startup analytics

Getting Started with Startup Analytics

At last week’s New York GrowHack Meetup, analytics expert Cassie Lancellotti-Young, dove into a case study of a new growing product called HighTable.

Although HighTable at first seemed like it’d be a bust, Cassie shows how her experience using the lean marketing funnel and running growth at startups like Savored (acquired by Groupon) and TheLadders.com paid off big in its turnaround.

Here’s what happened.

HighTable 1.0
HighTable first launched as an experimental product of the Gerson Lehrman Group – a Quora for Professionals. When Cassie the product had already been built to do six things:

  1. Ask question
  2. Answer question
  3. Vote on answer
  4. Follow person
  5. Follow topic
  6. Follow question

Then came the moment, everyone dreads when working in startups, the “Oh shit” moment..

Although HighTable had over 100,000 registered users, after running the analytics (Cassie teaches a cult-status cohort analysis class on how to do this) it became obvious the product was doing a horrible job of not only activating new users,  but also with retaining and engaging the users they had.

It turned out 85% of activity was happening on day 1 of a user’s life. On top of that, many of these initial users were inherited from GLG’s main business, rather than organically.

After the initial shock, HighTable needed to make changes if it was going to get off the ground.

Cassie took an inventory of the available information on potential users and the team built the next version of the application with a focus on directing users what to do and around simple calls to action.

These steps led to a great 136% increase in answers/visits. However, this simply wasn’t enough engagement that was needed to grow HighTable into an influential product.

HighTable 2.0
The next step was to take the information they’d learned from the now failed product and start from scratch.

Just launched in September, the new HighTable 2.0 has already posted some big early results including: 68% usage (compared to 10% before) and a viral coefficient of .5+.

The change in focus was dramatic. What did they learn? Instead of focusing on a broader “professionals” market, HighTable created and tested a small MVP which gained traction when it was tested with startup members. HighTable would now turn into a private community of early-adopter startup leaders.

Focusing first on retention and engagement, each potential user who wants to join must go through an endorsement process.

After connecting to LinkedIn, HighTable makes it easy to email 2-5 contacts and endorse you. HighTable makes it clear that the more and higher quality the endorsements, the more your application will be considered for the community.

Another check to keep quality level high is once a user is directed back to do an endorsement, each contact is asked whether they consider the applicant one of the “top 5 people” they know with a particular expertise.

So what are the major lessons from the HighTable experience?

Measure and understand your users. Are they coming back and using the product? Are they acting the way you’d expect?

Although there are different types of analytics to run, a cohort analysis is a solid first step. It empowers someone tasked with growth to gain context on their user base. This can be done using MixPanel, KISS Metrics, or crunched manually in Excel.

The point for this first type of analysis, is to separate out the time period a user segment joined and the revisit % rate. This gives you the ability to understand the health of the site and compare performance over time. As you make changes, your revisit rates should rise.

Learning that 85% of HighTable 1.0’s usage was on day 1 meant they needed to change their approach, quick. This change ended up being dramatic, with HighTable 2.0 being released without any of the initial six activities it started with.

How do you know what’s good retention? Well this depends a lot on your product, so investigate other products in your category. Flurry, which has insight into 230,000 apps, put out a useful report in October on mobile metrics, focusing on frequency of use and retention over 90 days.

It turns out, most mobile apps have between 1 to 5 uses per week. However, if your communication or music streaming app is in this range, you’d be underperforming the average of around 9 uses per week.

Second, start with a single simple feature set, then layer up. Facebook’s first product started with meeting a core need for college students, then went all the way to launching features like the Newsfeed.

After talking about HighTable Cassie goes into each step in the lean marketing funnel in great detail. This is a must see:

Now onto you. Leave a comment below on how you’re using analytics to get you to the next stage in your business.

Comments on this entry are closed.

  • http://www.vuzum.com/ Andrei Potorac

    Actually I think this is a very solid example. I bet many of the startups currently out there face the same problem, except they don’t know it.

    Building a product is a frist step, working on having the metrics properly setup is another. Than invite users, and use that data you collect, to iterate on your product.

    Loved this example, it’s frightening to know you can end up building a complete project and than you figure out pivoting is the right way to go.

Read previous post:
lightbulb2
Discovering Your Aha! Moment

One of the most interesting takeaways from the Growth Hacker Conference in Silicon Valley this past October was the importance of...

Close