It seems you can’t move on Twitter these days for tweets linking to blogs, reports and opinion pieces about automation in market research and suggestions that it won’t be long before researchers may be replaced by robots. (It would certainly make the tea round easier in my office.)
Musing on this, I found myself investigating the possibility of a bot writing this blog… I even tried asking Amazon’s Alexa for her input but not surprisingly, she didn’t understand. And I quickly found that blog writing wasn’t a task I was going to be able to delegate to a bot any time soon. For similar reasons (which I will examine) there are also certain aspects of market research which will remain in the human domain for the foreseeable future. There are also many tasks which can and should be automated. In this blog (written by me) I will examine both sides of the human and automated market research coin.
| Tweet This | |
| "Certain aspects of market research should be automated. others will remaim human" |
One of the areas arguably ripe for automation is the commissioning, set up and recruitment process of market research projects. Rather than having to enquire through a website, talk to an Account Manager, raise a PO, have a set up call, etc., you could simply supply your credit card details, select a research tool, purchase a sample and off you go. Sample purchases could be linked to external research platforms if required and agile testing can be conducted speed. Sounds great! In theory…
Reality Check: A little time invested at the research outset can mean a lot of time (and money) saved in the long run. Taking a few minutes to talk to the research experts supporting sample and/or research software allows you to benefit from their experience of it. And this heightened level of understanding will lead to greater accuracy in your sample selection and a tool methodology which ensures objectives are met.
When it comes to setting up your research activities, be that programming a survey, designing a focus group topic guide or planning a diary study, automation can help to varying extents. Bite size, off-the-peg research activities allow for the fast deployment of smaller tasks. Survey templates support the design process by removing the ‘blank sheet of paper’ syndrome. Meanwhile, creative and visual tools such as brainstorms and smartboards offer an engaging, user-friendly ways of gaining collaborative feedback without the need for twiddling with hundreds of configuration options.
Reality Check: A range of tools designed individually to fulfil every research objective holistically do enable the agile deployment of research activities. Just be sure to choose the tools with the functionality most suited to your needs. Again, this is where un-automated advice from your research provider can be beneficial.
It’s tempting to think of analysis as the primary opportunity for market research automation. This is certainly true of some basic tasks surrounding quantitative data, i.e. cleaning files, merging datasets and recoding variables. Indeed, such tasks have already been automated. However, a skilled researcher is still needed to make judgements as to how the data are interpreted, what’s important and what is not.
The automated analysis of qual data still lags behind that of quant because language and sentiment analysis is much harder to achieve. Often you will hear companies claiming that they offer qual analysis, which then turns out to be nothing more than a frequency count of words depicted in a wordle. Even when software can help you categorise and detect themes within your data, it still takes a researcher with an understanding of customers’ lives in context, business’ processes and market environment to draw out the most salient points.
Reality Check: You can automate a lot of the research analysis process, but the judgement about what is important still rests with you, the researcher. What’s great (for job retention purposes!) is that it is in this innately human function that the real insight value lies.
| Tweet This | |
| "Raw research data to thematic outcomes = Automation. Thematic outcomes to insight = Researcher |
Products such as Wordsmith, a ‘Natural Language Generation engine’, which allows you to turn data into a natural-sounding text appear, on the face of it, to be a way of getting that PowerPoint presentation written for you when time is tight. However, the long-hoped for ‘write report’ button that you could press and wander off to make a cup of tea is not quite there yet.
Whilst Wordsmith no doubt has value for turning market data and such like into reports, it still requires quite a bit of up-front effort in terms of writing a template, adding in merge codes, and specifying alternative text to be used depending on the results being merged in. Most of all, it still requires human input to decide what is important enough to be included in the first place and to ensure that this is communicated to stakeholders in the most appropriate and engaging fashion.
Reality Check: At present, there are a lot of tools which can help in the provision of content for your report, but the magic ‘create report’ button on your keyboard is still a long way off.
One of the big drivers of automation is autonomy - clients want more control and quicker turn-around times and a big trend here is DIY. Hypothetically, market research automation goes a few steps further than DIY making things not just quicker but also effortless. In reality, at least today's reality, that's not the case.
One of our clients recently said, ‘my brain isn’t working today, I want you to sort it out for me’. As we’ve seen, automated processed still require a good deal of thought and planning up-front. And they are a long way from taking over completely, or from being able to provide the reassurance, ‘hand holding’ and encouragement that a human project manager can.
Moving from the process of setting up and running market research activities to the fundamental purpose of the research itself, the actionable insight, gives us even more pause for thought. No one really wants ‘outputs’ from market research. What they want is guidance on what their brand should do off the back of these outputs. And that’s not something you’re going to get from a bot any time soon. Even with the most advanced automation, there is still the need for a crack team of research experts to be on hand.
Now, if you’ll excuse me, it’s my turn to make the brew round again…