A cross-platform voice experience for Amazon Alexa and Google Assistant to bring the world of Channel 4’s Humans to life.
In 2018 Channel 4 approached us to craft a unique voice experience promoting the upcoming season 3 of their critically acclaimed tv show Humans. Having won Creative Baftas for the promotional campaigns of the show’s first two series, Channel 4 sought to raise the bar and become the first UK broadcaster to promote a tv show through Amazon Alexa and Google Assistant.
Working together with Channel 4’s in-house creative agency 4Creative, we developed a sophisticated voice experience based on the concept of a Turing test to find out whether the show’s fans were human or ‘synths’ trying to hide among the human population.
The experience was launched in April 2018, accompanied by cross-channel promotion by Channel 4 on-air and online, as well as a companion website NSDU.co.uk. The experience received over 50,000 users in its first week, and was rated highly on both Amazon Alexa and Google Assistant platforms.
At the end of the second season of the show, artificial humans called ‘synths’ became conscious and nearly indistinguishable from the human population. By using voice assistants to administer a Turing test-style quiz to the show’s fans, Channel 4 sought to continue their award-winning strategy of bringing the ‘parallel present’ in which the show is set to our real world.
Working with 4Creative and Glasgow-based copywriting agency Blackad, we crafted four sets of unique questions covering a variety of question types, from True or False to open-ended questions. The question set a user receives each time they interact with the experience is randomised, increasing replayability and stickiness. We also crafted fluid and intuitive failure paths, meaning that if Alexa or Google Assistant didn’t comprehend a user’s response, they would not get stuck and could progress through the test with minimal frustration.
Please think carefully and answer all questions truthfully.Your answers will not be stored once the test is complete.Let's begin. What is your first name?
So, Dave. Here’s your first question: You have a burning secret. You can only share it with one person.Would you choose a friend, relative or a stranger?
You would trust a stranger Dave? Interesting.Here's the next question. Yes or no…Are all humans replaceable?
In your opinion you can't be replaced?True or false…Although we didn’t live as long, life was better 500 years ago.
Your answer has been noted Dave. Please relax.Think before you speak to avoid your case being escalated.Yes or no...Can you admit when you've made a mistake?
Your answer has been noted Dave.
The experience uses a mix of short audio clips and bespoke recorded audio in addition to the native assistant voice. This blend ensures that the experience stands out from existing offerings and exceed users’ expectations.
Part way through each experience, the native assistant determines that it cannot confirm the user is human, and escalates the conversation. To achieve this, 4Creative recorded custom audio with a voice actor to simulate a phone call to a ‘human colleague’ who took the user through the remaining questions. This quirk quickly became a fan favourite, catching users by surprise and enhancing the ‘real world’ feel of the experience. Each experience concludes with a radio-style advert promoting season 3 of Humans, and directing fans to the experience’s companion site NSDU.co.uk for more information and to share their results.
To reduce development time and ongoing maintenance we used a cross-platform approach to develop the Human Test. Hosting the experience’s content on Azure, we used Azure Functions as a facade between a shared backend codebase and the Alexa and Google Assistant platforms.
Using our own ‘Story Mode’ method of voice development developed in-house for use in media and entertainment voice experiences, we were able to control how a user moved through each experience, without jumping from question set to question set. By understanding where a user was in the test, we could ensure that the assistants were serving users the right questions and responses in sequence, resulting in a smooth, consistent, and engaging experience for users.