We have a pretty simple definition of research here at the zoo:

Identifying difficult user problems and turning them into solvable engineering problems.

There are three different kinds of research we do under that general charter…

1. “User Experience” — This is our qualitative user research work.

We constantly monitor the user experience for unintended funkiness, and also for opportunities where we could make it easier for users to experience full-fledged Aardvark awesomeness.

Then we go out and interview lots of existing users, and potential new users, to see exactly what they think.  We also recruit users (and not-yet-users) to participate in “dry runs” of our new ideas, to see if they’re worth building.

The knowledge we gain here is what drives our feature design process.

2. “Analytics” — This is our quantitative user research work. 

Whenever we hear ourselves saying “I wonder if our users like X” or “I wish we knew how to improve Y” or “How can we stop unpleasant experience Z from happening?”, we put that story on our analytics queue. 

It’s a chance to run the numbers, look at a heatmap, try to find some correlations, and diagnose exactly what’s going on.

When we’ve got a clue (or even something that looks like a clue) about the phenomenon, it’s on to designing solutions. 

3. “Algorithms” — This is our core technology research work. 

Once we know the kind of solution we want, then we work through our various tricks until we get a breakthrough.  Most of what would traditionally be considered “technology innovation” falls into this bucket; things like fancy person-to-person matching algorithms, special-purpose classifiers, and question analysis techniques. 

Whenever we come up with a new algorithm, we run it against our test benchmarks to see how it does on our activity datasets… and if it works, then it’s time to tune it for production engineering.

We’re fortunate enough to have a top-notch research team, with both extremely experienced veterans and young talent.  We’ve invested heavily here early on, because we see research as a key differentiator as others start to imitate our product.  And because it’s awfully fun to see what a clever machine can discover in the data.  

Stay tuned for more about the specific research projects we’re working on now…