Download the

Jan Hofmeyr Challenges The Future Of Market Research


By Ray Poynter

Last week, at the MRMW Conference in London, the future of market research as we know it was challenged by Jan Hofmeyr. Although there were many informative and interesting presentations, Jan was that the only person who was talking about a very different way of doing business.

In writing this post I am working from memory, so apologies if I misrepresent anything. The presentation in London appeared to be a continuation of a presentation that Jan made last year in Amsterdam, at the ESOMAR 3D Conference. A continuation in the sense that he had moved his thinking on, and an extension, in that he now appears to be offering a solution to some of the world’s largest agencies and clients.


The main points

Jan Hofmeyr’s main points were:

  1. The existing model of market research, in particular the large trackers, is broken. It is too slow, too expensive, and not sufficiently useful. Not many people would argue with this point of view.
  2. The best device, for collecting tracking interviews is the mobile phone. Jan’s key point is that nearly everybody has a phone and they have it with them almost all the time. And, by mobile phone, he means both feature phones and smartphone.
  3. The core of his suggested data collection should be text based and last about 120 seconds. It should be text based because more than half the world is still using a feature phone, and it can be 120 seconds because it is going to focus on just 3 key products per person (the three most relevant to each person). And, it needs to be 120 seconds to make it affordable and to reach enough people.
  4. Social media monitoring should be added to the tracking mix to make the information richer.
  5. Predictive analytics should be used to look at the data to predict what is likely to happen next – rather than reporting what did happen weeks or months ago.
  6. Artificial Intelligence or Expert Systems should be used to analyze the data, to produce reports that have not been written by people and which are automatically translated into the client’s preferred languages.

The key benefits

The research will be much cheaper, more insightful, predictive, and faster.

The implications for the MR industry

The MR industry would need far fewer employees. These employees would mostly be experts, salesmen, and accountants. Presumably, the sort of process Jan is talking about would not affect the small ad hoc projects, especially the qual projects nearly so much. What he is mostly talking about are the big projects; the brand, ad, and customer satisfaction trackers, the places where clients spend most of their money, and where most market researchers are employed.

However, if the big stuff were to change to the extent that Jan is forecasting, then there would surely be implications for other types of research too?

The reaction in the room?

Most people in the room showed no reaction to Jan saying they may not have a job within a year or two. Did they not believe him, not understand him, or not really listen?

My initial reaction was to focus on the bits of his plan that, as described, do not seem possible with current technologies. However, I do accept his central points about a) the need for change, and b) the possibility for change.

My reservations

My key reservations were, that on the evidence presented, I do not feel that the following items will work as well, or as accurately, as he intimated:

  • Collection of emotional data about brands, from short surveys or from social media.
  • Predicting brand movements. Jan seemed to be suggesting he could predict the score a brand would get in the future. I think that predicting a cloud of possible values is more realistic (a point covered by both Nate Silver and Nassim Nicholas Talib in their books The Signal and the Noise and The Black Swan).
  • The ability to create expert systems that could automatically produce non-trivial reports.
  • The ability to automate the tailoring of reports to different clients and languages.

What’s next?

From what I hear from the US and from Europe, Jan Hofmeyr has some very interested major clients/potential clients. So, some people are taking it very seriously. Perhaps some of the wackier elements of his presentation are not part of what he currently plans to deliver? Perhaps the core of what he is suggesting is more limited and more practical? Although I have some major reservations about what Jan Hofmeyr presented, I suspect he is closer to the truth than the majority of the room who seemed to believe that the MR industry in two and five years will be much like it is today.

Please share...

11 responses to “Jan Hofmeyr Challenges The Future Of Market Research

  1. Thanks for posting this Ray.
    I wasn’t at MRMW in London, but by coincidence, my Blog post last week on C3Centricity was on exactly the same topic, albeit from a slightly different angle to Jan’s. Clearly we need to join voices!
    As you certainly saw, also started some heated discussions on several LinkedIn groups with as many people defending the status quo as agreeing that it’s time for change.
    Don’t you think it’s time ESOMAR speeded up its review and started joining the conversation?

  2. It is always a puzzle to me why when some new methodologies come along why we suddenly get a flurry of doomsayers about this industry. The existing model of market research is hardly broken or we would not be seeing rapid adoption of new technologies. But let’s take Ray Poynter’s recall of the key lines from Hofmeyr’s speech.

    First was that large scale tracking is broken. The arguments seem to relate to slowness, expense and usefulness. What a load of rubbish. I would be recognized as the tracking guru in Asia and I can assure you it is alive and well out here. Online has meant we can turn around reporting in two weeks at much lower costs than in the past and sample sizes are healthy and insights as useful as ever. Just one example of market research adapting to new technologies!

    The second comment that mobile phones are going to be the medium for surveys in future defies logic. Even that “pink sunglasses” study by the UKMRS admitted that screen sizes and formatting are going to be barriers to usefulness. I would add an even bigger issue, just who are these respondents that will give up anything more than a few minutes of their mobile lives to participate in a survey? No one has been commenting on this as large scale samples obfuscate what I am sure is a huge bias in age profiles and attitudes among those who respond to mobile surveys.

    I also seriously wonder whether anyone in the industry at management level out there has actually ever tried to complete a mobile phone survey on the go. It is usually very annoying with screen formats making it hard to respond and with slow load-ups and dropouts as signals vary. Dropout rates must be horrific (again no reported data!) Participation rates in mobile, I should add “quality” participation rates, will prove as elusive as big data’s claims for predictive breakthroughs.

    The short survey that Hofmeyr postulates is also a bit of “hocus pocus”. Nano-surveys look interesting but think through the limited applications and the doubtful willingness of experienced marketers to buy data that has been stitched together across multiple nano-surveys. Sorry, not even suits in ad agencies will fall for this. Does anyone honestly believe that usage claims in one survey can be linked to reasons why in another, even with the supposed validity of fusion tools.

    Social media monitoring is another promise yet to make its value proven. All I ever read is how Twitter feed, sentiment analysis and whatever other social media data are simply not delivering on the major promise of building brand loyalty and sales. Big brands that are using these methods sure see a peak in comments, dubious Likes and sentiment, but as one reported, “no bump in sales”. So what is it all about and why is it seen to be so valuable in brand insights? The bigger issue we all should reflect on is the major impact of spurious paid blogging and other manipulative techniques that plague social media at the moment. Perhaps we should ask ourselves just how valid and reliable is all this data and how does it make information “all that richer”?

    As far as I can see, the blogspace suggests that the only people who seriously challenge all this smoke and mirrors stuff are the supposed dinosaurs of the market research industry said to be “defensive about their jobs” or “unwilling to recognize change is inevitable”. The reality is we are more like pure scientists in the realm of new age gurus and desperate futurologists trying to stake out a position.

    I wonder if some of these pundits would care to put their money where their mouths are and give a time line for all these prognostications? Probably not. Or perhaps some artificial intelligence might step up to the plate and prove Hofmeyr right?

  3. I think Jan is a marketing genious. I had the privilege of working with his Conversion Model while with MEMRB and afterwards Synovate. I think his model changed the way companies think about the elusive notion of Brand Equity. I fully agree with this connection of listening and asking quesitons. The only thing I want to add is that texting is very limiting and we should keep in mind that by 2017 5 billion people will have access to broadband through a smartphone. That will change everything in the way we try to adapt to fetaure phones.

  4. Doesn’t the call for automated reporting from tracking studies fly in the face of the many calls for us to move from data providers to insight providers? Pretty hard to do if a machine is producing your output and unlikely that the output will contain much insight.

  5. The existing model of market research actually is broken. If it weren’t, we wouldn’t be seeing so many clients press all of us so hard to take costs out but to deliver more actionable insight. The days of long tracking surveys with questions that bear little relationship to human behaviour are coming to an end. To take just one simple example of such questions: any question that relies on people’s memories of what they’ve done e.g. which brand(s) have you bought in the past week?

    A few more points.

    Two week turn-around: fine until someone comes along who can turn predictive marketing insights around in a day or two. This isn’t fiction. I know of at least one company that delivers insights in hours rather than weeks.

    Mobile surveys. I would not be talking about our ability to migrate tracking to mobile devices if I hadn’t seen the quality of the information that we can deliver with questions that take the limitations of screen size into account.

    Nano-surveys: as it happens, I’m not a fan of ‘stitching things together’. So I agree. But I’m a very big fan of a few ultra-impact questions that deliver high quality marketing information. Just to be clear then, we are not relying on data-fusion.

    Shorter, smarter surveys: When we ask people to participate in long and boring surveys we compromise the validity of their responses. But surveys do not have to be long and boring; we can reduce survey length by removing redundant questions and focussing on questions that are relevant to our respondents. So, with fewer questions we can improve the accuracy and validity of our research. My opinion on this is well documented –

    Automated reporting: this isn’t fiction. I’ve been in the business of delivering high-end IP for 25 years. We currently have a system operating that delivers a basic level of reporting automatically. By ‘basic’ I mean: to a standard that typically takes a researcher 3 years of training to achieve. Automating outputs in this way frees researchers up to be more consultative.

    Perhaps one last point: don’t underestimate what a machine can do in the hands of a smart person.

    Processes of change always take longer than breathless prognosticators predict. But someone has to be willing to take out the loud-hailer to wake people up.

    1. Great additional feedback Jan, and I agree on all levels. It’s my job to monitor such advances and I am aware of multiple companies that deliver on all the functions you have outlined. As you’ve said, none of this is fiction. IN fact, much of what you are discussing is being built by very large global tech companies such as Google, IBM, Intel, SAP and others. And in most cases these technologies have been in active use for years now; it’s the commercial opportunity that has brought them to “critical mass” now. That imperative is being driven client demand.

  6. Jan’s statement that “with fewer questions we can improve the accuracy and validity of our research” is a little off-base, even after I’ve read the linked article. Asking better, more relevant questions, for sure. But just cutting questions does not ipso facto make for a more valid and accurate survey.

  7. Jan I wish they had a Likes option! Loved your response. Lets agree to disagree. I just looked at the turnaround time on tracking for a major banking client and we are now down to less than week. Only a machine could beat that servicing rate. Maybe that is the future? And Leonard if that is the future there are probably machines that will prove as annoyingly challenging as you!

  8. I was there in London to listen to Jannie – spluttering into my coffee. A lot of what Mr. Hofmeyr provocatively suggested – the need to allow machines to perform tasks they can do better than humans, thereby “freeing up” researchers for….err….reading a book on the beach is what I recall Jannie saying, as being a researcher simply isn’t fun. I’ll out myself as a humanist with a degree of scepticism of a belief in machines’ ability to get near interpreting the complexity of many situations, human’s “irrationality” (or our less than rational approach to many situations). Lots fo BE insights point out how often it is difficult to accurately assess motivation. A small example – planning a trip to another town with your partner, you need to select a hotel – how many loops and re.thinks do you go through, with your partner? How many times do you change your mind? On another tack: if machine intelligence is so good, how come the Financial analysts (surely better quants that Research has) didn’t predict the crash in 2008/9? How many predictions have I seen from Consultants that have been so spectactularly wrong, but look oh so plausible? The further we get from an individual take on humans – observing them in their environment – possibly armed with some behavioural quant. data about them too, the worse. Yes, cost is a driver – but focussing on cost at the expense of adding value is a very dangerous strategy. That having been said, I of course took note of what the esteemed Mr Hofmeyr says – one would be foolish not to.

  9. One of the problems I consistently see is how people on both sides of the issue approach the perspective of the other side. Too often the people pushing new approaches take the following approach: “You traditionalists are too stupid or backward to see that you’re toast. You’re all going to be out of jobs with your inane surveys and focus groups because it’s all rubbish and I can do it better.”

    On the other side, too often the traditionalists take the approach: “You dreamers are idiots to think any of this is valid. It’s all unproven garbage and has so many holes that it’s useless. You’re just trying to pull the wool over everyone’s eyes to make some sales until you crawl back to the dreamworld you came from – and we’ll still be here doing what we do best.”

    At some point, I’m hoping more of the NextGen people stop feeling like they have to predict that doomsday will be here tomorrow and that everything traditional is useless, and that more of the traditionalists will take a valid look at some of the new approaches rather than just dismissing them as unproven (the car, the radio, the computer, and the telephone were all unproven once too). Maybe by collaborating instead of dismissing each other can we actually advance the industry to be more useful to clients without throwing out the experience of anyone who has the gall to still moderate focus groups or design surveys.

  10. Well said, Ron. On behalf of the traditionalists (by virtue of age if nothing else), I don’t think it’s too much to ask to see some validation of a tool instead of incessant hype. I love new stuff that works – does something faster or better or just as good but cheaper and easier. But I also want to know that it works the way it’s supposed to, that the answers are reliable, generalizable, and map the outcome to something in the real world.

Join the conversation