Every year I make the same mistake
About a month before the Cannes Lions festival, the biggest advertising/creativity event of the year - conveniently located in the south of France, I always make sure to get my yacht in order. I stock the fridge with Dom Perignon, hire a staff and make sure the sails are in order. Then I wait patiently by the mail for my Cannes invitation to come...and this year, once again, it didn’t come. Just to be sure, I set sail for Cannes anyway - maybe someone will text me last minute? But, year after year I sail past Cannes - past the parties, laughter and fun, only to dock at Monte Carlo, where I drown my sorrows at the casino. It’s a sad life.
This intro had a point...
The reason I bring Cannes is to focus on a company that made its first Cannes appearance this year: TikTok.
TikTok is a platform where users upload short videos. People love it. It was the most downloaded iPhone app for the past five quarters. Global daily users average above 500mn. In the US, TikTok users login in almost 8 times per day. How on earth did it get so popular?
TikTok is AI/ML
What I mean, is TikTok relies immensely on the use of AI/ML for everything. AI/ML determines what video you see.Every swipe, tab, pause is an indicator to the algorithm about what video to show you next. The filters you can apply to videos are powered by AI/ML. The effects you can add are powered by AI/ML. Everything on TikTok is driven by AI/ML.
You could say, “Ok, well other platforms use AI/ML as well. Just look at Netflix.” That’s true. But, what’s different about TikTok is they always knew AI/ML was going to be core to their business. From the very beginning, TikTok started integrating AI/ML features to drive engagement.
TikTok represents the second iteration of how apps will use AI/ML
TikTok got me thinking. Companies used to launch apps and then add small amounts of AI/ML to it, which I’ll call iteration one. Examples of this include Netflix serving you recommendations or Facebook filling your news feed. These companies had an app, wanted to increase engagement, and then added AI/ML.
Then, iteration two came along
Companies realized how effective AI/ML could be and started putting it everywhere. The Netflix version of "iteration two" means the entire app experience is driven by AI/ML. Each home page for each user is personalized, which titles and in which order Netflix shows you is them is personalized, even the artwork thumbnail - yes, the artwork thumbnail - of each show is chosen based on your viewing characteristics. All this personalization has led to significant results: 80% of content consumed on Netflix comes from personalized recommendations.
Which raises a question
Do all new apps need to include AI/ML to optimize user engagement from the very beginning? If you are going to be an aggregator, then the answer is yes. But even for any app, the answer may be yes. Current apps have such a dominance over user time spent with Facebook, Instagram and Snapchat users spending on average almost an hour in each platform a day. In order for people to even have a chance, you need to hook people from the beginning.
What does iteration three look like?
Iteration three is using AI/ML to produce content/interactions to keep people hooked on the platform. ByteDance, TikTok’s parent company, has already started to generate stories using AI/ML. As general adversarial networks evolve, even more content will be produced by AI/ML. This could be a way to bootstrap a network. Instead of relying on people to produce content from the beginning, AI/ML can produce a plethora of content so even if the network has one user, they have plenty to view.
At the end of the day - it’s all about ads.
TikTok’s global revenue was about $4bn in 2018 and the company is looking to expand into other offerings. According to the FT, they include “standard vertical video ads typically offered by rivals such as Instagram, ads that splash across the screen before the main content feed loads and branded camera filters, similar to those available on Snapchat.” As ad revenue continues to grow, expect TikTok to be a major force.
And expect AI/ML to be be most of that driving force.
Every year I make the same mistake