I really must be careful how I phrase this. Because I know that, if I position the sentiment too boldly, I am going to pay for it further down the line.

It is this: that we try to be very careful about just how many analytical papers we publish in each issue of the Journal of Database Marketing and Customer Strategy Management. Its not that we have anything against analysis. Far from it: we believe passionately that good, effective marketing starts from a basis of information and understanding. Nor is it anything personal. As current editor, I first began my association with the stable of journals, of which Database Marketing is one, when I established the Journal of Targeting, Measurement and Analysis for Marketing.

I have worked, off and on, as a data analyst: and I would like to think that my work has, on occasion, produced some moments of insight within a company that have, in turn, given rise to real strategic change. Like everyone who contributes to the commercial planning process, it would be nice to feel that one had made a difference.

Even so, what work I have done has always been fraught with difficulty. For instance: in one financial institution, I developed a multi-media response model. The principle was straightforward: most market measurement to date had taken the weight of promotion put out by channel, and calculated some sort of response rate as a function of that medium.

Response rates for direct mail and direct response press, or television, might be different by several orders of magnitude: but they were measurable. For the size of budget being spent, however, that was far from enough. The Marketing Director wanted to know what happened when television ran at the same time as Direct Mail. Or what about Press and television and Direct Mail.

Presumably, he reasoned, the Direct Mail response would go up. Which it did. Sometimes. Apart from the occasions when running two promotions in tandem led to a drop in response to the first.

Our initial response was that that felt wrong. Then we thought about it a bit more, and realised that it wasn’t wrong at all. If you imagine the total market as having an upper limit to buying capacity, then the real impact of putting out promotions through multiple channels was to capture a greater and greater proportion of the total demand: but the proportion of demand captured by each medium is likely to grow slightly, or even to fall, as demand is picked up by another medium.

When looked at in that way, the result was not too hard to understand at all. The next steps should have been the creation of some form of trade-off model that allowed the Marketing Director to play ‘what if?’ games with various combinations of the media being considered.

With a little clever algorithm-building, it might even have been possible to ‘solve’ the model, and to come up with the optimum combination of media to return maximum responses for minimum costs. Or it might have been possible to present alternatives: one media schedule would maximise volume; another would minimise cost; and so on.

I say ‘should’ and ‘might’, because the entire process foundered at the first hurdle. The Marketing Director did not like the results of a model that appeared to show that average response fell when he increased his spend on television. He had an agenda — which involved increasing the use of broadband media (don’t ask!) — and the news from the analytics front was completely at odds with what he wanted to do.

Well: it wasn’t. If he had read the small print, he might have understood that something complex was going on, and that he needed a complex model to capture it. Not only that: he also had at hand an analytical team that was more than capable of delivering. Sadly, none of this was to be. He rejected the initial findings. By the time we attempted to explain what was actually going on, he was in no mood to hear. In the end, the research was shelved, and the Company increased its spend on television significantly.

Let's look at another example of complexity. Deaths from prostate cancer have increased greatly in recent years. That sounds like very bad news for men: is something in the environment or in the way we live now making it more likely that an individual will ‘catch’ this disease?

The answer is yes: and no. Because the main thing that has changed is the lifespan of individuals. Prostate cancer is an illness that takes a very long time to develop and to finish you off. So in an era when the average lifespan was short, the chances of this particular illness actually turning up as cause of death on your Death Certificate was very slight indeed. Only as life spans increase does the likelihood of someone living long enough for this illness to mature become reality.

In a sense, therefore, the reason for increased mortality due to this cancer is nothing more nor less than better health. It is what is loosely referred to as an actuarial effect: something that takes time to emerge and often may be masked by other effects that hit sooner.

One side effect of a motor trade loyalty programme was, perversely, an increase in complaints. The direct effect of the programme was to increase the likelihood that individuals would either trade in their existing model for a similar one — or retain it for longer.

The effect of Loyalty was to create a pool of drivers who held on to their car for longer than they would normally: and since the cars were getting older, they ended up reporting more faults and, inevitably, complained more.

Look around your own business, and the chances are that it will not be long before you encounter one or more examples of this type: where a successful achievement appears to bring with it a downside; and where it takes relatively complex analysis to dig into the causes of that downside.

Surely, therefore, we should be encouraging analysis through the Journal? That would be my first thought, too. However, bear in mind the reaction, all those years ago, of a Marketing Director to complex analysis. It was rejected. There is a demand through much business for simplicity and straightforward messages. Not all analysis is capable of fitting into these categories.

So we sift the analytical papers we receive carefully. If the proportion of Greek algebra to text starts to look daunting, it is probably not for our readership: it is a paper about statistical technique. Perfectly valid as it is — but unlikely to be widely understood.

Equally, we are cautious of papers that carry out a small piece of research on the basis of a specific statistical technique. Science — and academic theory — advances by slow, incremental steps. Marketing cannot afford to. The pressure of this year's commercial plan is here now: and next year's is not far behind.

Papers that are too ‘small’ in this sense, tend to be put to one side. So, too, are papers based on samples and research that is too select or too focused. We are rightly scathing when the News Headlines are based on a piece of eminent research carried out on no more than a handful of individuals. Yet the same issue is at work in marketing. Entire PhDs are based around very small sample investigations, and if we are not careful, the findings from these small samples can all too quickly enter the marketing firmament as established wisdom — and be very difficult to shift thereafter.

All of which is a very long journey to bring us back to the main point: that information and insight are worth cultivating. They should, however, be cultivated carefully. Marketing is not the place for rocket science — statistical models that require a degree in maths to audit, let alone validate: nor for work in which the ratio of reality to noise is too small.

As with every other part of Marketing, a balance needs to be found. Analytics need to be fostered and developed, and good lines of communication need to be established between the specialists and the creatives. Analysis should be respected. It should also be intelligible.

With this in mind, this issue is slightly more of a mixed bag than usual. It starts with a useful insight piece by Koslowsky into how you can make use of analysis.

It is then followed by four papers, three of which, in their way, tackle a particular subject from an analytical point of view. Gunnarsson et al. look at how data mining has been used in the newspaper industry. D'Souza et al. apply tree induction to pet insurance. Añaña and Nique look at the role of personal values and their influence on consumer perceptions.

If you work in the newspaper industry, or pet insurance, then it may be that one or other of these papers relates instantly to an issue you have been poring over. That is unlikely. Their role, we hope, is to illustrate the type of questions that may be asked through analysis, as well as the sort of analytical techniques available — and to encourage marketers to make more creative use of analysis in future.

Meanwhile, we will continue to seek out and publish analytical papers whenever they have something interesting to say — or whenever they can help you to understand better how to do analysis within your own organisation.