Recency analysis on the web channel

Posted: January 2006 C: Customer Intelligence Posted by: K. Panayotakis

Recency analysis on a web site which is aiming to attract and retain repeat visitors, can prove to be very valuable. The visitor lifecycle on the web may be very short; the opportunity window to attract & retain a new visitor, is very limited.

The analysis of a recency diagram can yield valuable insight on the performance and the business prospect of an e-commerce site. (a relevant analysis can be found at RFM).

In order to monitor recency, an e-commerce site needs to answer certain questions (or provide definitions which allow measurement): Who is a web site ‘customer’ ? The definition may vary: one who has a cookie on his PC or is registered or is registered and receives a newsletter, or has made orders on the web. However the existence of a cookie, does not necessarily identify a single person: the PC may be used by a household or may be a shared PC. When is a party no longer a Customer ? For example: one who has not visited the site for 3 weeks (the interval may depend on the specific case, the rate of content refresh or the period of newsletter for a blog). Which stages comprise the sales funnel? An e-commerce site typically has the following stages: web visit – registration – shopping cart use - order. A blog which is not selling directly, has an equivalent funnel, the last stage of which may be a newsletter registration.

An interesting analysis would be to compare the sales funnel velocity vs the recency of a customer. It is expected for a more recent customer to move faster through the sales funnel.

Which event type should be used to measure recency on the web? It depends on the type of the web site: • Recency on a content portal (e.g. a blog) shall be measured on visits: the date of the last visit • Recency on a ecommerce site shall probably be measured on orders: the date of the last orders. This metric is expected to have more predictive power than the one based on visits.

The ability to identify the visitor, enables the execution of any recency analysis. This identification is usually based on cookies placed on the PC or on user registration (some analysis on this can be found here). However, certain users disallow the placement of cookies (estimated at 5%). Though this fact reduces the degree of accuracy of certain recency metrics, the recency analysis value is still high.

In an Enterprise which operates multiple customer touch points (CTPs), measuring recency requires monitoring customer interactions on all CTPs. The selection of the appropriate recency metric should be based on tests. The metric with the highest predictive power should be used.