How to create awesome Retention Curves from Mixpanel Cohorts - what you need to know
Retention is the Key to both Traction and Growth
If you've been paying attention to much of the great content out there on growth, you've probably been noticing that retention is the MOST important aspect of both early traction and long term growth.
Retention in a Nutshell
For your business to grow you need to have more customers (or users depending on your business model).
But, what if you lose customers? That's like going backwards. The faster you lose customers, the harder you need to work to get new customers to replace them and move your business forward.
Retention then is about making sure you keep customers each month, so at the end of each month you have new customers you have retained PLUS the customers acquired (and retained) in previous months. Over time this builds.
Retention is a MULTIPLIER
Retaining more customers for longer means:
- Higher Lifetime Value (LTV): you can make more money per customer (if a customer is worth $X per month to you, then a customer who stays for 10 months is worth DOUBLE a customer who only stays for 5 months.
- Acquire Customers Faster: for every $1 you spend on acquiring customers, if instead of making $2 you make $4 you will have more dollars to spend acquiring new customers.
- Afford Higher Customer Acquisition Cost CAC: given you make make per customer, you can afford to pay more to acquire customers. This may open up new channels that were previously too expensive.
- More Referrals, more Word of Mouth (WOM): given your customers stay with you longer, you will have more opportunity for those customers to invite new users, or recommend your product to other potential customers (and WOM is a very strong channel).
- More Up-Sell, More Cross-Sell: as your customers stay with you longer, you will also have more opportunity to upsell higher plans (as their business grows) or to cross sell other related products.
If you don't have good retention, fact is your business will struggle to survive and grow.
"If you have poor retention, NOTHING else matters." (Brian Balfour, VP Growth @Hubspot, Investor, Founder, Growth Advisor)
Understanding Retention via Cohorts
A Cohort is a group of people who have something common. If you graduated from Stanford university in 2005 you would be part of the 2005 Stanford Graduates cohort. The Millennium Generation is a cohort. Baby boomers are a cohort.
When it comes to your business, cohorts are most useful when we apply them to a start date and this is typically the Signup Week or Signup Month (January Signups, February Signups, March signups, etc..).
Our goal with Cohorts is to compare behaviour on each successive month, and achieve an improvement with each successive month. So, if we're measuring how frequently users come back, we'd want to see that our March users come back more than our February users, and that our February users come back more than our January users.
I want to keep this article light, but you should definitely read how to measure if users love your product using cohorts by Andrew Chen.
Retention Curves show the Trends
Cohort Reports are often provided in table form - but they aren't so easy to read in table format. So if we can get that data into a nice graph where we can "see" the data and the trends, its so much easier to see how our retention is (hopefully) improving.
How to Create Retention Curves from Mixpanel Cohorts using Google Sheets
Ok, now you know WHY retention curves matter, I'm going to show you how to turn your Retention (Cohorts) Reports in Mixpanel into beautiful Retention Curves using nothing more complicated than Google Sheets. This article was inspired by a post by Dan Croak with one KEY difference - that here we create the retention curves using percentages, which makes cohorts instantly comparable, regardless of the actual number of users in any cohort. It takes about 5 minutes.. Ready? Lets's go!
Export from Mixpanel - in 3 mins
PART 1 - Grab the Data from Mixpanel:
STEP 1. Login to Mixpanel and Select Retention
SIDE NOTE: If you aren't yet collecting this data in Mixpanel, then I recommmend you start collecting today. Its easy to get started and you can use a service to easily capture and send data to mixpanel so you can get setup without needing to go talk to your developers. As a minimum you should track user signups, and key actions they take on your website/webapp.
STEP 2. Select "First Time" - this will allow us to create cohorts by Signup date.
STEP 3. Select people who did Signup (in this case I'm selecting our REGISTER FORM event, it depends what you've called the signup event in yorr Mixpanel) - Then select did “Anything”.
NOTE: Later, instead of "Anything" you could select by a key action so you could build retention curves by key actions to find the actions that best retain users. For simplicity though, we'll start with people who came back and did "Anything".
STEP 4. Optionally Select: Past 12 Months, Past 12 Weeks, Past 14 Days. For this example, I'm going to select Past 12 Weeks.
STEP 5. Select Export CSV: In the bottom right corner there is a small icon that will let you export the data as a CSV file. Go ahead and download the data.
(NOTE: Mixpanel only exports the totals not percents, but we'll solve that in a moment.)
PART 2 - Upload to Google Sheets and Create the Retention Curve:
STEP 6. Open Google Drive and Upload the file. Then open the file.
STEP 7. Open the file in sheets - you'll see the same data as in Mixpanel. Remember these are totals, and that’s okay, but cohort sizes can vary, so to be comparable I prefer to look at percentages. (because what we're interested in is comparing behaviour of the cohorts - regardless of the cohorts size.)
STEP 8. We're going to take an extra step here and turn the totals into percentages, so our curves are more comparable.
STEP 9. Select, copy and paste ALL the data below, so you have two copies.
STEP 10. In the copy section, edit the first cell (shown) to calc using the data above.
The cell formula will be this: =if($B2, C2/$B2*100, 0)
STEP 11. Format the cell (pick any colour you like, its only so you can see what we're changing)
STEP 12. Copy and paste that cell into the other cells in the copy area. Also, delete the Cohort Size data from the copy area, because we don't want to show it in the chart.
STEP 13. Select all the data we've modified, and select Menu Insert > Chart.
STEP 14. Set the following settings:
Chart Types: Line (Curve)
Switch rows/columns = True
Use column A as headers = True
STEP 15. Using one colour for all lines can help make the data standout better. Try to use one column in the colour picker, and make more recent months be a darker shade. (If you run out of colours you can switch to thicker lines, or use colours in pairs.)
BOOM! You’re done! Easy huh? You should now have a retention curve that looks something like the curve below.
For this particular set of data, you can see the typical shape of a retention curve, with a fairly fast drop off during the first week (which is when people's attention is generally higher) and over time the curve levels out (showing around 13% longterm retention). Also note that some of the most recent curves are higher (showing better retention) which is what you would expect as a company works to improve retention.
Of course measuring user retention is only a part of your growth strategy and needs a good marketing analytics tool stack. If you want to set up an essential stack with key tools covering acquisition channels, conversion funnels & retention, and automated customer messaging read our 7 Essentials of Marketing Analytics - a Beginners Guide.
I hope this has been useful to you, and if you’ve struggled with understanding user retention I’d love you to tell me how you tackled it in the comments below.
p.s. If you ant to check your current Marketing Analytics Stack (or your competitors) you can use this FREE online Marketing Analytics Grader tool.
If you've enjoyed this post and don't want to miss my next post you can subscribe here.