WebProNews

Recency: Visitor Momentum and Friction

By the way, think about that Equilibrium area in the second graph above (Revisit Index, Figure 12). This area is where the ratio of people who visited again to people who did not visit again is dead even. Does this metric remind you of anything, perhaps the Latency metric? I mean, if you were going to set a trip wire somewhere, setting it where the average visitor is as equally likely to visit again as not can't be a bad choice. This is how Recency and Latency are related; Recency works on a "scale," a relative comparison of behavior. Latency is an absolute cut-off, looking to find the average or break-even behavior and use this point as a trigger for action.

As a matter of fact, you can use Latency and Recency together to track rising or falling Friction. Think about this scenario. The average Latency of visitors to a web site is 2.5 days, meaning on average, there are 2.5 days between visits. Sounds pretty good, right? Now, what if I tell you that these same visitors have an average Recency of 8.5 days - the last visit was 8.5 days ago. Think about it.

If on average they come back every 2.5 days, but the last visit date averages 8.5 days ago, what do you have? Well, fellow Driller, you have increasing Friction, of course. The Latency pattern has been broken, the wire tripped. These visitors are simply not (on average) visiting as often as they used to, and in fact are falling back on at a fairly alarming rate. They are having trouble, not finding what they need for info, not moving towards conversion. Likewise, if visitors had an average Latency of 7.0 days, but these same visitors also had an average Recency of 3.4 days, what would that mean? Of course. Friction is falling, the visit rate is increasing, these visitors are indicating an acceleration of the relationship, they are finding what they need, increasing their engagement. Now, what if you could see this data broken out by campaign, or search phrase?

(click the link below)

http://www.drillingdownbook.com/images/recency-latency.jpg

Visitor Friction by Search Phrase

Glad to oblige, my fellow Driller. This is a WebTrends report looking at visitors by Organic (not pay-per-click) Search Engine source with Average Recency and Average Latency data by Search Phrase. The report is tracking visitors by the first (Initial) search engine they used to reach the web site. The search phrases are hidden by a light blue color to protect the identity of the client:

What can you do with this piece of intelligence? Now you are talking about the Friction of visitors rising and falling by campaign source, and now you are into predicting the likelihood of a campaign conversion. Not just conversion today, but likelihood to convert tomorrow. Given any two ad campaigns, e-mail drops, search phrases, etc. with the same initial conversion rate, and faced with where to put the money on the next round, you would bet more money on the campaign generating visitors with falling Friction, because they are the most likely to convert in the future, and they have higher potential value. In the case of Organic Search in the example above, you would want to take a look at the pages these visitors are initially landing on and figure out why certain pages tend to increase Friction and certain pages tend to decrease Friction.

Capiche?

Think about this. Latency, or Trip Wire metrics tell you when something bad or good has already happened. Using Recency, you can predict the likelihood of something bad or good to happen. There is a huge difference between these two metrics, my fellow Driller. Customers who are more Recent have higher potential value than customers who are less Recent, for any given activity. Customers who made a purchase 15 days ago have higher potential value than customers who made a purchase 60 days ago. Customers who logged in last week are much more likely to visit than customers who logged in 30 days ago, and have higher potential value.

"Now hold on just a minute, Jim," you say. "Recency is a very cool concept, but I can think of some specific instances where it can't possibly work. A person who just filed a tax return 30 days ago is not more likely to file one than a person who filed one 60 days ago, and the same thing is true for people who bought a new car. Explain yourself!"

There are two issues to consider when using Recency - external forces and time frame. If there are powerful external forces shaping behavior - like the April 15th tax deadline - these forces may overcome the Recency effect. An accountant trying to manage customer relationships would probably look more to Latency and set a trip wire: I will call best customers who don't schedule an appointment by March 15, for example. The tax deadline is simply too powerful a force and overcomes normal human behavior.

One also needs to consider Recency in light of the cycle of normal behavior. It is unrealistic to think of Recency in new car buying in terms of 30 and 60-day periods, when the normal purchase cycle may be 3 or 4 years long. It's not a rational use of the Recency metric. However, for the dealer selling the original car to the customer, as this purchase gets to be 3 or 4 years old, the longer it has been since the purchase, the less likely the customer is to make the car purchase from this same dealer.

Recency is a very powerful metric, but there are times when it simply is not appropriate to use without some adjustments. If there are powerful cycles acting on behavior, Recency often takes a back seat to Latency. Often the two concepts can be used together - there is first a Latency trip wire and then Recency kicks in. For example, up until the April 15th deadline, the accountant is really operating in the world of Latency. If customers don't call by a certain day, they are unlikely to be using the accountant for their tax return. Once the April 15 deadline passes though, the accountant is in the Land of Recency - the longer it has been since the last tax filling, the less likely it is the customer will be using the same accountant next year. The accountant needs to get on the phone with these high value customers and find out what happened right away if the customer is to be recaptured.

The new car dealer is in a similar situation. Let's say the average customer trades in every four years. Up to four years after the new car purchase, the dealer is in the Latency world - there is a trip wire at 4 years, and any customer who has not purchased again at the 4-year point is in danger of being lost. After the 4-year point passes, the more time passing, the less likely it is the customer will come back - the Recency effect. As time goes by after the trip wire triggers, it becomes more and more urgent the customer be contacted and made an offer. And, the more time that passes, the higher the offer will have to be to get the customer to come back. This phenomenon is called the Discount Ladder and you can see an example of it here:

http://www.drillingdownbook.com/Recency-Discount.htm

OK, that's it for a description of two basic friction metrics - Latency and Recency - that you can use to rank the potential value of visitors and customers. Now we'll go into a specific Real World Example of using each.

Next Article: Latency: The B2B Software Example

This article is taken from the book Drilling Down: Turning Customer Data into Profits with a Spreadsheet. The first article in this series can be found here.

Jim Novo has nearly 20 years of experience using customer data to increase profits. He is co-author of The Guide to Web Analytics and author of Drilling Down:Turning Customer Data into Profits with a Spreadsheet. If you want more visitors to take action on your web site, try using the free conversion metrics calculator, downloadable here. If you need to sell more to customers while reducing marketing expenses, get the first nine chapters of the Drilling Down book free at http://www.drillingdownbook.com.
Ask Jim a Traffic Analysis Question!