Qriously was the only data company to predict the Labour surge in the last election, a story that Wired picked up and was one of their best-trafficked stories of the year. Company CEO Chris Kahler explains what happens next.
We are a research platform to understand and predict human behaviour in real-time, in particular with event-based predictions such as political elections.
The main alternative to our approach is to use traditional polling which is becoming increasingly inaccurate (as evidenced by the recent spate of hilariously wrong predictions).
We are in a state of political tumult and we think accurate polling is an important feedback mechanism for a government to function properly. This applies to readers ‘in tech’ and beyond.
What is currently happening in the world that makes Qriously relevant?
Ummm, the world is totally fucked. Russia, Putin, Trump, China… Name it! It feels like no one knows really what’s going on.
Right now it’s quiet, but the mid-terms in the US next year will be huge and we might be lulled into a false sense of security with Germany. Populism is threatening the EU.
Can’t argue with that summation. So what needs to be done?
Polling is broken but the need for it hasn’t diminished. That means something needs to be done about it if governments want to know what citizens really think.
Research and polling methodology has seen extremely little innovation since the invention of the first online panel. In particular, research methods have barely taken advantage of smartphones although it’s the most universal medium of information creation and communication of all time.
How, specifically, is Qriously helping to solve this problem?
Without diving into too much detail and giving away too much to our competitors, Qriously has developed a methodology that produces more accurate results and up to 10x faster than traditional methods.
The main problem our methodology is solving is that of sample bias: traditional landlines/panels capture an increasingly skewed subset of the population. Our method captures random people in random apps, making them more likely to random people.
So, it’s a mobile-first product?
Yes. We’ve developed a research methodology that replaces ads with surveys in smartphone apps. Instead of building a panel, we use machine learning to create representative samples in real-time.
Sounds simple enough, but describe it to somebody who knows nothing about technology.
Do you know those pesky banner ads you see in some smartphone apps and games? Qriously replaces those with surveys on smartphones that you can answer if you want to.
Because so many people have smartphones nowadays, we can get better data than other methods which use landlines (who has those nowadays anyway?) or online panels which are groups of people that are paid to answer surveys.
The Brexit decision, which you predicted, is making people nervous. Do you intend to stay based in London?
At the moment, yes, we plan to stay in London. Looking at how much work goes into predicting an election or referendum outcome, predicting something much more complicated like the political future of the UK is truly in the realm of pure speculation – we’re not that good yet!
We originally came to London because of access to capital, talent, and a diverse set of markets (important for young startups experimenting with business models). At the moment, those continue to be compelling reasons for us and many other startups to stay. Of these, however, talent is the main concern.
So how did you get this point and where did you start?
In 2010 my co-founders and I built a few simple consumer-facing apps for fun, just to see what would happen. In 2010 many apps weren’t designed really well and searching for apps didn’t work very well (it’s still nowhere near as good as web search).
So we discovered that if you built well-designed apps and gave them really boring descriptive names, you got tons of downloads (in our case, more than 30 million across all the titles we released).
We wanted to find a way to monetise those apps and began experimenting with display ads. However, we didn’t like the way those ads looked in our apps so we decided to do the only logical thing which was to build our own native ad unit. Once we had the unit, we began tinkering with other modes of engagement such as simple questions.
We were blown away by the response rates and once we did some basic data validation we knew this was something special, that we had stumbled across a new way of gathering data that wasn’t possible before. We also had our own user base to test out the technology and iterate quickly.
How much traction in the market do you have?
* Hundreds of clients across a range of industries, including government organisations, hedge funds, and brands.
* Hundreds of millions of answers
* Millions in revenue
Finally, Chris, you’re clearly good at predicting what other people are going to do? What’s next for Qriously?
Up until this point we’ve been busy getting the methodology right and building the technical infrastructure. That’s largely completed so the next area for us to focus on is opening up the predictive power of the data to everyone so we’re building a self-serve platform. We can’t say too much at this point, but we’re really excited to see what happens you open access to the opinions of >1b people all over the world.
That was awesome, Chris Kahler, thanks for sharing your story.
My pleasure, thanks for having me.