By creating a cohort analysis we learned that we are service different customers and the strategy change is needed.
A cohort is a group of people who share a common characteristic over a certain period of time. For examples: in our case: First buying customers in a specific year”
Cohort analysis is very useful to understand the customers repeat purchases over time. Is tells you:
- What portion of your revenue comes from new and existing customers?
- Are they spending more or less over time?
- Is one cohort more loyal to you then the other?
- What is the effect of strategy changes or of campaign?
We did a cohort analysis to learn more about customer loyalty or churn. To see if they spend more or less over time. From the Customer Factory we saw that churn was different then expected. With cohort analysis we dig deeper.
Steps we took
Define different cohorts
Start of all cohort analysis is understanding what to measure. We wanted to understand that how the different customers act over time. The business sales one once a year so the ‘fist year ordering’s used. Other business should use monthly sales as cohort.
Build your data
That turned out to take some time and we ended up bringing the data to one excel spreadsheet making use different backend systems. We used the customer email address is main identifier overcoming practical issues. Based on the ordering data we could add the cohort information. Extra attention was given to instructions for future use and on the decisions we took.
Finally we created the cohort overviews making use of this xxxxxxxx cohort excel template. We changed it to our needs, many adding the average spend per cohort per year. We did not use the Cost of Campaigns, but plan to do this in the future.
Compare and draw conclusions
With the overviews in place we were able to understand the customer behaviour, we could see changes average spend and churn patterns became clear.
Our business insights
Are we servicing different customer groups?
The average spend per cohort over of the the most recent years (2016, 2015 and 2014) dropped compared to the earlier years ( 2011 and 2012).
We see that new cohorts spend significantly less, while the products offered are unchanged. At the same time the average spend of elder cohorts reamins stable.
This leads us the question:
- Have we attracted a different type of customer with a different need?
It sure looks like the customer behaviour of newly attracted customers differs from the existing customers.
With the “Innovation adaptation lifecycle” in mind; does something like a ‘Wine life cycle adoption Curve’ exists?
2. The benefit of more sales moments exceeds the cost
We use to worked with the hypothesis that customers order wine for a full year ahead. Giving an extra opportunity to order halfway the year does bring extra revenue. The cohort analysis proved this hypothesis false.
We learned that over all the cohorts the total yearly send was 30% higher if we would offer wine twice per year.
3. The improved email marketing does show results.
We can see that in the year we changed our email promotion more customers for all cohorts procured again that year. Since this improvement is shown in all the cohorts it giving us the confidence to continue with this way of communicating.
Side note using the cohort analysis
Determine upfront how you will measure the effectiveness of your campaigns.
The cohort is more effective on a monthly time interval and needs roughly over 200 data points (sales) a month to give data you can trust.
The cohort analysis is a historical data and can’t be used to predict especially if over time the business model is changed, other customer segments are served or an other solution offered.
Benefits we have experienced
Comparing the cohorts was useful to understand the different behaviour of customers groups over time.
Having clear what revenue is from new customers and existing helps to determine and maintain focus on the growth strategy .
The cohort analysis helped to understand the return on investments on campaigns. What did we spend and what was the revenue generated?
But most important by creating the cohort analysis we have learned that our newest customers behave differently. They spend less and churn more. This is a clear indication that the new cohorts has a different need and wants. We see that our value proposition for this group needs change to maintain the growth we are looking for.
It sounds it’s time for renewed problem/solution interviews. Spend time on persona’s and on renewed Customer Development.
To be continued.