In identifying these three segments we have already crossed over the What and Why eras and now we’re definitely in the What If era because we are now going to have to do something to improve the profitability of these middle-tier people in particular because this is where we can grow revenue.

As we move into the ‘doing something’ mode we start to cross the bridge between what is becoming known as ‘CRM Analytics’ (the process we have just described) to CRM Operational – how we can set about trying to influence a person’s (customer or prospect) future behaviour, and here some examples might help.

  • Ø If we predict that a profitable customer will close his business with us, we must persuade them not to.
  • Ø If we predict that a person has the right profile to buy product ‘X’, we must offer the product to them to allow them the opportunity to purchase.
  • Ø If we predict that opening a shop in a particular place is likely to attract profitable customers, we need to tell the local population about it and open the shop.
  • Ø If we predict that an outbreak of flu is likely to happen in a specific part of the country, we must mobilise our field services AND warn the population what to do.

First, however, we must hypothesise what actions might alter specific predicted behaviours and we must remodel to see if we are correct. To do this we need to revert to our multidimensional and data mining environment to answer the following types of question:

  • Ø Will removing the annual fees persuade this customer not to close his account, and if we do, will this customer remain a profitable customer to us?
  • Ø Over what channel should I offer product ‘X’ to this customer? Direct mail is cheap but people like him often don’t respond – sending a rep is expensive but likely to succeed. Where is the balance?
  • Ø How do I attract the right customer base to my new shop? A paper advert is broad and will get the numbers, but will it attract the ‘right’ customers? Maybe a fashion magazine advertisement will bring the right clients, but it’s much more expensive.
  • Ø How do I warn the general public about an imminent epidemic? TV is easy but is expensive and really it’s only the ten- to fifteen-year-olds that are in any danger.

Answering these types of questions is necessary in order to be able to balance the cost of the operational CRM with its likely outcome in terms of revenue and margin. With no analytic capability it is always possible to advertise either to everyone or to no-one – the balance is to do something in between to justify cost with increased margin, but as I’ve said elsewhere, this is pretty difficult stuff.

Posted in Uncategorized | Leave a comment

So, What Needs to Be Done?


Sorry, It’s been a while since my last post but here in Sweden we’ve been busy preparing for winter!!

Referring back to my last entry, the first task is surely to segment a company’s customer base into two broad segments – those who are profitable today and those who are not. Using current techniques and tools, this process can be achieved in most industries today, and whilst profitability does not always mean revenue, a simple model of revenue minus cost may be sufficient to get started.

In order to identify the groups mentioned above we must now segment this same customer base according to what their profitability might be in the future. Now, the future probably needs to be at least five years hence, and in some cases it might be possible to derive real ‘lifetime value’ figures.

LTV is the total value that a customer provides to a company over the entire period that that person was a customer.

This task can be pretty complicated and lies in the realm of true Business Intelligence. Basically, data mining tools are used to build profitability models using past customer history, and then these models are used to score the current customer base to identify future value. In this art there are deep secrets and methodologies because getting this right can bring great rewards.

For many industries the distribution of customers into the three major behavioural segments lies in a common ratio. The high and low profitability customers will generally account for 20-25% of the total population, leaving around 75% or more that we might persuade to be just a little more profitable.

For the top tier – those that are profitable today and you predict always will be – you have to put together strategies to keep these people happy, as they are responsible for the major segment of your revenue. The good news is that it’s relatively easy to keep these people because they already like you, but the bad news is that if you lose just a few of them, your revenue will be dented big-time.

For the bottom segment, the segment Don Peppers refers to as ‘below zero’, we are not so worried. If these people stop being customers, margin actually improves a little, and as it’s likely that these people will then become customers of a competitor and burn their profits, we win two ways. If you can’t actively get rid of these customers, at least don’t spend any money on them!

If we look at the need to execute effective campaigns to grow the middle-tier customers (those that are today marginally profitable, but we believe can be made much more so) then we can see the dichotomy. To grow these discerning and high-risk customers we must behave in a much more personal manner. However, they are often vast in number so we have to put together offers that can be automatically tailored in a huge variety of ways. We need technology!


Posted in Uncategorized | Leave a comment

Segmenting on Value

We have discussed natural selection in a business context previously. Natural selection is driven by competition, and competition is being created by more companies selling more products to the same number of people living in deregulated economies with no geographic boundaries. In such a world it becomes much more productive to increase the value you extract from your existing customer base rather than attract new, unknown customers. To increase the value of a company’s customers, there are three avenues to pursue.

Your options are to:

  • Ø Keep the customer longer
  • Ø Sell a customer more products
  • Ø Sell to the customer in a more cost-effective manner (cheaper!)

Firstly, a company must try to keep its customers longer. If a customer spends $100 per year and it cost $100 to acquire him in the first place, then it is obviously important to keep him for a minimum of one year – every year he stays after the first represents opportunity for profit. Secondly, a company must sell more products or services to a customer to increase profitability. This seems incredibly obvious but is often overlooked. Selling a second product to a happy customer is usually fairly easy, and the more products a customer has, the more ‘loyal’ he is to the provider. Thirdly, a company can service a customer in a more cost-effective way and for this, customer preferences must be understood and a variety of technologies deployed, from call centres to internet commerce, as we shall see later.

However, all of these strategies are predicated on one thing, and that is the ability to understand who is a good customer in the first place, and this can be startlingly difficult. Broadly speaking, there are three types of customer: those that are always profitable, those that will never be profitable and those that are marginal. Whilst identifying these people based on current behaviour patterns is difficult enough, the intelligent business concentrates itself on what the future value of its customer base will be. This predictive modelling demands the help of technology. The business that can predict which of its current customers will be profitable over their lifetime (and the converse) is at a massive intellectual and operative advantage over anyone

Posted in Uncategorized | 2 Comments

More on marketing – I’ll look at the efffect of Social Media soon…..

The second driver alluded to previously is something that needs greater explanation. Since 1960 we have observed that the cost of computer processing has gone down by a factor of about one thousand every twenty years. This observation was predicted many years ago and is likely to hold correct well into this millennium, as I described earlier. Conversely, the cost of human endeavour has increased over that same period by some magnitude (don’t we all expect a pay rise every year?) and this has resulted in a shift of some 100,000 times in the ratio between the cost of computing and that of human resources over the time period since England last won a major football tournament (in 1966).

So, the race is on to replace human activity, including intelligence, which is not so easy, with automation. Hence the replacement of marketing staff with computer systems that can not only automate the creation of direct mail but also supposedly make intelligent decisions as to what should be marketed to whom.

So, marketing as we know it, and as we see it today, is set to make a remarkable change because of the competitive environment we are in and the sheer expense of any task that needs human intervention.

It is notable that as marketing becomes more important, we actually find fewer people doing it, and this drives even harder this tremendous need to automate. We also find that the nature of a marketing ‘campaign’ is changing rapidly. Not so far back, our big mail-order companies used to run four major campaigns per year – one each in winter, spring, summer and autumn. The catalogues that were sent each had hundreds of high-quality, glossy pages and were frightfully expensive to create, print and distribute. Today, whilst these big catalogues often remain, they are supplemented with other, more directed catalogues that are much smaller and cheaper to manage. This direction comes in two ways. Firstly, they are sent or directed to smaller segments of the customer base – those that are deemed more likely to buy; and secondly, the catalogue contents are subject-oriented. They are related and designed to hold the interest of specific ‘profiles’, or types of people.

Perhaps the major change, however, is simply the number of campaigns that we might witness in the future, their recursive nature and just how they are delivered to us. As we saw above, years ago, there tended to be fewer but larger campaigns which were sent uniformly to all customers without discrimination and, of course, with very poor response rates. Even today, a run-of-the-mill direct marketing campaign for a new product may expect to achieve one to three percent response rate (one to three people per hundred sent the mailing might respond with either a product purchase or further enquiry).

As we move into the future, response rates of single digits will not be acceptable, so the race is on for higher productivity.


Posted in Uncategorized | Leave a comment

I suddenly had lots of interest!!

Loads of people viewed my blog yesterday. thanks. I’m gona move on and talk some more about Big Data. The subject concerns me in the way it is being hyped up big time. Is it real? Well maybe it is but we must remember the lessons learned from Data Warehousing. This new stuff is not new at all but maybe needs a little more clear thinking…………

Posted in Uncategorized | Leave a comment

Analytical Platforms


In my step through contemporary architectures (platforms at least) I have omitted the current fad for ‘analytical platforms’ and the arguments for and against column based DBMS’s.

I will come back to these shortly but for now the ‘discussion’ between relational and columnar platforms has become ridiculous and is not even particularly relevant and the analytical platform is shaping up to be an independent data mart with all the pros and cons I have already exposed. More later.


 ‘Modern Marketing’

I’m sure all of us have noticed the way marketing invades our lives more than ever before. Some years ago, the term ‘junk mail’ was invented, but now junk mail has progressed from invention to acceptance and now pervades our very lives. For myself, it is common to be away from home for weeks at a time, and it is never so clear just how much junk mail finds its way into our homes as when, on returning from a few days’ business trip, I see a pile of mail and find that less than one in eight items need attention. The rest, the junk, is luckily not only very easy to identify (‘this is not a circular’, your name and address spelled incorrectly), but just as easily tossed into the bin.

There are, by and large, two reasons why we get so much junk mail today (and by the way, it is set to increase over the next few years). Firstly, we are now in a world of incredible commercial competition. In every major European country, we have too many suppliers of too many products for a population which is by and large fairly static. In such an environment it is very important to use every mechanism (channel) available to make offers of products or services to a wide market, and until the total acceptance of the web (or rather the acceptance of performing financial transactions over the web), direct mail is probably the cheapest method for most types of targeted marketing/selling.

Posted in Uncategorized | Leave a comment

BIG Data -Problems with Contemporary Solutions


Note, very soon I’m going to take a diversion in this blog to discuss Big Data………………so bear with me whilst I sort my feelings out!!

Firstly, it’s only right to say that Data Marts and Data Warehouses have often brought great benefits to those companies that have deployed them well. Some companies have been smart and gained real benefits quickly, but others have simply got bogged down in a morass of complexity and politics and simply wasted huge amounts of time and money. Failure comes in many flavours and it won’t do anyone any good for me to list all the pitfalls here, so when I talk about ‘problems’, I want to be specific and remain relevant to the theme of this blog

If we examine the biggest problem that confronts BI, it is the fact that the base data of a company is invariably spread across many systems, and is subject to many movements and relocations, subject to many modifications in different systems, and subject to very little validation.

The greatest driver in IT architecture must therefore surely be to remove all of this complexity, and drive to a point where there is no copy management of data because it is simply not needed, and it’s this principle that convinces me that our vision of the Data Warehouse must change. Today, in order to solve the problems caused by inaccurate data, we actually add a further requirement for copy management as data enters the Data Warehouse. In effect we take lots of unorganised data and make a further copy of it, change it, and summarise it until it often becomes unrecognisable and certainly irreconcilable back to its original.

Now, there are no clear alternatives to what I have outlined above, but it does seem incredible that in the 21st century we are still solving problems caused by one flawed strategy by adding a further requirement for that same flawed strategy.


Posted in Uncategorized | Leave a comment

Two Things to Get Right


    Over twenty years of building Data Warehouses we have learned a lot, and there are many published documents on best practices and pitfalls to avoid. I just want to say a few things on two particular factors that are critical.


    Business Involvement

    We will examine the criticality of business in detail in an overall Business Intelligence project later, but suffice it here to say that it is vital. Any Data Warehouse project designed by IT people with no business involvement is almost destined to fail. Business people need to be involved in nearly all aspects of the project because only then will they become enthused with the Data Warehouse and find innovative ways in which to use it.


    Scalable Platform

    The underlying ‘platform’ for the Data Warehouse is a computer and a relational database, and the way these work in combination determines scalability. A key methodology employed in building Data Warehouses is the ‘think big, start small’ concept, which means that you start with a well-scoped project that will take, say, four months to complete, and then build it out to fulfil other business requirements in an incremental, project-by-project way. This is absolutely the right approach and is one of the reasons that I advocate TNF data models and relational databases. I also advocate hardware platforms that can be grown, so as the Data Warehouse grows, the computer and database can handle the growth with ease by just adding  hardware and software without changing anything, and anything includes:

    •  Physical Database design
    •  Already deployed applications
    •  Already working ETL logic
    •  Existing data layout
    •  Business understanding of the data

    When we think of scalability, we commonly think about the ability of the platform to cope with an increasing amount of data, but that is only one facet of growth as it applies to our Data Warehouse environment. In fact, you must consider an array of dimensions that will grow if the project is successful, and all will put heavy demands on the platform, which must therefore be able to grow seamlessly.

    Such dimensions will include:

    •  Volume of data
    •  Number of concurrent users
    •  Number of concurrent queries
    •  Complexity of queries
    •  The need for very recent data
    •  Mixed workload
    •  24/7 operation
Posted in Uncategorized | Leave a comment

Key Business Opportunities for the Data Warehouse

There are many Business Opportunities that can be enhanced by using a Data Warehouse such as:

  •  Revenue accounting
  •  Revenue assurance
  •  Yield management
  •  Operational costing
  •  Maintenance and engineering
  •  Fraud tracking
  •  Product packaging
  •  Parcel tracking
  •  Pricing and re-tariffing
  •  Load balancing
  •  Financial reporting       
  •  Tracking    
  •  Demand planning
  •  Location planning         


Perhaps my favourite is analysing the supply chain. Billions of dollars are spent every year in automating and understanding supply chain dynamics, but still the consumer – the one at the end of the chain – is subjected to goods out of stock, planes and hotel rooms that are full and out-of-date products on the shelves. Why is this? Well there are many reasons, not the least of which is that a full supply chain for any product on the market today is a very complex ‘thing’. Consider that when you eat an olive bought from Sainsbury’s today, the whole chain that enables you to have that pleasure possibly started years ago with someone negotiating with an olive grower in Greece. Supply chains are nearly always very complex, often involving many companies, often reaching over many countries, and are often long-winded in terms of time. This complexity ensures that the data needed to smooth the workings of the chain’s integral parts is distributed in many systems and cannot be shared in any reasonable way. The result? Guesswork as to what should be planned, what should be manufactured, when it should be distributed, where, how and who will buy it at what price and when. You can see why supply chain management and analysis is a key application for a Data Warehouse.

Posted in Uncategorized | Leave a comment

What Is the Relationship Between Data Marts and the Data Warehouse?

There are five possibilities at least worthy of discussion:

Possibility One

There is no Data Warehouse and just a single Data Mart: in this case, the Data Mart is really a Data Warehouse – it may be small or it may be poorly defined but there is no clear difference between the two concepts at this level.

Possibility Two

There is no Data Warehouse but more than one Data Mart: this type of deployment is a mess and needs sorting out as a matter of priority. Such Data Marts are called ‘independent’ but should be called ‘unreliable’ instead.

Possibility Three

There is a Data Warehouse and one or many Data Marts which get all of their data from the Data Warehouse: this is a sound basic architecture and in this case the Data Marts are termed ‘dependent’ as they are dependent on the Data Warehouse for all of their data.

Possibility Four

There is a Data Warehouse and one or several Data Marts, but the Marts get data not only from the Warehouse but from other systems as well: this is the worst case scenario and disaster is looming.

Possibility Five

There is a Data Warehouse and no Data Marts: the preferred solution architecture, enabling the actual ‘single version of the truth’ – well, almost.



Classic DW Application Areas

Key to success with the Data Warehouse is selecting the right application areas and deploying applications (reports) in an orderly fashion against a pre-defined roadmap. Data Warehouse techniques and capabilities favour some types of application above others, with ideal candidates sharing some of the following characteristics:

  • Ø They need detailed, historic data.
  • Ø They need data from different systems.
  • Ø They process data in complex hierarchies.
  • Ø They tend to look for trends and correlations rather than run short, simple queries.
  • Ø They can live with data that is not current.
  • Ø They do not calculate complex figures where 100% accuracy is needed.
  • Ø The reports created are unique  across the enterprise.
  • Ø Requirements are always changing.


Bearing in mind the list above, perhaps the most common use of the Data Warehouse is in marketing, because marketing is a process that is notoriously difficult to measure in terms of performance. Traditionally, campaigns are created and executed and if there happens to be an increase in sales at around the same time, then the campaign is deemed a success. My guess is that 90% of all marketing campaigns are designed with no scientific or factual input. I also guess that 90% of all responses to all campaigns cannot be tracked back to the responsible campaign, and that 90% of all campaigns have minimal effect on a company’s bottom line. This is pretty depressing bearing in mind all the data that is available to the marketers, and the impressive tools and huge computers at their disposal. The use of a simple Data Warehouse has the potential to improve the efficiency of marketing many times over.

Posted in Uncategorized | Leave a comment