Forget Realtime, Sequoia-Funded Taykey Shoots For Pre-Time

What makes Sequoia invest an undisclosed multi-million dollar round in a 22 year-old’s startup? The future. No, not in the sense that young entrepreneurs hold the future of our industry. Literally the future. Or to be more exact, selling advertisers future trends. Of course to pull that off one needs some precognition.

One route was to comb the earth for mutated humans and exploit their gift, but Taykey took the other route – algorithms. Watch out, Agatha.

Taykey spent the first two years of its life developing algorithms that crunch extraordinary amounts of user-stream data, to determine trends, before they happen. The company is founded by Israeli intelligence corps graduates so rest assured that there’s true trending analysis and prediction technology inside. Also, Sequoia’s investment implies to me a diligence process that looked beyond vaporware and buzzwords and found real IP instead.

Before Taykey starts analyzing and predicting, it must first collect user-generated-data from across the web. It does this by monitoring real-time stream sources like Twitter’s and MySpace’s data hoses, diggs, items publicly posted, liked and shared on Facebook, location data through check-ins, what users are watching on YouTube, and Wikipedia updates. It also keeps tabs on what it deems to be ‘leading blogs,’ and news sources to see what’s new and what people are commenting about. All of this is fully automated and happens continuously, in realtime.

At this point a ‘Privacy’ flag should be popping-up in your head. In this respect it’s important to note that Taykey neither ‘cookies’ users, nor does it save any personal data. It says it cares exclusively for the public data, with no significance as to which individual contributed what.

Here’s how Taykey forecasts a trend before it actually happens: data such as videos being posted on YouTube, Likes and Tweets about something called ‘Susan Boyle,’ all flow into Taykey. Within a matter of minutes, their algorithms determine that there’s something new going on—an object—called ‘Susan Boyle”.

Taykey then analyzes and employs mathematical matching history upon the volume, the destinations and the duration between each engagement with this new object. Doing so allows it to tell just how big this ‘Susan Boyle’ trend will become, which demographics are and will be engaging with it, and that it’s related, in this case, to music.

And this is where the commercial side of things become really interesting. Taykey includes granularity for micro-trend forecasting. Let’s say you’re a fashion retailer and want to target 16 year olds. Taykey knows that right now this demographic is talking about Justin Bieber so it would place a first set of ads along side such mentions in order to create a psychological association.

However, Taykey already knows that, in five hours, users in this very same demographic will be talking about Lady Gaga and it would then show a second set of ads next to Lady Gaga mentions. Think about this for a moment: Taykey basically claims to be aware of what the user will be talking about before the user does.

Taykey’s reporting includes benchmarking, enabling advertisers to constantly quantify the lift it provides by running a trend-based ad vs. an “old-fashioned” demographics-based ad.

From a business model stand point, Taykey is basically a Demand Side Platform (DSP). When it recognizes a trend, it goes out and buys the relevant audience for the relevant amount of time. It then charges advertisers a premium CPM to deliver the ads, and uses arbitrage to maximize its cut.

All this talk about forecasting makes me wonder… Can Taykey forecast the future success of its own technology? Might want to call back Agatha for that.