Burger King’s ad: the problems & potential with voice interaction
Yesterday, Burger King launched a TV ad which cleverly tries to invoke the Google Assistant, asking it to describe the ingredients to one of its burgers (you can watch it here).
Beyond just a funny story, it shows us some of the dangers inherent in all voice interactions, as anyone can trigger an interaction. It’s like leaving your computer unlocked at all times, to anyone, even if they’re not in the room.
The Google Assistant on a phone – along with previous incarnations of Google voice search – to an extent can have only trusted voices set to unlock. Google Home, and the other AI assistants and devices (like Echo/Alexa, Cortana and Siri), don’t have that functionality to the same degree.
More than that, though, it shows us the safeguards which will be put in place and the care we must act as marketers when working with voice. Google has already prevented the specific recording in the ad from invoking the Google Assistant, and it’s reasonable to assume they’ll act similarly if anyone else tries it. That is, if they don’t act far more harshly, potentially stripping an offending brand from appearing in (voice) search, in a similar manner they have done for historical infractions of their webmaster guidelines.
Finally, for a brief period of time, people vandalising the Wikipedia entry returned by Google were able to get Google Assistant to claim the whopper was made with some unpalatable ingredients. If I was watching this ad and had my Google Home trigger innocently, only to hear that “toenails” or “cyanide” were in the ingredients list, I wouldn’t rush out the door to buy myself one.
So after we’ve laughed at this encounter, what should we bear in mind? Three core questions:
How do I make voice interactions safe?
- Issue: Voice interactions are easy to trigger, potentially allowing for mischief.
- Solution: Make sure that any important customer interactions require a verification step – e.g. sending an email to the account holder with a link they must follow to confirm changes or purchases.
How should I use voice search in advertising?
- Issue: Google and other AI agent developers don’t like anyone abusing voice interaction.
- Solution: Be careful when writing any ads to ensure that they don’t trigger an action. This isn’t just limited to obvious things like saying “OK Google”, as the Echo sitting on my desk is often accidentally triggered by people in the office saying something like “Alexa”. Given that voice recognition is still learning, it’s more common than you think.
How do I make sure the results I want are returned?
- Issue: AI Assistants will faithfully read out the answer they’re given, which might match your answer.
- Solution 1: Don’t rely on external data which could be manipulated for your key information.
- Solution 2: Make sure that your information is up-to-date and easy for Google and the rest of them to crawl, index and understand the context of, so that it chooses your answer.
- Solution 3: Build a brand assistant of your own, like an Alexa Skill or an Action on Google, which lets your own AI-driven chatbot directly interact with your customers when they use voice search.
We’d love to hear your thoughts on the ad, so why not tweet us @homeagencyuk.
Written by:Neill Horie
Category:What we think
You may also like
/ 16 Dec 2019
The link between Margaret Thatcher and safe passwords
Passwords are generally accepted by the security community to not be a particularly good way of securing online accounts or sensitive information. This is mostly because of the way that society have been taught to create 'safe' passwords, i.e. by usiRead more
/ 25 Nov 2019
Targeting in a cookie-less world
Due to both GDPR and the introduction of 3rd party cookie restrictions from browsers such as Safari, Firefox and Chrome, fuelling targeted advertising with 3rd party cookie tracking may soon be a thing of the past. The ICO (Information CommissioneRead more