Analytics: not just for optimization
When we pitch Mixpanel to people, a lot of them tell us
that they’re busy working on their core product, and that they’re not at an
optimization stage yet. I’d like to point out how valuable analytics data can
be for learning about your customers – something that is crucial to product
A couple of really valuable aspects are
- No selection bias
- Understanding feature use
Selection bias often occurs when companies ask customers for information. The
customers with the strongest opinions are most vocal, and skew the dataset.
This means that the data you get back from surveys may not actually be a valid
representation of your customers.
A great thing about automated analytics is that there is no selection bias.
You track what EVERYONE is doing, so there’s no way to get only a subset of
opinions. That means you can trust the analytics data you gather a bit more
than the feedback you get from talking to users.
As a startup, we spend a lot of time cranking out new features. However, we
can’t afford to spend too much time working on something before we know it
works. This means we push out new stuff way before we’re entirely happy with
it – but it lets us start getting actionable feedback immediately.
This feedback comes in the form of data. We can instantly see if the new
feature is getting use or getting ignored. If it’s getting use, we see what
aspects of it are most popular and we can focus development on those.
If it’s getting ignored, that is just as telling. A new feature that no one
wants is useless, and we either lower its priority or scrap it entirely.
We heard a real-life example of this from one of our users the other day. Joey
Liaw of Voxli (YC W09) is using Mixpanel, and he posted this to
Top three push to talk bindings for our users this week: 16% use Left
Control, 11% use Middle Mouse Button, and 10% use Tilde. I put mouse support
in because David A. asked for it, but I thought he’d be the only person using
it. Glad I was wrong. Would be interesting to correlate this against what apps
users run, or their Arena ranking.
He only knows this because he started tracking it. It sounds obvious, but so
many people are flying blind – they release new things and have no idea what
I’m sure you can think of times in the past when you added something to your
product and never knew if it was getting any use – what were they?