Better insights with improved segmentation
We released our first customer segmentation feature in August 2020. That version borrowed concepts from the RFM model used by marketers. We made some tweaks and we added a metric for how well customers pay, and it was good. The automatic grouping of customers into segments like Champions or Give Attention helps small businesses tailor sales and marketing to fit their best customers and the ones they risked losing. After working with our customer base, we learned a lot and realized there was room for improvement.
What we learned
Although our segmentation helps businesses better categorize their customers, one of the biggest places for improvement was to increase the importance of momentum. For example, two customers who both bought 15 times and spent $60k over the same 3-year period fell into the same segment. That happened even if Customer A started small and is growing quickly while Customer B started big and trailed off — and a business owner or sales exec would definitely prefer more Customer A’s, and would consider Customer B a churn risk.
Another issue was a tendency to overvalue frequency. For example, a customer paying $120K upfront for a year of service scored lower than a customer who paid $10K per month over the same year, even if both had done that for multiple years. But most businesses would consider the annual commitment to be at least as good if not more valuable.
Our improvements to RFTM+T
So we went back to our (virtual) whiteboard. We tried to create better customer grouping by applying more clustering using various machine learning models. But each SMB tends to have hundreds of customers, not millions of them, and they can be pretty different from each other. So the models had a hard time generalizing customers into useful clusters.
The best results came from applying more of the experience and knowledge that we’ve accumulated after working with hundreds of small and midsize businesses across all industries — and understanding how SMBs think about their customers and their growth. That led us to make the following improvements to recency, frequency, monetary and timeliness (RFM+T).
Recency: The last time the customer made a purchase is still the key input, but we also consider the typical customer retention period for each business. So if a business has an annual sales cycle, then customers who bought one month ago or 10 months ago are treated more similarly.
Frequency: Instead of looking at just the number of times customers made purchases, we now incorporate whether their purchase frequency is steady, increasing or decreasing.
Monetary: Customer lifetime value (LTV) is a great metric, but it can also overweigh the importance of sales made a long ago. So we now give more weight to the sales momentum of each customer.
Timeliness: The average number of days each customer takes to pay on invoices (aka Tally DSO) remains the key input. The new improvement adjusts for whether their average days to pay is less than or more than their typical payment terms.
A better picture of your customers
Each business has a unique customer mix and business model, and each one gets their own application of the segmentation mode, so the results will definitely vary! But a few changes are common across most SMBs.
First, it’s harder for customers to be in the Champion segment. Just being “big” no longer tips the scale as much. True Champions are still engaging and growing, and therefore have increasing momentum in LTV and buying frequency, while continuing to make timely payments.
But many businesses will see more customers in the Good and Promising segments. These are customers who are engaged and buying, but don’t show the same high growth rates as Champions. Now you’ll be able to identify these ideal customers and can spend time nurturing them to grow with you.
The ability to identify churn risks was boosted by increasing the importance of falling momentum and velocity. Now businesses will see more customers move into the Don’t Lose Them and Given Attention segments, giving them more opportunities to reach out before it’s too late!
| Customer Segment | Customers | Avg LTV | # of Purchases | Total TTM Sales | Avg Days to Pay |
|---|---|---|---|---|---|
| Champion | 22 | $163,450 | 133 | $1,711,005 | 17 |
| Good | 71 | $65,999 | 17 | $781,216 | 19 |
| Promising | 44 | $11,772 | 3 | $56,976 | 1 |
| Don't Lose Them | 14 | $25,148 | 8 | $44,980 | 12 |
| Give Attention | 16 | $16,299 | 6 | $27,362 | 20 |
| Hibernating | 29 | $14,325 | 4 | $0 | 12 |
| Possible Mismatch | 45 | $1,631 | 1 | $0 | 1 |
| Late Payer | 18 | $102,227 | 34 | $435,249 | 86 |
Use Google Sheets and CRMS to take action
A great way to put customer segments to good use is to push them into other platforms. A good place to start is to connect Tally Street to Google Sheets. This creates a “live” version of the Tally Customer Sheet inside Google. Then you can build out your roll ups on other tabs and easily share the results with your team members.
An even better solution is to sync customer segments to CRMs. Making the connection between Tally Street and HubSpot or Salesforce only takes seconds, and your company and account records are enriched with not only their segment, but also their sales and payment histories. Sales, Marketing, Finance and RevOps teams can then add this new information to dashboards, build smarter lists and trigger new workflows.
