Find Your Voice – how to develop a killer voice application
Chris Drake
23 September 2019

Have you ever gritted your teeth as you’ve had to pick up the phone with greasy hands when you’re in the middle of cooking, missed a bus stop because you’ve been checking the sports results or dangerously taken your eyes off the road when driving to check “what was that ping!” on your phone? Wouldn’t it better if you could live your lives more heads up than constantly downwards and into a screen. This is the promise of voice – making people superhuman by extending their most natural of senses and making their lives easier. If brands can tap into that they will reap the rewards bestowed on all productive brands – increased brand preference and long term growth.

Google assistant only launched in Sweden in August 2018 and this development of more intelligent assistance has been a key factor in Sweden’s desire to start talking with tech. Indeed 28% are now using voice at least once a month (we are fast catching the more talkative nations of Spain 54% and Italy 58% respectively). Source: OMD European wide research piece – the Retail Revolution AI perceptions and adoption.

So how do you get it right? At OMD we are using our understanding of customers and their needs, desires and behaviors to extend our service offering from paid, earned and owned communications, through to service design – including voice. Our latest project being the design and development of the McDelivery Google Home application – delivering McDonald’s favorites straight to your door as quick as you can say ‘Big Mac’ and we wanted to share our learnings and considerations when building a voice application that (we hope) will help brands deliver on the promise of voice.

"This is the promise of voice – making people superhuman by extending their most natural of senses and making their lives easier."
  1. Right use case

To find the right use case, you first need to be clear on your goals, so which audience/s are you going after and how do you need to engage with them to drive growth….. so do you need to Prime (drive brand equity and pre-disposition), prompt (deliver reasoning), purchase (convert all sales opps) or post purchase (up-sell and reward loyalty).  The McDelivery Google Home app for example is primarily going after a new home delivery audience smoothing purchase by empowering and supporting the customer, but there are great use cases for all the above. The most important thing is of course to add value, whether that be experiential or functional.


  1. Keep it simple

Simple brands, experiences and ideas tend to always outperform more complicated ones (see Siegel and Gate). So if you’ve identified a strong enough use case then don’t over-complicate it with unnecessary additions. We tend to say its one solution per app. For example SJ have focused (at least initially) on providing real-time travel advice rather than extending into complicated ticket booking process. For McDonald’s there were a number of ideas on the table (experiential and functional) but we focused on McDelivery and “McDonald’s food delivered straight to your door”. Make this clear for first time users in the welcome statement and prompt what they can do. Inside the application, its also crucial to keep the user experience as simple as possible. Voice is a relatively a new technology where the user has no visual cues and is essentially walking into a black room. You need to give the user clear guard-rails and limit or break-out choices into manageable pieces. Voice is linear so users have to 1. Wait for options to be read 2. Remember what the options are. This was a major part of the McDelivery app guided menu design.


  1. Persona

Most strong brands have a very clear visual brand identity, you can probably visualize the logo of at least a hundred brands, but there are also other elements – the typeface, colors,  a full library of graphic elements, images and even people. A strong sonic branding identity is just as important when developing voice applications, but most companies haven’t thought about what their brand sounds like (voices, signature sound/jingle, brand anthem, functional sounds, soundscapes, playlists). Volvo for instance have given this considerable consideration and have a ‘Scandinavian sound’ – a combination of simple, peaceful and functional that extends to how the sounds of the cars themselves. In the case of the McDelivery application, McDonald’s have a strong base to build upon having been very active in audio communications – with one of the most recognizable signature sounds in the world. But perhaps the biggest consideration with regards to sonic branding is the persona or voice itself, after all voice applications are judged as if they are real people. With this in mind a recorded voice for McDelivery (versus the Google standard voice) was a no-brainer. But whose voice? We had determined the voice should be varm, trygg och smart med glimten in ögan. We found that via an internal competition at McDonald’s to find the best voice from their own drive-thru operation (the service synergies between drive thru and voice are obvious). This co-creation served to not only have a pre-packaged persona, but also ensured we stayed on-brand and had a useful piece of employer branding.


  1. The long tail

People expect voice applications to behave like real people, and their success will be judged as such. They can ask any number of questions that brands need to anticipate and have answers for – from long-tail service questions (opening times, job applications etc), to “personal” questions and even inaudible noises and silence. The first thing is to minimize the scope for asking such questions by giving clear prompts and choices as to what they can do, but this is not sufficient. For the McDelivery app we linked to FAQ pages on the website to capture many of these queries but also looked at related google search queries and brainstormed questions in large groups or with drive-thru employees. As a result the McDelivery persona, Linnea, can answer questions about her age, favorite food and even what they call a Big Mac in France. Also remember users can’t can skip content in the audio stream so keep your answers crisp.


  1. Test, test and test

Testing is a crucial part of any service development – you will want to find the bugs rather than your customers to provide the best possible user experience. The importance is even greater in voice interfaces which allow an endless spread of interactions, unlike visual interfaces which tend to have more fixed choices. A lot of the programming and continuous testing can be automated by the development team, but end-to-end testing needs real users stressing the application using different types of voice, environments, phrases and unexpected inputs (and not just stakeholders who intuitively know the company language and processes but real customers).  Google also has a comprehensive certification testing program that includes secret shoppers who stress the back-end of the system for transactional apps like McDelivery, so be sure to allow ample time for that.


  1. Keep learning

Finally, the journey doesn’t end when you launch. There will always be new platforms e.g. Amazon will soon launch in Sweden, and new opportunities to improve your service e.g. apply new AI capability to your application to make it more personalized and useful, so keep learning and adjusting.  Your customers have busy lives, if you can empower their voice and help make them superhuman they will not just buy your products but become lifelong advocates for your brand.

Stay in the know

Sign up
Successfully subscribed! Thank you!