Sometimes when you’re forecasting LTV, you want to see how projections based on recent performance compare to projections based on longer time periods.
Up until now, Everhort’s LTV forecasts were always based on a linear projection of the blended average performance of the last 12 monthly cohorts. There were reasons for this. If customer retention rate drops off after 6 months, you typically want to factor that into your forecast. But if average order value has been increasing recently, you want to account for that as well. A trailing one year period strikes a good balance between considering enough historical context to know how customers engage with the business over time, while still incorporating recent data, which is why it’s still the default time period for LTV forecasts in Everhort.
But sometimes your business changes significantly. Maybe you’ve been experimenting with new offers or eCommerce subscriptions. You’ve seen big changes and you’re curious how that could impact future LTV if recent trends continue. For this reason, we’ve recently enhanced Everhort’s LTV forecast tool to tailor its projections to the currently selected time period:
You can see from the example above that the one year forecast based off the trailing 6 months of cohort performance is $1,943, compared to $1,276 when using the last 12 months of cohort performance.
One last thing to note: When a shorter historical time period is selected, Everhort will also shorten the time period used in its forecast.
We believe having the flexibility to compare forecasts based on different historical time periods will help you get a better sense of what you can expect LTV to look like in the future.
We’ve recently added some new tools that work really well together to help you better understand customer behavior. These tools are available in your account now if you’ve connected your store using our automatic Shopify data import tool.
New Filtering Options
We’ve added two new ways for you to isolate different segments of customers for analysis in Everhort.
When you click to add a filter, you’ll now see a dialog like the one below, with tabs to select filters at the customer level or the order level.
On the customer tab, you can now select to filter by customer tag. Tags will be populated automatically from your store if you are using Everhort’s Shopify import connection.
If you click over to the “Order” tab, you can filter customers based on certain criteria about their orders. On this tab, we’ve added the ability to filter by product property:
Like customer tags, product properties will be imported automatically from your Shopify store.
You can use several different types of matchers when filtering by product property, including Equals, Does Not Equal, Is Set, and Is Not Set. You can learn more about how these matchers work can in our support center.
New Baseline Average LTV
After applying a filter, Everhort will add a new green line to the LTV by Cohort chart showing how the “baseline,” or unfiltered average LTV compares to the performance of your filtered customer group.
For example, let’s say we’ve filtered by customers who are tagged “Subscriber:”
The new “12 mo. average (baseline)” curve helps us see that all monthly cohorts in the “Subscribers” segment of customers, especially recent cohorts, have superior LTV curves. After 11 months, customers in this segment have an average LTV of $2,000, compared to an average LTV of only $1,226 among the customer base as a whole.
We believe the new filters and baseline average tool work well together to help you understand how different segments of customers perform relative to each other. We will be adding more types of filters and matchers in the future to allow you take this even further.
When you first start digging into how cohort analysis can help you learn about the health of your business, you’ll typically encounter a lot of triangle-shaped retention charts that look like this:
Orienting yourself to these charts can take a minute, even if you’ve done it before. What do all these numbers mean? What’s with the colors? And why is it triangle-shaped?
These charts tabulate the percentage of customers who return each period after their first purchase and make a repeat purchase. Each row represents a cohort of customers acquired during a given period (usually a week or a month), which is listed in the first column. Cohorts are ordered from oldest to newest moving down. As you move to the right along a row, each column shows how many customers from that cohort returned or made a repeat purchase in each subsequent period. Older cohorts have more periods of possible retention than newer ones, which is why each row is shorter by one column than the row above it, producing the triangle shape. As for the colors, typically these charts will highlight cells above a given retention percentage in green, cells below a certain percentage in red, and those in between in yellow.
Once you’ve familiarized (or re-familiarized) yourself with the structure of these charts, you’re faced with another, more challenging question: how do you make sense of the data? Should you read it top down? Left to right? Diagonally? As humans, we have a natural tendency to see patterns in data, and your eye is probably drawn to multiple patterns in the above chart. Are those patterns significant, or are they just noise? What do they mean? How is my business actually doing?
Retention charts are confusing and overwhelming. There’s a better, more intuitive way to understand the health of your business.
Lifetime Value by Cohort
Customer lifetime value (LTV or CLV) is the single best metric for understanding the health of a business. If average order value goes up, lifetime value goes up. If margins improve, lifetime value improves. If customer retention improves, lifetime value improves. The opposite is also true. If order value, margin, or retention drop off, customer lifetime value is also going to suffer. LTV should be your go-to metric for the simple reason that it reflects the aggregate impact of many of the underlying drivers of profitability.
If we view LTV through the lens of cohorts grouped by acquisition date, we can open up another important dimension of understanding: how things are changing in the business over time.
Let’s see how this works by looking at an example LTV by Cohort graph exported from Everhort for a fictitious company:
Orienting yourself to this chart is easy. Each line shows the cumulative average lifetime value of a cohort over time. The longer (and darker) the line, the older the cohort. The Y-intercept, month 1 for all cohorts, shows the average value of their first order. From this chart we can see that first order value has been increasing steadily on average over the last 12 months, because the shorter, lighter lines have higher Y-intercept values than the longer, darker lines.
By comparing the shape and slope of each line, we can understand how much value each cohort contributes to the business over time. Lines that increase more sharply provide more value more quickly. In the chart above, we can see that recent cohorts have steeper slopes than older cohorts. This means the business is improving at recognizing more value in less time.
Here’s an example of a different fictitious company:
Like the previous company, this company is steadily increasing average first order value, as the Y-intercepts are rising for shorter, younger cohorts. Unlike the previous company, however, this company’s cohorts quickly flatten out after about 6 months. Flattening cohorts are a warning sign in these graphs. It means customers in the cohort have virtually stopped engaging and are no longer producing value.
LTV by Cohort charts help us quickly answer important questions like:
How long do newly acquired customers engage and generate value for our business ?
How does the rate of engagement of new customers compare to older ones?
Is the business improving the rate at which it generates profit from customers over time?
Answering these question can help confirm if tactics being worked on to improve underlying drivers of LTV are working or not, or what areas might need to be investigated further.