For years and years there has been a drive to make direct marketing more scientific, by insisting that everything should be measurable.  This approach took hold when direct marketing started, and has become even more intense with the advent of email.

The notion is that you might write a mailshot or an email and then perhaps vary it slightly.   Version A and Version B are sent out to randomly selected people on the same day at the same time, and the result compared.  If there is a significant difference between the two versions, then you can put that down to the change you made to the copy.

That approach is of course very valid, and I have used it myself for years.

But it can’t ever tell us everything, because there are so many variations to be tested.

Because of this I believe we also need a second approach – an approach that allows the creative team to fly with different ideas, without the need to test each and every one of them.

By way of example, let me relate what I did with a series of adverts for a client earlier this year.

First, I came up with a general idea of how the advert should work, and sent that out in week one.   In week two, approaching a second product, I changed the way in which I developed the advert.  And again in week three and so on.

This of course is completely unscientific.  But it is still of interest.

Obviously I am mailing on different days, advertising different products and working in different ways.  All that stays the same is the mailing list.  But I have a general feel as to what sort of response rate I might get.

In working through these adverts I gradually modified my approach, until eventually by the sixth advert I hit a much higher response rate than one would normally expect.

Still nothing scientific about it, but it was at this point that I came back to the serious testing mode.  I looked at the results, and constructed the theory that if I changed one element within the advert, the response rate would go down.  If I put it back in, the response rate would go up.  Over the next couple of ads I did just this.  Not scientific in the sense there was no split list, but a good scientific test. because there was a hypothesis that I was testing out.

The results were as I predicted, and now we have our model – our way of writing adverts for this particular product.

The approach has taken up the response rate by 400% – something I could never have achieved if I had simply done the traditional split testing.

None of which is to say that the split testing approach is wrong – as I say, I use it myself.  But it should never be allowed to sweep aside some good old fashioned “I wonder what would happen if we did this…”

That is, after all, how I managed to created the Toppled Bollard series of stories – probably the most successful campaign I have ever produced.  (www.blog.toppled.info)

Tony Attwood

PS I am now going on holiday – but my colleagues in the office will be pleased to help with all questions and enquiries.   01536 399 000 usually works.