Well, there's a shocker, thinks the owner, who was getting used to this kind of slap upside the head from the Recency metric by now. But it made sense. The areas with specific, targeted content had the lowest traffic but this traffic was on average more Recent - visitors to these areas didn't just "repeat", they were Recent Repeaters. The high traffic areas had relatively untargeted content so they drew a lot of activity but not a lot of loyalty; after a few visits that was it.
"How interesting" thinks the owner of IMissAsisa.com. Perhaps the high traffic / low loyalty areas are frequented mostly by new visitors and potential customers, where the low traffic / high loyalty areas are frequented primarily by current customers. Clearly there was an actionable idea in this chart, though it would take some more crunching with the visitor analysis tool to draw it out.
"Wait a minute", the owner thinks. "I have been tracking newsletter subscriptions by Content Group. You don't suppose..."
Well, fellow Driller, can you guess what the owner of IMissAsia.com is on to here? What will the data say? Here it is:
Content Groups, by Traffic, Recency, and Newsletter Conversion rate
You guessed it. The more Recent the Content Group, the higher the conversion of visits into newsletter subscriptions. The owner, once again slack-jawed by the power of the Recency metric, sums it all up:
"On average, the more Recent the visitor is, the more likely they are to subscribe to the newsletter, relative to other visitors.
The more Recent the newsletter subscriber is, the more likely they are to make a purchase, relative to other newsletter subscribers.
The more Recent the last purchase is, the more likely the buyer is to make another purchase, relative to other buyers."
The owner continues on, a bit breathless with the rush of all this stuff coming together at the same time.
"What I am seeing is that becoming a "best customer" on IMissAsia.com can be seen as a process, one that starts with a visit, moves on to a subscription, then a purchase, and hopefully multiple purchases. I always sort of knew that; now I see it in action. But what's really powerful is I can rank each member of a group - visitors, subscribers, buyers - against all the other members of their group for likelihood to move forward in the visitor - subscriber - buyer - multi-buyer process.
Knowing this provides me with three benefits:
1. Having the source of the visitor, subscriber, or buyer, I can "track backwards" and find out what sources (media type / offers) generate visitors most likely to become subscribers, buyers, and multi-buyers - and using Recency, predict which of those are most likely to complete all the steps of the LifeCycle process.
2. I can customize communications to the members of each group based on their likelihood to move forward in this LifeCycle using Recency. By addressing specific people with the right message at the right time (like I did with the discount ladder), I will generally get higher response and conversion of the visitor to multi-buyer while lowering my marketing and discount costs.
3. I can track retention and failure to progress in the LifeCycle with Recency, and be proactive about trying to "save" customers who are in the process of defecting. At any point in the visitor - subscriber - buyer - multi-buyer LifeCycle, I can track decreasing likelihood to progress and take special action with those who have high potential value or current value based on their source or past buying behavior."
It was late, and the owner was exhausted. No point in trying to map out all the implications of these discoveries now; this Recency thing was obviously quite powerful and it would take some time and testing to fully develop. For example, rather than determining "what's hot" just by visit volume, if I look at the Recency of visitors I can make a better guess on whether the issue is important to core customers or casual visitors, and adjust the message and offers appropriately. To think all of this came out of trying to answer one simple question on newsletter response.
Having now discovered the secret of the Recency Chain, the owner was confident IMissAsia.com could be taken to the next level of profitability. Yes, things are sweet, the owner thinks, as she turns off the light and "heads home" - down the stairs to her living room.
End of Recency: the Web Retailing Example
Jim's Note: The capability to track Recency (or Latency) has up until now been addressed with either a home-built system (all you need is a way to recognize users and store last visit date) or a data warehouse solution like WebTrends Intelligence Suite, CoreMetrics, Quadstone, several others.
What is new is the availability of Visitor History tracking without building it yourself or getting into a solution that has too much horsepower in other areas for what you need. In other words, it just got a lot cheaper and easier to track visitor history - and also to have these critical metrics integrated into the rest of your web site reporting.
WebTrends Reporting Center provides Visitor History capabilities in the Enterprise Edition, storing Recency and Latency data, plus many more "source" variables such as 1st campaign, 1st referrer, etc. Intellitracker out of the UK also provides reporting on Recency / Latency.
Bottom line: You've got no excuse now...
Q: As of today, I am in the early planning stages of an Operational database. This Operational database communicates with merchant terminals. Loyalty cards will be distributed to merchants' customers. The loyalty and gift card functions are not problematic, but the expense to track migration of customers at the Operational database level is questionable. With a marketing database created using the software from your book...
A: Ummm... I don't know how many customers you are talking about, or what kind of operational systems you have, but the scoring application that comes with the book is just a Microsoft Access application. My application doesn't "create a marketing database"; it runs on a database you have already created. You probably know that, but just to make sure...
Q: Can various reports be made on the Marketing database that track: who (of a particular segment) had a particular offer (specific to their buying behavior) and report the redemption or non-redeemed offers in a specific duration of time? Can I avoid this expense on an Operational database with a Marketing database?
Q: Well, you can report on anything that makes it into the database. As long as you can get "who the offer was made to" and "who responded" data into the Marketing database, yes, you can report on it, and use it to create offers.
A: The programmers of the Operational Database encourage using a Marketing Database for various reasons. First, the specific offers made to particular customers at a particular merchant will be communicated to the merchants' terminal using the Operations database. This protects merchants from customer fraud or abuse of repeat redemption. Second, they want to make more money. What do you recommend?
Q: I'm not sure I am following this completely, but I think the situation is this: your ops people don't want you running marketing stuff on their database, because they are afraid it will slow down performance, and that is bad for them. So they want all the coding, scoring, tracking, and development of offers to take place "offline". This is quite typical and standard procedure.
Usually the marketing database doesn't have to be "real
time" where the operational database does. So each night,
the operational database updates the marketing database with
transactions and you do whatever you want on the Marketing database -
analysis, scoring, creation of promo codes, and so forth - without
slowing down operations. You get
Now, if you want the ops database to be able to respond in real time based on customer scores and provide promotional codes or other data, what you do is send "customer codes" back to the ops database from the marketing database after you run your scoring.
For example, you get data from ops at 1 am, run scoring until 2 am, then send promotional codes back to the ops database at 3 am. So the next day, the operational database can respond to customers in real time based on scores and codes but it doesn't have to do any calculations because it has the scores and codes already. This dramatically reduces the "load" on the operational system, while still allowing it to be "smart" in real time.
Scoring of transactional data for real-time use in operations is a classic benefit of RFM, because the scores are "standardized" and are the same format each time. Since the scores rank likelihood to purchase and customer value, it is fairly easy to set up a rules-based system to make offers according to customer value and predicted reponse.
Am I understanding the question correctly? If not, feel free to ask again!
What would you like to
the book with Free customer scoring software at: Learn Customer
Marketing Models and Metrics (site article
Get the book with Free customer scoring software at:
Learn Customer Marketing Models and Metrics (site article list)