Don’t over hype social listening – it’s useful not revolutionary

The other day, I found myself, yet again, marvelling at the “that’s bleeding obvious, but so true and I wish I had thought of it’ wisdom of marketing professor, Mark Ritson.

He had written an article recommending that the best way to get a top job in marketing was to pretend to buy into vogueish, digital bullshit, rather than correcting your potential employer and giving a more balanced strategic perspective.

“To put it more bluntly: if you are a proper marketer, your brain might answer a recruitment question correctly but you will consequently lose the role to a lesser marketer. So, ignore the technically correct answer and go with the vocationally prudent one instead.”

His premise – that marketers have been swept away by digital tactics, and have lost the plot when it comes to more rigorous strategic thinking – rings very true with me. The expression ‘digital first‘ marketing is the epitome of the tactical tail wagging the strategic dog.

One of the areas of digital hype in Ritson’s firing line is social listening:

“Social listening is a very cool, very unrepresentative real-time barometer of brand sentiment that you should look at but it should never be more than 5% of your insight pool “

This area – sometimes called online anthropology (cool eh?) has long fascinated me. I even attended the BrandWatch annual conference specifically to learn everything about social listening, so I could wow my friends with the most current buzzy techniques. I was disappointed, nothing genuinely compelling here, but concluded that I just didn’t get it. Simply not cool enough to capture the zeitgeist.

Then a couple of ears ago, I was working on a marketing campaign in a therapy are that was new to me – autism. In these situations, the first thing to do is to furiously hoover up every bit of research, medical coverage, editorial and opinion around the therapy area, so I immersed myself in all things autistic.

One research source, commissioned at great expense by the client, jumped out at me – a study which beautifully segmented the audience of autistic kids, teens and their parents. It was packed with insight. It described a spectrum of perspectives ranging (and I’m not remotely doing it justice here) from those who had limited horizons and were largely defined by their ‘condition’ to others who seemed to excel in many ways and could almost be said to treat autism as their ‘super-power‘.

For the purposes of developing an advertising strategy, this was a compelling piece of work. If the ‘super power’ angle was tenable, and if it represented an aspirational, yet realistic proposition, it was a massively fertile creative territory. The agency was salivating at the prospect. A host of exciting storylines rapidly presented themselves – the narrow line between madness and genius, the unsung hero (tortoise) who quietly excels ahead of the flamboyant charlatan (hare), the power of concentration to go beyond what was thought possible etc.

Being a bit of a research nerd, I was interested in the methodology for this segmentation, so I dug a bit deeper. It turned out the segmentation was based on a deep-dive, social listening exercise. What does that mean? It means they ‘scraped’ a large number of conversations happening online, around autism, and set about grouping the attitudes expressed, into segments. Nothing wrong with that – it was incredibly insightful, yielding some fascinating angles.

But there’s a huge assumption here – that the opinions expressed in this ‘scraping’ are in some way representative of the population concerned, in this case autistic kids, teenagers and their parents. A moment’s reflection confirms this is likely the opposite of the case. Online conversations, of the kind elicited by the researchers, almost certainly reflect a sub-set of our population who are atypical. They are by definition high-performing, literate, tech-savvy and opinionated. A brief dive into all the other autism literature tells us that this is very far from typical.

So when Ritson says social listening is “a very cool but very unrepresentative real-time barometer of sentiment” I can absolutely concur.

Unfortunately, we never got to create a campaign about autism as a super-power. It would have been a fantastic creative opportunity. But it would have been based on a tiny, unrepresentative insight based on a small, atypical sub-set of our audience.

Narrow escape or missed opportunity?

Can’t decide, but when it comes to social listening, I don’t feel so bad about ‘not getting it’ any more. Having said all that, it’s a technique that can add something really valuable to the insight armoury, For example, when good tracking study researchers report their findings, they often supplement the survey data on attitudes with concurrent metrics covering online sentiment. It makes perfect sense.

If you don’t believe the hype, this digital revolution can be genuinely helpful.

Don’t ask me

Sometimes research can be your worst enemy.  “How so?” I hear you cry.

Consider this example.  Many newspapers are confronted with a conundrum.  More people are reading the paper (mostly online), but fewer people are buying it (paper edition).  Revenue falls while the demand for more content continues to rise.  One response is the pay wall.  But this threatens to marginalise your title and it appears to give other free offers a big advantage.  The Guardian’s response has been to appeal for voluntary contributions through different levels of ‘membership’.


Given the paper’s history and its loyal following among those who share a broadly liberal ideology (myself very much included) this seems pretty sensible.

But, if we look a bit further, here’s the ‘rationale offered to prospective members:


Let me draw your attention to the bit about revenues falling fast.  Behavioural economics and the writings of Mark Earls (‘Herdmeister’) give us some clear learning around this.  Telling people that others, like them, are deserting the paper in their droves is the worst thing you could possibly say.  Much better to say that more and more people are subscribing online.

There’s a significant body of experimental evidence for this.  Richard Shotton quotes a fab example from the Arizona national park, in which visitors were taking away start wood in alarming quantities.  The park authorities set up an observed section and tested three approaches: i) said lots of other people were taking wood home and this was depleting the area so please don’t take wood home ii) said most people don’t take wood home, so please don’t you take wood home either and iii) control saying nothing.  The second approach asking people to fall in with ‘herd’ behaviour was by far the most successful Indeed approach (i) actually seemed to encourage the taking of wood.  This result is echoed in lots of other studies.

So why is The Guardian – a smart operation with sophisticated marketing – getting it so seemingly wrong?

I would venture to suggest it’s because they have relied on asking people questions – using market research.  I know this is true because I have been a respondent in some of this research too.

In short, if you ask people what to do, they will give the common sense answer.  If you want me (and others like me, who support you) to remedy this revenue problem, it seems sensible to explain the situation and, because we want the same thing, we’ll offer our help – we’ll subscribe.  But we won’t.  because behaviour doesn’t follow those rules.  Marketing doesn’t work by explaining what I need you to do which prompts you to behave accordingly.  People don’t act to maximise their utility and they don’t do what you ask them to do – at least not directly.  It’s oblique.  The world of brands and marketing doesn’t make sense if you treat people as rational utility-maximising agents.  Most of the world’s most successful brands wouldn’t exist (and we’d be poorer for it).

Sadly, modern marketers seem to be using market research as a simple mechanism for asking people their opinions.  What’s needed is a proper model of influence – a set of hypotheses around which stimulus will provoke what response. Then, based on the answers we get to our questions, we diagnose the solution.  We don’t simply ask people what to do then do it.

I do hope the Guardian gets over this myopia.  We’d be much poorer without it.


Generational generalising

Millenial bollocks

We read a lot these days about Millennials.  My HR team say they’re revolutionising the world of employment because they have different hopes and dreams to other cohorts.  I say bollocks.

Millennials are in a younger cohort.  But mostly, they’re just younger.

Cohort generalisations are seductive.  They allow us to make sweeping generalisations which lend themselves to neat stories.  Boomers were the first generation to enjoy contraception and they gave us the swinging sixties.  They carried those hedonistic values through and now they refuse to grow old gracefully.  Their children – Generation X – rebelled against this rebellion and adopted values which are more staid and rigid.  Their children are Gen Y who…. blah blah blah. I’m not buying it.

Why?  Because twenty years ago, we were saying exactly the same things about Generation X, when they were in that pre-family lifestage.  They too appeared to value authenticity.  They were less materialistic.  They would save humanity.  They wanted experiences over possessions.  Simple pleasures.  They sought control over their lives.  They were getting used to the idea there were no jobs for life and embracing a future of self-determination.  Climate change hadn’t hit the headlines, but if it had, they’d be in the vanguard of fighting it.  They had ideals.  It’s called being young.

So why are Millennials more accepting of their ageing looks?  Because they’re not ageing.  They’re young.

Worse, this whole conversation betrays a misunderstanding of how segmentation works.  When we use consumer segments in marketing the idea is to start with a behaviour we want to explain – say who does or doesn’t buy our brand – and then relate this back to the various dynamics that influence it.  Could be demographics, psychographics, generational cohort, whatever.  It doesn’t work the other way around.  Define a group which may or may not be meaningful and then try to describe how they’re different to another group.  This can be interesting, but it’s always likely to lead to this kind of spurious analysis which doesn’t help much.

It also flies in the face of all the evidence that age is increasingly a very poor predictor of behaviour.  We really don’t act our age any more.

Add to that the now unfashionable idea of need-states, which reminds us that we drift in and out of our various roles, making generalisations misleading.  In the parlance…the same person is more different on two occasions than two individuals are on the same occasion.

So after this eminently balanced and reasonable piece, I say to you:  “Millennials, my arse”.






Bad research

Conducting market research is a dying art, it seems.  I’ve seen a lot of really bad research recently and it worries me.  I’m a bit sad like that.

Here are just a couple of examples of questions from surveys commissioned by significant clients who probably should know better from significant research companies who definitely should know better.  I have loads more.  I’ve been collecting them.  There, I told you I was sad.

What’s wrong with this one?bad question 1

Just try answering it.  You can’t possibly know this information without having kept a diary on the subject for the 30 days in question (which would be a better way to get the information, no?)

What about this one?

bad question 2

This one’s a bit more insidious, because you might be tempted to have a stab.  But it’s obvious we don’t know how far we are influenced by these things.  Imagining we do is a bit worrying.  Any worthwhile basic training in research would tell you that very early on.

In real life, respondents will do their best to answer, but the data will be nonsense.  It will be used as though it were the truth.  Oh dear.

The problem is that marketers in some companies have seemingly lost the instinct for research. Like understanding that it’s more about designing experiments than it is about asking questions.  It seems the modern way is to write down what question we need to answer – like what influences our target audience – and simply ask them.

Come on chaps.  We can do better than this.