Updated 4/13/2010: Our interface has changed significantly since this post was first written, so I’ve modified the images and descriptions to reflect that
A brief introduction
Funnel analysis is a way to measure how users go through a specific set of actions on your webpage. Anywhere you need to get conversions, like a signup or purchase, can benefit from this kind of analysis.
An example funnel would be a signup process for new users, which could look something like:
front page -> signup impression -> signup -> confirmation
Users would drop off at each of those steps. The goal of measuring the funnel is to minimize this dropoff rate.
Tracking funnel steps
One really cool thing that Mixpanel can do is track events as funnel steps instead of relying purely on pageviews.
This means, for example, that you can figure out exactly what percent of people add an item to their cart – which is an action that doesn’t change the page but is crucial to the checkout process. It’s also a major dropoff point, so you definitely want to be able to track it in your funnel.
Once you’ve figured out the steps of your funnel it’s really easy to integrate. After integration, we get to the fun stuff – the report.
Viewing the funnel
Here’s what a funnel report looks like (with a few notes):
So, with this display you learn:
- Number of unique visitors at each step
- Conversion rate between steps
- Overall funnel conversion rate
Which are all amazingly useful things to know.
But wait – there’s more!
We actually take this a step further – the same concept of properties, when applied to funnels, allows you to store any extra data you know about that step of the funnel and see how it effects user behavior.
You can track things like:
- ad campaign
- search string
- anything you think would effect conversion rates
Below we are tracking which promo offers users are looking at, and comparing the overall conversion rates for each one:
We can see that users who viewed the Â“next purchase discountÂ” promo converted the best through Step 1. This is interesting, but since we donÂ’t make any money unless a user actually clicks pay, a more important question might be which promo offer has the highest overall conversion rate. We can see that the Â“half offÂ” offer converts the highest at 32%.
Being able to track specific things across the full length of the funnel allows for some interesting applications – such as site-wide A/B testing. You can show different versions of the site to different users and see not only how it effects one level of conversion – clicking a signup button, for example – but how it actually effects the things you really care about which happen later.