant for KLM. We tried very hard to make it fully conversational. We focussed on interpreting many answers and also built in options to make sure people did not leave the conversation and could jump in and out of different intents. The result was technically advanced and very forward thinking piece of conversational voice technology that even got featured at the Google Play House at SXSW 2018 as a case. However, what we see now is that it was too complicated and elaborate for people to really use. That’s why we are now working on simplifying it drastically, to make it far more accessible.

Reality is that people still need to learn what they can expect from a brand on voice. At this moment people use their Google Home mainly to turn on the lights, add things to a shopping list and set timers, with short commands to arrange one thing. There is very little actual conversing going on. And people don’t expect to have a conversation yet. They expect to give short commands and get brief short answers.

In development things can also get complicated quickly. Especially once you start to bring in multiple API’s and run on multiple platforms in multiple languages. The work involved in testing, training and maintaining the services can explode.

These are all arguments to start small, keep it simple and move from there. You don’t really know how the platform will develop in the near future and you can only evaluate how to improve once you’ve got user feedback. The beauty of the platform is that feedback is very direct. If you want to know what people think of your website or app you really have to ask them to tell you. Within the Google Home environment people are already tell you what they expect, whether you like it or not. This can help you to prioritize your design iterations and give focus to improve and expand your service portfolio.

For us, the main goal that remains is to keep learning and to transform ideas into services as soon as possible. It’s the only way to move forward. Ship quickly, learn quickly and take it from there.



Erik Rave, Technology Director, CODE D’AZUR

Erik Rave is Technical Director at creative digital agency CODE D’AZUR. In his position, he oversees the development and technology department and ensures the technical realization of show-stopping creative concepts for the likes of KLM, Red Bull, ABN AMRO and LeasePlan.

With a passion for tinkering, Erik has spent close to two decades honing his skills and knowledge all things technical, from coding web platforms and building installations, to developing augmented reality games, artificial intelligence systems and voice technology solutions. When he’s not knee-deep in code or arms full of tech-parts, Erik enjoys hip-hop and techno tunes to fuel his inspiration.

Headshot-Jeroen Thissen-LR

Jeroen Thissen, Creative Director, CODE D’AZUR

Jeroen Thissen is Creative Director at CODE D’AZUR. He creates digital campaigns, services and products that people love for brands such as KLM Royal Dutch Airlines, LeasePlan, Red Bull and ABN AMRO Bank. Recently he’s worked on creating AI ecosystems, voice applications as well as AR games for various brands. His projects have been recognized by The Webby Awards, The Lovie Awards and Cannes Lions.

Voice. Six Lessons Beyond the Hype – Part 1

Google Assistant Taps into a Much Larger Number of Voice Search Sources Than Alexa or Siri for Questions About Brands (Plus Webinar Tomorrow)