Know-How Exchange : Strategy

About 6 years ago I cancelled my Discover Card and started using Visa (in some countries Discover is not accepted) for several reasons including what I thought was terrible customer service.

Yesterday I got a mailing that said "We'd like to give you a new Discover Card". I know it's been 6 years but I would have said:
"We still can't believe you're gone!" At least I'd know I was different than the 6.5million people who got the same letter.

AOL still sends me postcards saying "You're still a valued customer". I haven't been one for years, but I notice the fact that they remember....yeah I know it's automated.

Why not try something along those lines? Something like: "Oh no! We just realized we're neglecting you! No, you don't want to have coffee with us weekly, but a Thanks for the business would be nice right?

THANKS FOR TRUSTING US WITH YOUR BUSINESS!!



Go to our website www.weforgotyouandweresorry.com and request a free back-to-school safety list. "

Something like that.

Hope that stirs the juices.

Michael

leepearce,

Of course you have to start with real segmentation of your customers.The most important thing is to find out which ones of them are or rather may be most profitable to you.The starting point is a bit queer in your case as you do only one-off sale and that's it.Anyway you have to start some place.Start with the analysis of what they bought with you as per brand,price, recency or frequencyif they bought more than once with you.
Then on you should start out Recency, Frequency and Monetary Scoreboard for your customers.
The retention plan can can be in place as soon as you know who your customers are, what they buy, how often, what they need and expect from your products.

Hope it helps

Adam

Hi

Your question, if the way you phrased it is correct, makes it a curious one indeed.

You have a large CRM database, but it consists of records in which there is one purchase entry! You acquire the customer, sell them something and then do nothing.

If that is the case, you have no business to retain, because you have never set out to acquire further business and therefore never failed to hold on to an existing customer. You will have lost no business, because you never invited anyone to purchase further products and suffered a rejection.

OK, some customers must have come back for more purchases, presumably through some marketing activity which is not related to the database. If these additional transactions have not been de-duplicated, then your data will consist of many records, some consisting of the same business, with one purchase transaction per record.

If they have been de-duplicated, then some records will have multiple transactions, but the repeat purchases will have little to do with the existence of the original record – it will have simply been the most convenient place to record further purchasing activity.

That is very unusual, but I suppose someone’s got to have done it!

If this is the case, RFM (Recency, Frequency and Monetary) analysis will be of little use, as the RFM codes will be of little meaning.

R will equate to the purchases unique customers made when they responded to your marketing initiatives. 5 will be the most recent 1/5 of all your records, 1 the last 1/5 of all your records. 5’s will be alive or still in their post, 1’s will be dead or have moved on. There is some use here, but not in a classical RFM context.

F will be meaningless in the context of your statement, as everyone will have made only one purchase, so frequency of purchase across all your records will have a spread of only 1, making the standard RFM coding useless. An F value of 5 would usually be appended to the top fifth (Quintile) of your records, regardless of when they purchased or how much they spent and a 1 to the bottom quintile. If the numerical purchase value was only 1 purchase per record for all records, then there is no behavioural difference between a 1 and a 5 as far as database marketing is concerned.

M value is of little meaning as it equates to the purchase the customer made when they chanced upon your marketing activity. If it was an expensive product, it will attract a value of 5 and if it was an inexpensive product, it will attract a value of 1. The M value might tell you that those records have a behavioural trait which displays a disposition to larger purchases, but it just as likely might indicate that the particular offer which precipitated the purchase that month was for a high value item which was good value for money at a time when there was a demand. For example, low cost air-con systems in the middle of a heat wave.

So, retention in the conventional sense is not the key and classical RFM analysis doesn’t mean what it usually does, where you can divide the database into 125 cells coded from 555 (Mail them and you’ll make a profit) to 111 (Consider throwing them away!)

What your R data will tell you is if the contact or the company is likely to still be around, so that’s a good starting point. F is of little use, but it does show you what the company purchased and coupled with the R value, what the company bought at the time. The M value is of almost identical use, in this case to the F value as you only have one transaction. The conclusion from this is that there is little you can utilise from behavioural analysis.

To write a customer retention plan, if your description is correct is therefore a bit pointless. I would think that you would be better off to recognise that you have a very useful data source where the most recent records are likely to be still of use for direct mailing and that the older records need qualifying through either mailing or telephoning.

I’m reasonably certain that if you havn’t done it, merging duplicate records will produce a multiple purchase history for a lot of records. Now you will have the basis of having some customers with repeat purchase histories and the chance to go after further business.

If you take the de-duped and merged database as a source of marketing data, a sample mailing to every 20th record will allow you to see what kind of response you should expect to see on the R values. R=5 should be 5 to 10 times more responsive to an offer than R=4 in a classic RFM quintile coding.

If that is the case, you can then start to go for the records where R= 5 or 4 and that through merging duplicates, your F value indicates that the company has made more than one purchase.

Then you can go for some of the larger M values and test those.

Once you’ve had the chance to see what the distributions on R and M look like from a test mailing, you will be able to start to infer what the purchasing relationship might be for small, medium and large companies.

This is a big job. What you are trying to do looks like a business database utilisation plan not a customer retention plan. From your stated question, you have a retention rate of zero because you’ve never sought to retain customers.

Once you start selling to your customer base via your database or offering repeat purchase products, then you’ll need a retention plan.


Steve Alker
Unimax Solutions

Whoops, I missed this post, I wrote a retention plan a year ago, if you email me, I'll dig it out for you. It might help a little (B2B Telecoms)