Skyn Condoms’ new Facebook Messenger bot does more than just encourage dirty talk. It rates it.
Conceived by Sid Lee Paris as part of a global campaign (see the video ad below), the bot was designed and built in Toronto by Jam3. Released after approximately eight weeks of coding and testing, the bot invites users to submit a short voice sample through its Facebook Messenger channel. (It offers a sample script for those too shy to improv their own dirty talk.)
“The bot looks at three factors: voice amplitude, pitch and pauses,” says Mike Dobell, head of production at Jam3. “They all feed into a matrix that measures each and spits out results.”
After seeing their voice analyzed on “mysterious” “sensual” “sophisticated” and “intense” scales, the user is shown a link to Skyn’s product line in their chat feed. Clicking through finishes the bot’s interaction arc by recommending a product based on the voice sample. My “mysterious” and apparently non-intense recitation of a Shakespearean soliloquy, for example, resulted in a reco for the Unknown Pleasures Limited Edition line.
“We opted not to get too scientific about what a person’s voice says about their sensuality,” Dobell says. “We’re trying to create a balance of cool without over-sensationalizing what was being done.”
Dobell says Jam3 has been getting more briefs from clients and agencies asking for bot tech. Skyn’s “voice-activated profiler/condom recommendation engine” was the “third pitch we’d gotten like that that week.”
Jam3 creative director Greg Bolton says the increase in bot work has shown the agency that the nature of voice interaction software means their business has to increasingly rely on more than pure coding skills.
“The key to getting this right and being user-centric is really a writing exercise,” he says. Whereas a traditional smartphone app must look at button placement, font size, screen size and the like to make sure users can intuit how to use software, clear language is most important to a voice-enabled bot, so “writing is the new UI” in a voice interaction world, Dobell says.
Consumer and strategic expertise are also playing an increased role as bots come up against the more destructive elements of human behaviour.
“The first thing most people try to do with bots is break them,” Bolton adds, by testing the limits of their programming. Jam3’s job, he says, is to anticipate how users will try to do so and plan accordingly.
“You’d never think that that would be such a big piece, but if you want to do something really great, you do have to do that extra work, because there are so many of these things and otherwise they’re going to get ignored and abused.”
[iframe_youtube video=”oT78KHIEUE4″]
