Jim Novo - Turning Customer Data into Profits

It wasn't uncommon to simply mail everyone on our list a catalog. Of course, this was an expensive proposition, but we didn't have any way to determine the relative value of our customers or their level of engagement.

Fortunately, we had just hired a new database marketing analyst and he quickly educated us on the importance of using both customer Recency (date of last transaction) and Frequency (number of transactions within a defined time period). This is known as RF Scoring and it's a hot topic in Jim Novo's fabulous book Drilling Down.

By utilizing RF Scoring, we were able to increase our overall direct mail response rate as well as save money by not mailing expensive catalogs to the folks that were least likely to respond.

It turns out that there is a lot more you can do with database marketing techniques than just RF scoring. Jim's book is literally the Bible on the subject. And here's the best part - you don't need expensive or sophisticated software in order to mine your own customer data.

In fact, all you really need is a simple spreadsheet and the tools and techniques given in Jim's book. Of all the marketing books I've purchased over my career, this is one of the few I find myself referencing again and again. It's a great investment for your small business.

I was quite happy to pin Jim down for a few moments in order to ask some questions. Enjoy the interview!

A Discussion With Author Jim Novo

In the 90's, Jim revolutionized the Home Shopping Network by focusing resources on the customer instead of the products. For the 00's, the Internet lies squarely in Jim's sights. The innovative yet simple customer analysis techniques found in the Drilling Down book puts two decades of High ROI Customer Relationship Management experience in the hands of today's interactive marketers.

Corte Swearingen: I remember purchasing your book Drilling Down in 2003. I had some basic familiarity with Recency and Frequency scoring (i.e., RF Scoring), but I was blown away by the depth and breadth of your presentation. What was the main motivation for writing the book?

Jim Novo: Most people think "behavioral modeling" is something you need to have special software for or be a genius to do, so they don't even think of how it could help them. Also, when they do decide predicting customer behavior could be a benefit to the business, people are often talked into buying all kinds of expensive "solutions" when they could make a huge leap forward by just using simple models they create themselves. The book was meant to bring these issues out in the open.

Corte Swearingen: What is your most important recommendation for a small business owner (with no experience in database marketing) to start their own data mining program?

Jim Novo: Most business owners I've met have at one time or another had a simple, powerful thought: I wonder why (insert best customer name) has not been in for awhile? People know intuitively that "not been in for awhile" can be a sign the customer is not coming back; they've seen this pattern. This is a powerful concept; the problem is there's usually no way to quantify or act on it; they lack a "system."

The simplest possible thing you can do to understand how important this idea might be to your business is this: pick 10 or 20 customers you consider to be "best customers." Make the effort to record each transaction you have with them - the date and amount - in a notebook or a spreadsheet. Then once a month or a quarter for a year, look at the data and compare for each customer how long it has been since they last interacted with you. I think you'll be surprised at what you see.

Corte Swearingen: In your book, you place a lot of importance on Recency - that more recent customers are more likely to act again soon. However, does this concept apply to all business types? If a customer purchases an expensive item, it seems that they might be less likely to make another purchase soon. In that case, spending marketing dollars promoting to them based on Recency might be a waste.

Jim Novo: Recency is a relative system - the question you need to ask is not "how long" but "how long relative to similar customers"; it's a comparison. So if your business is long cycle the model still works. If high ticket buyers purchase only once a year, then you should become concerned if an entire year goes by without a purchase.

Corte Swearingen: Never before have business owners been able to acquire so much information about their buyers and website visitors. From Google Analytics to inexpensive user testing and click-analysis, how does one zero in on the data that will most help grow profits without being completely overwhelmed?

Jim Novo: They should focus on the most important information relative to business goals. For example, "goal" reporting in Google Analytics is really simple to set up, and goal accomplishment by source of visitor is easy to report on. That's all you really need to start doing something significant to improve an online business - which ads, search phrases, web sites produce traffic that is likely to achieve your goals? Knowing that, it's simply a matter of allocating more resources to the traffic activities that have the best outcome, that generate visitors who achieve your goals. From there everything is fine tuning.

Corte Swearingen: Your book describes a very detailed method for scoring customers based on recency and frequency of purchases. This leads to a 2-digit score with "55" being the best customers and "11" being the least responsive. How do you recommend marketing to customers that score in the middle of the pack - say an RF score of around 35? Should business owners worry about defining specific RF boundaries when scoring?

Jim Novo: Tracking the behavior of 10 or 20 best customers as described above will give you an idea if this data is meaningful for your business. If you want to take action on the data - what can I do about this - then you need a "system," and RF scoring is such a system. How to use the system takes more space than we have here, but a detailed small business case is the Web Retailing Example found here: www.jimnovo.com/retail-recency. If your business has a "natural rhythm," like oil changes or hair cuts, there's a slightly different approach covered in the Hair Salon Example here: www.jimnovo.com/hair-salon.htm.

Corte Swearingen: Are there any good inexpensive software packages out there that can help small business owners automate their customer scoring process?

Jim Novo: Most don't need anything other than a spreadsheet or a notebook. If you have a lot of customer transaction data you can get a scoring package - RFM for Windows - here: http://www.dbmarketing.com.

Corte Swearingen: I have known some very small companies with only 200-300 customers in their database. With a database this small, is RF scoring still viable?

Jim Novo: Sure. You might use 3 segments instead of 5, or track Recency only, or simply pick a threshold like "We should call any customer we have not had contact with for over 3 months." The fundamental idea is the same: how long has it been since we have had contact, what is the likelihood we are losing this customer (Recency)? And then the action you take depends on the customer value (Frequency).

Corte Swearingen: Any recommendations on whether a B2B should perform RF scoring at the individual level versus rolling up a single RF score at the company level?

Jim Novo: It really depends on the nature of the business, you simply have to think through what you want to accomplish. I think most B2B models would operate at the contact or buyer level, unless you are dealing with "committees." A company score combined with a contact score could be useful at the Strategic level, it would tell you something like "Company A still buys from us, but it looks like they are going to stop buying Line B."

Corte Swearingen: As a long-time marketing professional, what are some of the interesting trends you have seen in data mining over the past few years? Will there always be a place in the marketer's arsenal for simple RF scoring or will it get eclipsed by something more sophisticated?

Jim Novo: There are certainly companies that go way beyond RF Scoring but these folks usually start with the simple models and prove the business value. Then they move up to regression and so forth. This path makes a ton of sense because the data generated by RF Scoring is highly valuable when constructing more advanced models, it makes developing the models faster with better end results.

Corte Swearingen: Any plans for a follow-up to "Drilling Down"?

Jim Novo: Not necessarily a followup, because I think (by the 3rd edition) I finally got that concept just right. But I have 3 or 4 related books I'd like to do if I can find the time.