Generally, the answer is yes, the viability of a prospect falls as time since first contact grows. But it depends a lot on product, the more complex the purchase decision, the longer a prospect can remain viable.

A technique you can use to help determine the answer for your business is to look at where closings are coming from in the prospect Lifecycle. Example: if 80% of closings come from prospects less than 100 days old, and 1% come form prospects over 400 days old, then the answer to your question above is yes.

You can probably tease these numbers out of your lead management system, even if you don’t have a “report” that makes it easy to do.

Found the concept of recency very interesting. I also want to know if the “age” of an “opportunity” in a sales pipeline of a typical B2B company can also indicate the willingness to buy. Has there been any research in this area in the past? In other words, if my pipeline has 40% of open opportunities more than 400 days old, is this pipeline necessarily better or worse than another pipeline where say, most of the opportunities are less than 100 days old? In other words, does the quality of an opportunity deteriorate with its age?

You would realize that my question focuses on opportunities with customers, rather than customers themselves…

Regards,
Vivek

It is very inspiring, our company provides free service to our visitors, but we may apply the LifeCycle to our clients, which is we have been thinking of for a long time but no idea how shall we start. Many thanks for the comments and articles, have a nice day!

Best,

Echo

Drop a promotion tagged by Recency across the population and you will see response that looks much like the curves in the first chart above. What you need to find is where profit is maximized along that curve, and then as each customer reaches that point, drop the specific offer that maximizes profit.

This is in fact a “series” of emails based on LifeCycle; those that respond pop back to the top of the curve and go through the cycle again.

In general, for commerce, this means a “ladder” of discounts where the discounts get higher as people fall down the curve / are less likely to respond. You pay for this margin hit by reducing discounts to people at the top of the curve who are highly likely to buy anyway.

See “Discount Ladder” for example of how to test this:

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

This is one example of how to customize message by which cell of the grid people are in. You can also customize message using suggestions by cell color above. Doing both at the same time I find most effective.

The net effect of above is maintaining margin while driving higher response in the lower half of the curve.

Agreed with you, I am also using 90 days. I thought my company customers are following the 20-80 rule, just a bit depressed to know it turned out to be 10-90 only, or maybe I should not say it 10-90, as most of them are new customers and existing customers kept drifting away from our site. My orange cells got big count, it seems there should be some action taken to retain those customers, maybe can send eDM, but the view and click rate is really low..

Regards,
Echo

Also, there’s nothing “magical” about the boundaries used for the colored areas; I used 90 days because it’s a common breakpoint for activity (over 90 days with no activity, people become much less responsive). If you have a longer activity cycle business that boundary might move out in time (B2B) if you have a shorter activity cycle it probably should move to 60 days or even 30 days (social network).

Testing tells you where the boundary is most actionable; look for steep drop in response by Days since Last Action.

Many thanks for sharing this segmentation analysis, I have done a similar analysis following your method and the result is very surprising, my green cells is really a small portion!

Hi Brian,
May I know what model are you applying? Are you taking the 4 segments as a category variable and together with other variables to fit a logistic or survival model?

longing for having your feedback, thanks in advance!

Regards,
Echo

Thanks…. your research looks thorough and well done. This will aid me greatly in the next coming months in training some new staffers and taking some to the next level. Keep it up.

“The model parameter estimates are fed into a dynamic programming
model that determines the optimal number, sequence, and timing of promotions to maximize retailer profits.”

“For our application, a discrete time formulation is more appropriate because it allows us to explicitly account for marketing activity in periods when households do not make a purchase.”

“We use a discrete-choice framework with time-varying coefficients to capture the duration dependence in the consumer’s purchase decision and promotion effectiveness.”

Then there is a whole section on trying 20 variations and looking for the best fit for predicting purchase *and* non-purchase. This bit about predicting non-purchase seems to be quite import to the researchers and I’m guessing this is where they are making some kind of academic stand versus prior art.

Does not sound like logistic regression to me – that I can understand!