Monday, September 21, 2009

To test or not to test? That shouldn’t be the question!

An other guest posting from my friend & Ideal Interface colleague Marc Ames:

One of the greatest advantages of online is the level of accountability it gives marketing people with their budgets. It can also be provide in near enough real-time with the analytic tools available in the market. So with this accountability comes the obvious conclusion that online also provides the advantage of being to conduct multivariate testing – whether it be on a web page, an e-mail, Pay Per Click copy or an online advert.

Yet, there are still many companies out that invest massively online and yet do not conduct even the simplest of A/B tests.

But A/B testing is nothing new for marketers. Direct marketing has embraced this for decades and if you want to read up more about this then I recommend you get Commonsense Direct and Digital Marketing from one of the Direct Marketing gurus – Drayton Bird. The “Direct” version was first published in 1982 and was and probably still is on the recommended reading lists for the CIM qualifications when I first came across this book back in the early 1990s.
It is still relevant, in fact probably even more so today, 27 years on, but despite the wealth of evidence that supports testing, then why do some marketers still not do it?

It is not as if all marketers have the insight to know what is the exact page on a web site to maximise response or sales as a survey by Maxymiser proved in May when only 21 out of 452 marketers correctly identified the best performing web content from an A/B/C/D test.

So what are the reasons for not doing this?

I think it comes down to one of the following reasons:

1) Resources – many think that simple A/B testing will involve increasing the time to develop the web page, or e-mail, or online advert. But the reality is that if you start off simple then A/B testing can be easily accommodated within the current processes. The change could be as simple as colour, or ordering, or a headline on an e-mail, or the time of day or actual day an e-mail is sent. How many times have you sat there and presented two or more options for the final sign-off? Why not test both?

2) Building the site for me and not the audience – despite the fact that you build a website or design an e-mail for your target audience, I can guarantee that in meeting rooms and offices across the world that websites and e-mails are being sign-off or changed because of the “gut feel” of the decision maker. For no better reason than that is what the decision maker likes most. The fact that they may not fit or understand their target audience is an irrelevance as long as it makes them happy. If you are in this position – bad luck, and good luck with the job search!

3) Can’t build the business case – perhaps you are stuck with the conundrum of how can you prove that it is worth testing unless you know what the results of the uplift are and go around in circles as you don’t want to make a prediction and the test fails? Well consider the lost opportunity if you don’t test. Look at elements that you know are not performing, perhaps a high drop-off page on a website, or copy for your PPC campaigns, or the fact that your competitors are sending e-mails out at different times to you. If you look hard enough there is certain to be some element of your online marketing that can be improved – and use a test to prove it – and that will help prove that testing works.

4) Marketers not knowing what real marketing is – coming from an engineering background I have to admit I have an axe to grind but unfortunately I have come across too many marketing people who talk about “creative” and not enough about the numbers. There would be a lot more marketers in the boardrooms of the UK if they got to grips with the numbers side of the business and unfortunately this does present an image problem for Marketing, which still has not been addressed.

In the current environment, it is easy to say that because you are so stretched that you need have so many priorities for “business as usual”, but my challenge to that is that this is exactly the sort of thing you must be doing in the current environment to prove your worth and ensure that you help your business through these tough times.

So if people in your organisation see multivariate testing as the equivalent to “How do you an eat an elephant?”

The answer is quite simple – “One bite at a time”.

Note;
You can still read Marc's previous guest post Did the BBC and not Video kill the Radio Star?

No comments: