via Data Direct
How do you measure, or more appropriately, identify your best customers? If your business is like most, 20% of your customers contribute 80% of your profits. Do you know this segment? How do you identify and measure them in order to implement a successful contact strategy?
RFM is not new. In the direct marketing world, RFM provides both a scoring model and a segmentation schema to help maximize the efficiency or productivity of a mailing or campaign. It stands for RECENCY / FREQUENCY / MONETARY VALUE. Savvy catalogers once segmented their list by RFM – the premise being that a scoring value for each variable would bubble the best customers to the top. For example, a customer who bought within the last six months would “score” higher than a customer who bought six to twelve months ago. A customer who bought three times over the last twelve months would “score” higher than one who only bought twice. A customer who spent $200 over a defined period would “score” higher than a customer who spent $100 over the same period. RFM provided the foundation for later, more complex, multivariate predictive models and more complex segmentation.
It is likely that your highest scoring decile or pentile group would be, by far, your most profitable, and would merit the highest spend to drive retention. These customers were, at a minimum, loyalists and more than likely were brand advocates.
I’ve been doing a lot of thinking about how RFM could be used in today’s digital marketing environment. Just the framework of the schema itself is helpful in clarifying our thinking around how to best optimize our marketing investments as they relate to customer contacts.
The original thinking was based simply on purchases that could be tracked by a household address. In today’s digital, multichannel environment, we may only have an email address, a cell phone number, or a cookie. But are these any less valuable? Consider this scenario – customer X downloaded your iPhone app within the last 30 days. Customer Y downloaded the same app over 90 days ago. Which customer is most likely to respond to a follow up SMS text containing a promotional offer? Using recency as a predictor, the most recent customer is most likely to respond.
On the surface, this seems obvious and simple. Diving deeper, things begin to get a little more complex. Consider all of the interactions with your brand in the digital space. Consider a Facebook fan who posts frequently, say daily, on your wall. Are they more or less valuable than the fan that posts weekly? Maybe. Maybe not. But if we were to look at brand interactions the same as purchases and actually assign an RFM value to each customer based on them, a defined segmentation would begin to emerge. If the daily Facebook poster only interacts with that brand in that way, but the weekly poster also recently downloaded your mobile app, is the mayor of one of your brick and mortar locations on Foursquare, clicked through on your most recent email, and visits your site frequently, perhaps they are a “higher scoring” RFM customer and therefore a true brand advocate instead of simply a loyalist.
How could you use RFM thinking to help define a segmented contact strategy? How could this strategy maximize your digital ROI and move customers through the relationship funnel as clever direct marketers once did using print and direct mail?
This article is from
Greg Chapman is a seasoned veteran with over 30 years of experience in direct response marketing and merchandising. As a former Divisional Vice President at Sears and VP of Merchandising at both Skymall and Value Vision (now Shop NBC), Greg is the ultimate frontman. He knows what clients want because he’s been in their shoes; he understands the fears they face in transitioning from traditional advertising to digital. His neo-corporate methodology focuses on building a strong, performance-based internal team to create a better performing external product.
has written 11 posts.
{ 1 comment… read it below or add one }
It seems to me that the key would be to identify the appropriate “RMF-type” attributes in order to provide a level of measurement consistency across your sample. These actions would have to be measurable in the sense that RMF is measurable. The nice thing is that online interactions are generally much more measurable!