Big Data Overload – Is Research The Answer?

by Michael on 9 April, 2012

In the good old days, we used to test everything that moved, measure responses and purchases, analyse the results, produce lifetime value calculations for customers, rank media and  lists according to actual results and just got on with developing the business. We largely ignored post campaign research, as we had hard facts, and we had those more quickly than any research could deliver results. So we may have used pre-campaign research in focus groups, but we believed in long copy, benefits, calls to action, testing and experience – because they worked. We had a control, and tested to beat it.

I remember mailings for a large credit card company 30 years ago where we’d mail 2.5 million people and have over 250 test cells for lists, copy, design. And of course life was a lot simpler, as online didn’t exist, things moved more slowly, customers were more loyal for longer (if you got it right), and we understood what we were doing both technically and in a marketing sense.

Now, we have big data and data overload, and companies simply don’t know what to do with a lot of the data they are generating. Analytics are clearly more difficult the more channels there are and the more they interact, but I think that a lot of the problem is because a lot of the new channels have developed in silos and the people working in them don’t understand either integration or marketing, nor do they test or even know how to test.

I appreciate that testing integrating TV and door to door by omitting door to door from TV fringe areas, so creating control groups, is a much simpler concept from over 30 years ago when we had TV fringe areas (I’ll happily explain to anybody who is interested). And that for example when I went on a cutting edge database marketing course for 5 days in 1985, it covered far fewer channels, but in much more depth, and was still cutting edge 15 years later with minor updates.

Now, we have a much more complicated landscape, and far more data, some of it real time, and perversely, far less testing. We also have much less experience and knowledge in the business, and most people live in silos, without clear goals at times. I cringe when I hear some digital marketing specialists talk about testing as if they’d just invented it. They’d have been laughed off a stage 30 years ago for being too simplistic – but we will get there as marketing is introduced to new techniques, along with true data understanding and analytics.

But that will take a while and we’ll still have a huge volume of data to analyse, and it all takes time. And we live in a fast moving world where we want to understand today and not in 3 month’s time. So we need immediate results analysis and the understanding and insight that goes with that. Which brings me back to research, and specifically to on-line research. This isn’t a research treatise, so I’m not going to discuss the relative strength of panels, enquirer and buyer questionnaires, except to say there are some very large on-line panels, you can build your own, and it is simple to research amongst on-line buyers and responders, and to look at the impact of off-line activity too.

So whilst we are waiting to find out what people have done, why they have done it and what they are likely to do next, and then discovering our analysis is out of date, I’m coming round to believing that timely online research is a better option if done properly.

One of the figures drummed into my brain 30 years ago was 95% significance, and that was simply saying that with the right sample size and predicted response, you could be 95% sure of the outcome. I suspect that some online panels aren’t measuring up to this, but that’s a separate issue.

I’ve been aware of the online research done by Fastmap for some time now, and I’m impressed with what they do in our area of business, but here we are talking about researching individual marketing campaigns. What also grabbed my interest recently was an article in the Winter 2011/12 issue of data IQ magazine (free at www.dqmgroup.com/dataiq ) written by Joanna Reynolds of Reynolds Busby Lee. It just so happens that another attendee on the database marketing course I mentioned was Joanna, who was client side at the time and we invited some select clients onto a course we funded. So she worked for Consumers Association, Time, Readers Digest and others along the way to founding her own agency. What Joanna said in her article was

 “We need to go beyond data, beyond the customer contact cycle, beyond post-purchase research and beyond monitoring social media if we are to fully understand our customers’ experience.”

And of course, Joanna’s company has a solution based on a real-time diary of customer experience. Diaries are not a new tool, although the ability to deliver online in real time is very powerful. But I’m not advocating this as the solution to all our needs, just one of them in the general area of panels, diaries, real time and immediate post research – but there are enough tools out there to validly analyse behaviour across multiple channels, tested offerings and the ‘why didn’t they’ as much as the ‘why did they’, much more quickly than waiting for a big data solution.

I suppose it’s inbred, but what I really believe is that constant testing is the right thing to do, adapted to a more integrated and fast moving marketing world – but it needs a lot more effort, budget, understanding and skill sets than it used to do, and the direct and digital communities have less not more right now.

{ 0 comments… add one now }

Leave a Comment

Previous post:

Next post: