on target

How to identify missing clients and visitors

What do a web site, a barber shop and a supermarket store have in common? Recurring visitors: no web site, and few business can survive without a constant flow of returning visitors. Find here some simple but effective formulas to identify missing visitors ASAP to prevent what is usually called churn or attrition.

Let's distill our rules from a plot of some sample cases:

You will immediately deduce first rule:

Rule # 1: visitor frequency depends on each visitor

In fact somebody goes to barber shop every 20 days, other people every two months.

Also, if you look at customer 1 you probably feel that something went wrong with him. His visit pattern was very precise and he missed next expected visit based on such pattern. You can't absolutely affirm the same with customer 2 because, based on his pattern, next expected visit still has to happen.

And what about customer 3? He actually started at low frequency like customer 2, but then he shortened the time between visits and, again, you would expect a visit that did not happen. So we can add two more rules:

Rule # 2: visitor frequency shall be updated at each visit.

Rule # 3: recent information shall have higher weight than older information.

Finally let's consider customer 4? He has a very irregular pattern and you can't make an exact prediction, but you can predict a certain range. This is in practice the most common case and you shall always think in terms of range instead of a single date.

Rule # 4: estimate error shall also be constantly updated and used to determine the range of next expected visit.

Using exponential smoothing you can quickly apply those rules each time a new visit occurs at date D, to compute T (expected time to next visit):

T = ALPHA × (D - D') + (1 - ALPHA) T'

Where D' is previous visit date and T' is previous value of T. Initially (when there is no previous T), you compute T as D - D'. ALPHA is smoothing constant that you choose. I suggest to start with 0.3 and possibly make some simulation to find its best value that shall be between 0 and 1.

You shall also constantly update the smoothed error:

E = ALPHA2 × ABS((D - D') - T') + (1 - ALPHA2) × E'

ALPHA2 is another smoothing constant you can adjust through simulation and E' is previous smoothed error.

Once you have T and E, here are three dates you shall consider if your visitor does not return:

• D1 = D + T + E: After this date something is possibly going wrong with your visitor.

• D2 = D + T + 2 × E: After this date something is probably going wrong with your visitor.

• D3 = D + T + 3 × E: After this date you probably lost your visitor.

Are you a management consultant?

Would like to try or coach this technique with your clients, do you need any additional or technical detail? Please let me know! Component Bases Solutions has great interest in partnership with business consultants. We can help you automate your proposed solutions in a very short time. We can also help to increase your visibility through links from many management tools we make freely available on the Web. Please contact for more detail.

Did you like this post?

Please, spend a few seconds and click on button below to socially bookmark it or tweet it. It will help to promote this blog and improve its content, Thank you!

Posted at 12:00AM Feb 09, 2010 by admin in Sales & Marketing  |  Comments[1]