Group 8
+31 (0)20 8466 855
Page 1 Copy 8
Group 3

Latest news


Market research is a) magic, b) madness or c) all of the above.

Market research can be a wild ride - ours has been, so far at least!

- article by Hans Lingeman, Winkle CEO

This is the first article in a series that's being published on LinkedIn as well as in other places on the internet. We're putting them up on our website for you to enjoy, as well. Like what you read? Drop us a line at Hans' Linkedin page.

Having been in market research for over twenty years, I’ve seen a lot of what our profession has to offer. Earned some credit for it, too: received the MOA award for ‘Researcher of the year’ in 2013, nominated for the Innovation Award in 2011.

Together with the smart people at Winkle, the market research agency I co-founded in 2007 and part of Happen Group, we conducted some several hundred research projects for a dizzying variety of clients, in a dazzling diversity of markets.

Market research can be a wild ride - ours has been, so far at least!

Today, we’re launching a series of articles that will articulate some of the insights we’ve developed over the years. We’re calling it ‘Riding the Curve’ and we’re going to try and publish new additions at regular intervals - say, once every two months. By we, I mean myself and the people that have helped shape these ideas for the past decade or so: amiable experts at Winkle, at Happen, and a host of others I’ve come to know and appreciate. We’ll be inviting them in from time to time, to drop a line where it fits.

The ambition is to add a useful series of smaller and larger bits of insight to the already substantial pool of knowledge in market research. The main question we’ll ask:

How can the science of market research best aid the art of innovation?

I hope you’ll join us for the ride.

Riding the Curve

The ‘Curve’ to ‘ride’ here is of course the S-Curve as first described by Gabriel Tarde in 1903. Tarde discovered that the ‘life’ of most successful innovations followed the same S-shaped kind of pattern of launch, uptake, maturity and expiration. Since that discovery, the S-curve has been extremely useful to innovators everywhere. Big brands and start-ups alike have used the Curve to try and predict the path their cherished invention might take.

And such predictive power is usually in high demand – hence the popularity of market research in general. Introducing new products often feels like shooting at a fish in a pitch-black pond – at night, while blindfolded…! The hopeful entrepreneur, or marketing director has only one launch, a single shot to fire - at a market that appears to behave completely unpredictably (see image to the right). It’s insightful theories such as Tarde’s S-curve, that at least give the hopeful innovator something to plot a successful product lifecycle through - even if the targeted market remains a mystery.

Which is where this series comes in. We’re going to explore how a new application of the S-Curve may aid innovators, start-ups and most certainly market researchers alike in their pursuit of the holy grail: a successful product launch. It’s a take on the Curve that keeps raising eyebrows among my audiences of entrepreneurs, intrapreneurs, innovators and marketers.

"It’s the market, stupid"

What we realised over the past years of research is that markets… do evolve in patterns that are just as S-curve shaped as the life cycles of the products that enter them.

Innovators seem to treat the market into which they’re about to launch as a given. The market is unpredictable, they say. Not in the least, we say. Markets may be analysed through the exact same lifecycle frame. And viewing the subject from that perspective, well, let’s just say it tells us some interesting things – so stay with us.

The why behind this writing

Tipping the hat to Simon Sinek’s Golden Circle, I think I should spend some words here on why I’ve decided to start this series. The reason isn’t just that I’m now in position to reflect on a rich and fascinating string of enquiries; it isn’t because I’m just so happy to be in this fantastic line of business, either.

But it often seems like I’m the only one who feels great about being a market researcher.

Market research gets a lot criticism and mistrust from the public at large. What’s worse, I often get the feeling that fellow market researchers themselves aren’t even proud of the work they do. Ours is a business where under-appreciation seems to be the norm, and… I think that’s wholly inappropriate.

Market research deserves a break.

When we do our work properly, market researchers are the Sherlock Holmes of business. We figure out the dots and we connect them. We separate astute deduction from casual speculation; we apply method to the apparent madness of markets. And in doing so, we often help business bring fantastic things to people everywhere. So I guess… I’m in it to help market researchers everywhere a little prouder of what they do and what they can achieve.

To do that, we’re going to illustrate, in an alternated series of case studies and theoretical exposition, how market research can contribute to a successfully planned and executed product launch and how it has done so time and time again.

Separate the wheat from the chaff

So, if market research can help secure a successful launch, why doesn’t it prevent all those misfires and wasted millions? Let me cut to the chase here and say that there's no such thing, as the golden approach and I'm pretty sceptical this will be found during my life time. But there are ways to assess the potential of your research.

We’re going to show how many of the millions routinely spent on ultimately not-so-useful research might have been identified at the start as a waste of time. And I’m not just talking about faulty scientific rigour. Neither am I talking about the vilified, trumped up ‘straw man’ research commissioned by brands that just want something that looks scientific to help them make a point.

I’m talking about honest-to-goodness, scientifically formulated research questions, meant to help genuine commercial ambitions – that fail to find a meaningful answer.

Market research is a tool, and like any tool it only works magic in the hands of a skilled craftsman. Ah, but how to identify the skilled market researcher? Something we aim to bring across in this series, as well…

To summarise:

1) we’re going to dive into an interesting bit of theory for the innovation-minded;

2) we’re going to help everyone feel a lot better about market research;

3) we’re going to identify what makes a solid market researcher;

4) and help those market researchers feel a bit better about themselves;

5) we’re going to make excursions into whatever takes our fancy along the way: ethics, anecdotes, world views, secrets of the trade…

In short - what’s not to like? Riding the Curve is going to be an interesting romp I’m sure, and I hope to see you for the next instalment. We're also publishing this series on, so join us there - or here - for comments.

Why all Shampoo smells like Fear to me

"Here's to the crazy ones." - Apple commercial, 1997

- article by Hans Lingeman, Winkle CEO

This is the first article in a series that's being published on LinkedIn as well as in other places on the internet. We're putting them up on our website for you to enjoy, as well. Like what you read? Drop us a line at Hans' Linkedin page.

For market researchers, a lot of time and effort is put into designing an endless stream of variations on a base product - on the assumption that the base product itself is no longer worth considering: seemingly eternal, taken for granted as much as the air we breathe. The S-curve suggests that, when this practice sets in, it’s best to take a long hard look at that base product: it may no longer be around in five years’ time...

Welcome to the second instalment of Riding the Curve.

Two types of response…

When I talk about the S-curve in my work, I generally get two types of response. The first is “Oh, that one again” and the second is “Hmmm… tell me more”. I also find that a lot of people who think they’ve learnt all they need to know about the Curve nevertheless either apply the concept in a superficial manner, or they’re outright doing it wrong. So even if you’ve heard about S-curves in product-market combinations before, you might want to keep reading.

…from two types of people.

The audience that responds to the perspective I offer also roughly falls into two categories. The first is the entrepreneurial (or intrapreneurial) type: at the helm of a start-up or an experimental department. They have nothing to gain from the status quo, from me-too versions and tweenager spin-offs. Hungry for change they are, as change might mean their revolutionary brain-child may get to have its time in the sun.

The second type is who I’m writing this edition of Riding the Curve for. Consider the following job type: a product group manager at any major brand in FMCG. A position they themselves often perceive as rather rock solid. From their perspective, disruptive change is a threat - but it’s generally deemed so unthinkable that it isn’t worth considering.

The message to both audiences is the same. Either repent or rejoice, depending on your interest, because change doesn’t care. It happens anyway.

Dealing with the unthinkable

Here's a little thought experiment:

What would happen if we suddenly lost all electricity?

Here's what I think would happen.My guess is, we’d be chopping down local trees for firewood within a week. Do we hoard food and water for such an event, though? Of course not. It’s something you don’t prepare for because the mere thought is unthinkable. Who prepares for something that unlikely? Dutch law decrees that every household must own a battery-powered transistor radio in the event that nation-wide electricity and all dependent communications channels cease to operate. Do you own one? Do you know who does?

For a product manager in the shampoo category, the idea that people might one day lose their appetite for shampoo can be called just as unlikely: impossible, unthinkable. Yet, it’s quite possibly happening as we live and breathe. Recent studies have suggested that washing one’s hair every day is not entirely as necessary or as healthy as we always thought. And a society is slowly following suit: the ‘no poo movement’ is slowly gaining traction. Their popular wisdom: we should consider lathering up only every other day. Sounds trifle? Consider what shampooing only every other day means for global sales. A sales drop of 50%. Unthinkable? It’s a change that is already underway. Click any of the links in that sentence to see examples.

Of course, commercial history abounds with examples of rock-solid products becoming all but obsolete, almost overnight.

  • Kodak and Polaroid, in the wake of digital film;
  • Palm Pilot, in the wake of the smart phone,
  • video rentals (I’m looking at you, Blockbuster),
  • book stores (Barnes & Noble),
  • road maps (Michelin),
  • fax machines (Xerox)…

the list goes on. That last example is a gift that keeps on giving, by the way: after losing a major cash cow in fax machines, it now looks as if the entire printer industry is floundering in the wake of the digital revolution.

Better safe than stupid

So, once we have it in the open that the unthinkable might well be inevitable and the question is not if it happens but when it will, two follow-up questions present themselves: 1) how to signal imminent fundamental change in your own product category? And 2)… how to prepare for it? That’s where market research comes in.

Crowded, complex categories are a red flag

If we ask ourselves where the decline of a product (or a product category) sets in, the first observation to make is that the Decline phase is always preceded by Late Maturity. And Late Maturity is characterised by two things: product "complexity" and crowdedness. The process can take decades, but it can also happen quite quickly.

An example of growing complexity that took decades comes from the automobile industry. The industry started out jaw-droppingly simplistic, with Henry Ford immortalising himself when he quipped “You can have a Ford in any colour you want, as long as it’s black.’ Over time, both the category and its language have become intensely competitive and complicated. The image to the right already targets a subcategory of car enthusiasts - the up-and-coming 'young business man'.

Considering the fact that not cars, but horses had been the standard for transportation for 5,000 years before - you'll agree that the speed with which the automobile market arrived at its current level of complexity and crowdedness - just about 50 years - is actually pretty darn quick.

A lot quicker still has been the rise of complexity in a category such as mobile phones. The first mobile phone was built in 1973; since then the complexity has risen dramatically with each new generation. As soon as the market smells a hit, new brands, features and editions pop up everywhere. Pretty soon, there’s a mobile phone for every lifestyle and budget. How about this for a maxim:

As category complexity grows, so does the possibility for radical change.

And when it does, it does so in two ways: discounters and disruptors.

Be wary of discounters…

The first change is to beware the discounter. This is a complete no-brainer: once a market matures to the point where zillions of different options are available, someone is going to undercut the market with a budget variant that once again claims simplicity as a selling point.

But beware the disruptor!

The second cause for alertness in complex categories is that complexity invites disruption. Now, I’m not claiming complexity must lead to disruption; indeed, there are plenty of well-established categories (beer, orange juice) that have become extremely complex - without anyone coming along with an innovation that rewrites the rules completely. What I would claim is that complexity is both a necessary condition for disruption, and a catalyst for it. It’s the only way out for a complex system.

This is where I claim that markets follow a rule that looks a lot like the second law of thermodynamics. Remember how entropy in a system can only ever increase, not decrease? Thought so… Well, I’d think product categories behave rather similarly: the longer they exist, the more complex they get – and this complexity is never reversed. You simply cannot un-invent a variation, once invented. Also, inevitably, someone is going to engineer a change to the base product that is so fundamental that it can be called disruptive. At that point, all bets are off.

How to stay alert in a complex category?

I’d say a proper research-driven approach to dealing with change, for the wise product category manager, falls into two categories:

1. Keep one ear everywhere, and the other to the crowd

Change will manifest itself in the crowd – after all, it’s their behaviour that constitutes the actual change. But the driver of change can be a lot of things – external to the crowd, most of them:

1. Scientific breakthroughs

When science publishes a report saying shampooing every day damages your hair, P&G will be wise to take note, But will the crowd respond?

2. Government policies

If government raises the tobacco tax by double digits, BAT will notice. But do smokers care?

3. Gamechangers

When the E3 tech conference launches a new gaming platform, Sony watches closely. But will the gamers move?

4. Iconic behaviour

When stars start wearing beards, Gillette must take heed. But will the hipsters stop shaving? (spoiler: yes, they did.)

Resumé: All of these changes are 'kicked off' outside the crowd. So you monitor what happens in science, government, showbizz and whatnot. But sales are of course only actually hurt once the crowd starts moving. So how do you keep that other ear glued to the crowd?

At Winkle, we have several tools in social listening that apply software engines to monitor and analyse the torrential stream of verbatim comments by consumers of a narrowly-defined subject such as a product or brand. We also build groups of well-articulated consumers that are passionate about a certain product, brand or category, willing to engage in meaningful dialogue with those behind it.

2. Go controlled crazy

Despite the somewhat flippant tone of the introduction, I would probably be the last one to say brands shouldn’t invest in pivoting their product to serve new age groups or life styles. It’s often an absolute necessity if you want to stay competitive in your category.

But I’d also be the last to say that you should keep churning out variants without checking to see if anyone’s interested to buy them.

3. Go uncontrolled crazy

The main objection against that option number 2 is that it tends to only walk the safe side of innovation. And while that option will allow you to radiate creativity, and to stay one step ahead of your regular competition, it won’t save you from a radical disruptor, or from a radical disruptive event.

So aside from putting resources in incremental innovations such as ginger-flavoured shampoo, I argue brands should always keep a small, dedicated group of creative minds on hand to explore the unthinkable. Unconventional thinkers that are given the space to tackle the potential category killers: what if water becomes so expensive, we can’t afford to rinse our hair anymore? What if baldness becomes the fashion? Et cetera.

Market research isn’t just the science of building big statistics and using them to drive minor adjustments. It’s also a discipline of making wild observations with far-reaching implications, reading big changes into small signals. And validating that hunch, even those most unthinkable. Being proven wrong. Being proven wrong again. And then, suddenly…. correctly noticing the changing winds before anyone else.

Robots can read. Soon.

In this (shorter) episode we’re looking at how Natural Language Processing (NLP) helps us to make sense of the consumer data torrent.

- article by Hans Lingeman, Winkle CEO

This is the first article in a series that's being published on LinkedIn as well as in other places on the internet. We're putting them up on our website for you to enjoy, as well. Like what you read? Drop us a line at Hans' Linkedin page.

Judging your place on the S-curve, or how your category unfolds itself, is all about reading your consumer audience. So, what’s on today’s consumers’ minds? You may assume that the flood of daily user-generated content has the answer and you'd be partially correct: there is a golden qualitative narrative hidden beneath the quantitative blizzard. But bringing it to the surface is astoundingly complex – it’s the very forefront of today’s market research analytics.

Welcome to the third instalment of Riding the Curve. You can have a look at our earlier episodes here:

1. Market research is a) magic, b) madness or c) all of the above

2. Why all shampoo smells like fear to me

In this (shorter) episode we’re looking at how Natural Language Processing (NLP) helps us to make sense of the consumer data torrent. Before starting, I would like to write a special thank you to my colleague Julian Kievit, who is deeply involved in the drive to advance this promising science and who helped me with tips and suggestions in preparing this instalment.

What’s in it for market research

The business case of using artificial intelligence in market research isn’t hard to explain. Where conventional market research involves going out and asking crowds of people for their participation in questionnaires and whatnot, the age of the internet has seen those crowds take to the Web and write, write, write. Resulting in petabytes and petabytes of user generated content (UGC) every day – and a lot of it involves reviews and comments on products, innovations and experiences. Written from the heart and from the gut – almost no politically correct answers, and certainly no researcher bias. A goldmine for market research – if only we could read and analyse all of it fast enough.

Limited comprehension

Computers – or more specifically, analysis engines - can ‘process’ consumer-generated data - gazillions of times faster than humans do, even. When it comes to reading, however – engines need to be ‘trained’. Which is quite like how we - adults – have built our social skills and ability to communicate through years of experience and learning. Engines can reliably ‘read’ lines that have been written specifically for them, i.e. literal programming. When processing lines that were not written for a computer, such as unstructured verbatim, we need to instruct the engine very precisely on how it should process what it’s looking at, or we need to feed in truckloads of data for it to derive sense of it all (machine learning). And to do this, thousands of scientists are creating ever better models of how we build sentences and convey our thoughts - which requires us to fully understand exactly just how mindbogglingly complex our language really is.

NLP: what it is

The science of teaching engines how to process our natural language is aptly called Natural Language Processing. It ties into the wider movement towards Artificial Intelligence, which may see us all under a man-made god someday. Let’s just say that if our robot overlords one day start screening our emails for signs of subversion, we’ll have NLP in part to thank for it.

Why it’s so hard

When it comes to language, our creative, associative minds are almost impossibly hard to follow for a rules-driven machine. Our language is riddled with what we call ‘lexical ambiguity’: we may use a single word to indicate vastly different things (‘square’ as a shape and as a mindset; ‘hard’ for a surface or a difficulty level). We may also put words together to form vastly different concepts (‘space ship’, ‘rain check’).

We humans use idioms, sarcasm, deliberate puns; we deliberately say the opposite of what we mean ("Don't you like eggs, Sam?"); we use syntax, grammar and whatnot to convey meanings far beyond the meanings of separate words alone… the list is endless, and this all still assumes we’re writing what we want to write. Considered how careless and error-prone people are when typing up reviews online, the magnitude of the problem increases exponentially.

In short, getting an engine to understand how it should process language, it needs models of how we build language which is a lot like modelling how we think.

Only human

Decades of hard work by dedicated scientists and programmers have slowly begun to advance the theoretical beginnings of NLP into an ever more detailed, pragmatic approach to capturing the quizzical contours of our speech and writing. Nevertheless, despite the expectations brought on by sales pitches from commercial language processing engines such as Radian6, Meltwater, Infegy Atlas, Lexalytics – the skilled hand of a trained human expert is still crucial to arriving at meaningful conclusions.

Let’s look at the learning curve NLP is expected to follow on its path to fully grasping what it means to talk like a human being.

The development curve in NLP

The earliest stage of unlocking our writing is the ‘Bag of Words’ stage, or the Syntactics Curve. The analysis engines can recognize words: lots of them, at speeds much higher than human readers. They report on frequencies of occurrence as well as on predefined patterns that emerge.

The next step is what has been dubbed the ‘Bag of Concepts’ stage or the Semantics curve. Here, engines can understand words not just in terms of recognition but also a beginning of meaning. Rather than merely recognising and counting a word like ‘father’, an engine would comprehend that it stands to ‘mother’ as ‘son’ does to ‘daughter’ – and then thousands of similar couplings to build more and more comprehension.

We have begun making progress here, using machine learning and concept databases to identify bigrams and trigrams (concepts composed of two and three words) and using ‘word2vec clouds’ to help engines put words in a context. We use machine learning to represent relationships between words as vectors, a very important technique in NLP that only became mainstream in the last 4-5 years (see pictures). But the hassle doesn’t end there.

The next - and presumably final stage we would call the Bag of Narratives, and it would involve an engine understanding the concept of Christmas for example - and how ‘egg nog’, ‘presents’ and ‘Father Christmas’ tie into these (without concluding that Father Christmas is an actual family member!). This stage is projected well into the 21st century, as you can derive from the diagram above.

Where does that leave market research?

Realising the immensity of the challenge in NLP opens our eyes to what is possible today in market research. How do we use online conversations today to uncover meaning and assess consumer sentiment?

This is where the acquaintance and commercial application of NLP comes to play: The art of social listening. Here, the raw power of the analysis engines we have at our disposal has been crucial in arriving at some fascinating observations which I’ll share with you below.

Talking ratings

Online ratings have always been hugely important, but these days the game is changing. Take (online) retailers, like Amazon, Best Buy, or Walmart, representing huge sources with millions of reviews posted. We’re seeing ratings creep closer together across the site in recent years: relatively more products are seeing ratings averaging 4.31, 4.32, 4.30, rather than a wider difference of 3.1, 4.8 or such.

The trend has important implications. Firstly, the bar has apparently been raised – with many products ranking positively. It also challenges brands to exceed the average, if the average is already quite high. Finally, it means trends and patterns are harder to single out, because a minor drop in ratings can already cause a major drop in ranking.

Talking emotions

We’ve made significant improvements in assessing the emotion behind a review in areas of the internet where no quantitative ratings are given. The words ‘fantastic’ and ‘awful’ are dead giveaways of course, but the human lexicon of delight and frustration are flowery, creative and quite rich indeed. Today, in full appreciation of the ‘bag-of-words’ stage of NLP, we use an all but exhaustive lexicon that spans the full range of human emotions in written form. ‘Jubilant’, ‘dismal’ or ‘spiffy’, anyone?

Talking price

Price talk is extremely important, but very tricky as well. You’d think it’d be cut-and-dried simple: people dislike expensive stuff and they like things cheap. From this you’d imply that where people talk price, they’ll either give high ratings for low prices, or low ratings for a high price. Right? Wrong. People tend to value a hefty price tag. It gives them a sense of status and quality. Does this mean all wine makers should raise prices as much as they can? Yes, and no: Price on itself is meaningless. A price tag makes sense if there is a reference: For wines to be considered expensive, there must be a bunch of cheaper ones around. Let alone budget restrictions - at the risk of bankruptcy - and our very personal, very situational behavioural, emotional and attitudinal traits towards pricing: Price is quite a nut to crack for computers.

Talking about learning curves

When researching reviews for a personal care brand, we discovered an interesting curve in ratings and reviews. It appeared that early ratings were often relatively negative, for the first couple of weeks after purchase. After that period, they improved significantly. The learning curve for the product thus emerged from the ratings. The insight it uncovered, was that the product takes a little time to appreciate. Which prompted our client to change their packaging communication.


There is massive progress been made by the relentlessly dedicated NLP community to creating smart engines that get better and better at understanding human language. What we learned is, that were you to successfully apply this technique and drive business performance, you should start with the right questions first. Then, identify the sources needed and the appropriate tools for running the analysis and coming up with meaningful findings. Lastly, hire market researchers with expert knowledge to turn the output into insights.

That latter artful discipline will be the subject of the next Riding the Curve – we hope to see you back here, then!

More interesting articles
Page 1 Copy 14

Market research is a) magic, b) madness or c) all of the above.

Page 1 Copy 14

Why all Shampoo smells like Fear to me

Page 1 Copy 14

Robots can read. Soon.