How a Time-Tested Segmentation Strategy Can Maximize Your DTC Potential
By Alex Caffarini
In an April 28 article, I discussed how wineries can harness analytics to thrive amid uncertainty. One customer segmentation technique I briefly mentioned — RFM analysis — deserves deeper discussion, since it’s both simple and time-tested.
RFM stands for Recency, Frequency and Monetary Value. It segments customers based on how recently they’ve purchased, how often they buy and how much they spend. Marketers discovered in the 1930s that they could predict a customer’s likelihood to buy again knowing just those three things. RFM’s approach was so effective that it’s still used today across industries — including retail, banking, gaming and nonprofits.
Given how heavily wineries rely on DTC channels, RFM is a natural fit. It gives you an easy and powerful way to identify your best customers and make strategic marketing decisions quickly.
Getting Started With RFM
To begin, decide on a timeframe to evaluate: two to three years is sufficient, while more than five is overkill. Pull transaction history for the period from your POS system. For each unique customer, calculate:
- Days since last purchase (Recency)
- Number of purchases (Frequency)
- Total amount spent (Monetary)
Your POS system may generate these stats automatically. If not, you can calculate them in either Excel or Google Sheets.
Then, for each of these three dimensions, sort and rank your customers from 1 to 5, where 5 indicates your top 20%, 4 your next 20%, and so on. Each customer ends up with a three-digit score. For example:
- “555” = recent, frequent and high-spending
- “523” = recent, occasional buyer who spends moderately.
- “111” = least valuable customers.
Each RFM score represents a segment, or cell, you can target or track. Once you’ve scored your customers, here’s how to transform those segments into strategy:
Identify Hidden Buying Patterns
Summarize trends within each RFM cell. You might discover:
- High-frequency buyers favor white wines.
- Certain cells over-index on wine club memberships
- Some cells show higher event attendance or tasting room visits.
These insights can help you tailor messaging and product offerings.
Boost Promotion Effectiveness
Run a test campaign to a random sample of your customers that includes all RFM cells. After the campaign, analyze:
- Response rates by cell: Who bites on your offers?
- Average order value by cell: Who’s worth marketing to?
This lets you fine-tune your promotions and customer targeting.
Measure Customer Profitability
Say it costs $1.50 to contact a customer, and the average profit per order is $60. Your break even response rate is 2.5% ($1.50 ÷ $60). RFM helps you identify the cells that clear the hurdle — and which to skip next time.
Personalize Offers with A/B Testing
Assume you offered half your customers free shipping on $200+ orders, and the other half $25 off the same threshold. RFM can help you pinpoint which cells respond better to which incentive.
Some Tips to Keep in Mind
Although intuitive, RFM comes with caveats:
- Don’t treat scores as math. They’re categorical. While “555” and “111” are your most and least valuable customers, respectively, the value of each cell in-between is much grayer. A “522” isn’t necessarily better than a “225.” Don’t blindly rank or sort the cells. Rather, analyze them contextually. Otherwise, you’ll place too much weight on recency and too little on monetary.
- Not all of your cells will be populated. Using five tiers per dimension (as we did above) gives you 125 cells (5x5x5). While that seems overwhelming and unwieldy, many cells will have few or no customers. Approximately 80% of your customers will be clustered in 20% of your cells, so you may be working with just 25 cells.
- Don’t ignore low-value cells. Some of today’s “111” customers may become tomorrow’s “555”s. And if you oversaturate your best customers, they may tune out or churn.
- It’s transactional, not demographic. RFM focuses solely on transaction behavior. It doesn’t account for age, income or whether someone is a club member or a tasting room visitor. You’ll need other tools to segment on those dimensions.
Maximizing RFM: A Few More Best Practices
RFM analysis isn’t a one-and-done. You should expect some trial and error before you get it down pat. To get the most out of RFM:
- Rescore regularly . Customers evolve — some become more loyal, others drift away. Rescore your customers quarterly, semiannually or yearly, depending on your winery’s business needs and your time and resources. At a minimum, rescore annually.
- Track customer migration. Record each customer’s historical RFM scores to track changes over time. Are high-value customers slipping into lower-value cells? Are some newcomers moving up fast? This can reveal churn risks or signs of rising loyalty — and help you estimate customer lifetime value.
- Compare and evolve. RFM is a great starting point. As your marketing becomes more sophisticated, you may test more advanced segmentation, such as clustering or machine learning. But don’t abandon RFM too quickly. If it performs just as well, save yourself the complexity.
A Perfect, Simple Start for Wineries
Whether your winery is new to marketing or wants to make existing campaigns more effective, RFM analysis is a great place to begin. It’s straightforward, low-cost and immediately actionable. And, in a DTC landscape where every marketing dollar matters, RFM helps ensure you direct those dollars to the segments yielding the highest return.
_____________________________________________________________
Alex Caffarini
Alex Caffarini is the founder and president of CompassDTC, a strategy and analytics consultancy that helps small and midsized food and beverage brands — wineries included — unlock growth through strategic DTC marketing. With over 30 years of experience in marketing analytics and data science, Alex has led high-impact CRM and direct marketing initiatives for banks, catalog retailers, and national grocery chains, and has built segmentation models for premium California wineries. He specializes in developing predictive models that identify high-value customers and optimize promotional performance. Reach Alex at alex.caffarini@compassdtc.com.
