From Utilty to Futility: Demographics in Marketing

Our friend Peggy Ann Salz over at M Search Groove mentioned the diminshing utility of using demographics in marketing segmentation and targeting. I wanted to return to this topic, and argue loud and clear, that the evidence is overwhelming, that we (marketing professional) have experienced in the past few years a total shift where customer demographics have gone from utility to futility. Yes, futility. They are now counter-productive. You, reading this blog, need to start to remove all references to demographics in all of your company marketing.

What is demographics? It is consumer data based on statistics relating to our lives. Our age, our marital status, our address (typically the zip code/postal code) etc.

Demographics were a powerful way to segment and target our marketing in the last century, when we had nothing better. There are often significant statistical correlations with our address or marital status or educational achievement level or our age, to our propensity to spend on a given product or service. For example, a married home-owner with kids, is far more likely to buy home and life insurance than a single young adult or indeed a teenager. Equally a young single adult is more likely to go out partying on the weekends in the hottest clubs with loud music etc. A person with an address on Manhattan's upper East Side is likely to be very wealthy, and ten miles up North still within the city of New York, in the ghettos of the Bronx, the people living there are very likely to be very poor. If you want to promote your newest model BMW or Cadillac Escallade, it made sense to target those in Upper East Side Manhattan rather than those in the poorest neighborhoods of the Bronx.

This was all very advanced marketing in the analog era of the previous century. It was far better than Henry Ford's one car for everybody ("Any color as long as its black" about the Model T Ford, which Henry Ford actually never said.)

For a while early in this decade there were some false promises of a change. Now we know, and the change is total. In the digital interactive age of marketing, wen have seen the total shift of demographics from utility to futility.

FALSE NEGATIVES

Demographics were far better than nothing, but they were also very imprecise. When we had nothing better, it made sense to use demographics in marketing segmentation and targeting. But take me for example. What good does it mean to know that I am a man? If you sell ladies underwear, it is certainly a reasonable assumption to think, that men don't buy your product, so to focus the advertising to women, who are more likely to buy.

But this introduces "false negatives" into your customer prospects database. Tomi is a man, therefore is not customer. Now, what if I am a cross-dresser (a man who likes to dress in womens' clothing)? The demograpic data will not identify that. It is very likely that the cross-dressing man would go to great lengths to hide this tendency. Now, granted, there probably are not very many cross-dressers as a percentage of all men, but even if its one percent or one tenth of one percent, if you target by demographics, your segmentation system excludes real prospects for your business.

Its not the only false negative. It is also possible, that I am in the process of a sex-change operation. Yes, this is probably even more rare, but they do exist. What was accurate once, as a demographic measure, has in reality changed.

But we don't need to be that sensationalist, I just wanted to make a point. In reality, many men buy lingerie to their wives and girl friends as gifts. Why not? if I am excluded as a customer, while the demographic facts are correct, and the analysis fits the majority of my gender, there is a significant size of a demographicminority, which is now locked out, as a false negative.

But note, if you pursue demographics, you are very likely to use proxy data for determining the gender of the prospective customer. Like Mr Charles Smith is very likely a man, yes. So if you are in Finland, and you see Tomi Ahonen as the name, you can rather safely guess, that the person is a man. But what if you are in Japan? Tomi is a woman's name in Japan. Now, if you are in New York, and are of Scandinavian ancestry, you probably would correctly identify me as a man, but if you are of Japanese ancestry, you would almost definitely classify me as a woman.

But wait, the demographic data may be collected accureately, but the entity has changed. The house is in the name of Mr Charles Smith, but he has recently split up with his (non--married) girl friend, and it is now the girl friend who lives in the house, as Mr Smith lives with his new girl friend in Los Angeles, as he waits for his house to be sold. but the behaviour in that address today, is no longer that  fo Mr Smith, it is that of his ex-girlfriend.

The exact same applies to all demographic data. Take age. Yes, we know that the youth tend to party and consume various youth services and products. And yes, on the whole, as we get older, we tend to "mature" and "settle down" and "get a life". Married, kids, home all that. Fine. Obviously there are some who never grow up (more of us among men than women ha-ha). But the more obvious part - a married 39 year old woman, two teenager kids, employed. Suddenly starts to behave exactly like a teenager 19 year old girl, same parties, clubs, nights out, text messages into the wee hours and all that etc.

What happened? She is suddenly single again, probably in the process of going through a divorce or separatation. If your system "discriminates" by age, you are missing this type of customer, totally.

Again, this is not a criticism of classic marketing textbooks. They were right, in the last CENTURY about trying to learn as much about customers as possible and then try to offer them targeted services and products. Today we have something FAR BETTER.

BEHAVIOR BASED SEGMENTATION

The only thing which matters is actual behaviour. How do we act. Not what we say we want, not what we say we would do, but the actual behaviour. And how can we capture that? Only with digital interactive services. And the most complete and rich data we get through engagement marketing methods, the tool which was first described in our book Communities Dominate Brands, the signature book for this blog.

All marketing communications will soon be behaviour based. All of it. Look at Google Ad Words. It is the first mass-market ad system that offer contextual ads. Not based on my demographics (on the whole) but based on what I did just now. If I search Nokia, it will offer me some links to phone websites. if I searched for BMW, I get cars, if I searched for lingerie, I get Victoria's Secret etc. Google does not care whether I am a man or woman searching for lingerie, if they've sold that top spot to VS, I get the Secret there, trying to lure me to visit their site, totally regardless of whether I'm a man or woman, married or single, 18 or 80, college educated or not.

This is only the beginning. You think Google Ad Words is radical, cutting edge? It is not. The internet is the 6th of the mass media. It is better than the first 5 legacy mass media, yes, and the internet has inhereant advantages over TV or radio or print or cinema etc, but mobile is the 7th of the mass media. A far more powerful platform to deliver customer info than the internet. Why? Because mobile can replicate everything that the internet can do, but mobile adds 7 unique elements that the internet cannot replicate. Mobile is a far more compelling mass media channel than the internet. And before you start to argue, Yahoo CEO Terry Semel, the next internet is on mobile. Google CEO Eric Schmidt says the same, the next internet is mobile.  Why did Apple Computer drop "computer" from its corporate name when they announced the iPhone in 2007? Because the future of the PC based internet is mobile.

Understand what I mean. The TV cannot replicate the internet. A newspaper cannot replicate the internet. You cannot do Google Adwords through the cinema screen. But the mobile can replicate  everything that the internet can do. Everything obviously including Google Adwords. Not just as contextual search, but mobile can take the concept much further, like I showed in the Flirtomatic First Face case study as one of 50 case studies in my eBook Tomi Ahonen Pearls Vol 1: Mobile Advertising.

In the future, all marketing communication will be targeted by our actual behaviour, not needlessly limited by any archaic measurements of demographics. Oh, and that zip code based segmentation in New York City? Yes, there are plenty of wealthy people in Manhattan, on the Upper East Side. But there are also individuals living there who can't afford any car, far less a BMW or Escallade, take the live-in maid or au-pair for example. You can have false positives in demographics. And as to the ghetto in the Bronx? yes, it may be a stretch, but there can be drug dealers and pimps and other criminals, living in he ghetto, who have a desire to have their "bling" and be quite capable of making purchases equal to or even exceeding that of millionaries on the Upper East Side. So we get a false negative.

I don't mean these as typical but every lost sale is your marketing actually hurting you, if better methods exist. Now, I am not a cross-dresser nor contemplating a sex-change. But honestly, I am a fan of rap music. I am a 49 year old white dude from Finland. I am TOTALLY off the demographic radar to get any marketing about hiphop music. And I don't dress like them, I don't own a pair of blue jeans and I don't own a pair of flashy tennis shoes. I dress even on casual days in my designer suits and ties (and hats). If your marketing system (segmentation/targeting) makes any assumptoins about me, about my age, my ethnicity etc, it will never know to sell me some new rap music. And if your system makes assumptoins based on what most rap music fans wear, you'll never sell me another Armani or Canali. Or if you are Armani or Canali, and assume something based on my music preference, you miss me again, as a customer.

But if you remove all demogrphic filters, and only measure actual behaviour, you will find that yes,. I do buy designer clothes -suits and ties for men, not teeshirts and jeans, and yes, I do listen to rap music (and buy it), and even occasionally buy some ladies lingerie - as a gift obviously, as if you track that tendency over time, they would be for women of a different size as per my girl friend of that time.

Demographics were a powerful marketing tool in the last century. Now we in marketing need to get rid of that outdated concept, and start to market based on actual behavior. We ahve some rudimenatary methods on the interent to capture customer data. That ability is far more enhanced on mobile. And as we learn to deploy engagement marketing methods, we will get vastly more satisfied customers , as so many of the recent mobile based campaigns have proven, with 30% response rates and above! not click-through rates, response rates.

I am not in any way suggesting "spam" unsolicited unwanted intrusive ads on mobile. That is equally dumb. No, I mean the newest engagement marketing methods, using mobile and the targeted and permission-based advertising. To see more, pick up a copy of my newest hardcover book, Mobile as 7th of the Mass Media.

Comments

[tim harrap]

Hi Tomi Great blog for Eddie Izzard!
At a slight tangent to your message I twittered yesterday the question, "Many Google Zeitgeist searches trending downwards over 5 yrs. What does that mean?". If you search for big theme's like "England", "Japan" the graph presents a downward movement. Is this evidence of the global communities improved experience with asking search questions i.e. deeper more detailed queries? Is this increasing granularity dissipating the brand identity of core themes?

[Roi Perez]

It's true that demographics are a not always the full picture. We use them as a handy way of segmenting audiences. Your right that they are proxies because we don't know the actual behaviour. But that doesn't make them futile. It just makes them imperfect.

[Paul McEnany]

I agree with the gist, for the most part, but I probably wouldn't go so far as to call demographic info worthless. Mostly because behavior is fantastic in capturing existing interest, but equally imperfect when generating new interest. So if you're comparing to direct marketing tactics, I'd probably agree, but not always for brand marketing. I guess it's really about using all of the cues we have, whether that's behavioral or demographic, to make our best guesses.

[Tomi Ahonen]

Hi Tim, Roi and Paul

Thank you for the comments. I'll respond to each individually

Tim - thanks. And good point, interesting experiment. I would think, that at least part of that trend is an increasing sophistication by users, learning to use more targeted searches maybe?

Roi - good comment and I totally understand what you say. Yes, demograpics are used (obviously, this blog and our themes are so new, that almost no university MBA courses yet teach marketing "beyond demographics" - thus by necessity, what you are reading here on our blog (and in our books) and debating here with me now, you are truly among the first thousands of marketers on the planet to come across this thinking, and it will take a decade for marketing to "grow up" and understand.

So from a factual "today's status" angle, you are totally right, we do that today. What I am arguing is, that using demographics is a COUNTER-PRODUCTIVE tool today. It will actually HURT more than help. It will create completely useless distinctions and worse, it will eliminiate market opportunities for your company and brand (in most cases, in most markets).

Now, there are some specific exceptions where age for example is very important - children's food for example. They have acutal age brackets, this food is for those babies 6 months to 12 months old, etc.

Age is also relevant to matters where there is a legal requirement for restrictions to purchase, such as cigarettes, alcohol, adult entertainment, gambling, credit cards, etc. For these rare exceptions there is a specific cut-off point for a given age. But most of our services and goods that we sell, any demographic data is truly counter-productive.

A 37 year old married man or woman, who is not yet divorced, but is in the process of splitting from the spouce, that person will start to behave like a single young adult. But is far too old for it, to fit the demographic. If you go by married status (wrong) or age (wrong) you get an error in your targeting. Or else, if you count your "youth" segment to go up to age 37, it is that meaningless to again be useless in the targeting we should be doing in marketing.

Demographics were an excellent tool last century when we had nothing better. It is now an outdated tool and should be abandoned (with some obvious exceptions such as baby foods and the legal age of being an adult)

Paul - I hear you and I appreciate it that you see a lot in it. You make a great point that for generating (genuinely) new interest, the existing usage is not available and other methods have to be used. Then I'd say use psychographic info (our profiles based on our preferences, lifestyles etc) but again, to exclude demographics totally from the mix. There is no doubt some statistical correlation for example with younger users being willing to experiment with new products and services more so than older users. A statistical correlation yes. But there are so many young users who are not willing to experiement, and so many older users who are still willing to experiment, that by looking at their psychographic profile, gives a far better measure than their age (or marital status or gender or zip code etc).

Thank you all for writing

Tomi Ahonen :-)

[mitzi]

Hi Tomi,

Nice blog. Nice post too. Thought provoking. Initially, felt for paul in his comment about not totally disregarding demographics in the mix. In your reply to him you said to use psychographic info BUT TO EXCLUDE DEMOGRAPHICS TOTALLY FROM THE MIX, at first I was thinking that to do so would actually do the opposite of maximizing possible consumers because demographics could also still serve as a guide in terms of identifying where marketers can find these targets (i.e, schools for univertsity students, etc).This however, with the thinking of includeing more than one demographic variable in identifying the target (i.e, a wider range in age, say 19-40 if we talk about the woman who starts acting like a teen again, or breaking it up into two age ranges of 16-25 and 35-40 for example)

But as I start to write this comment, i realize how much more effective it is indeed to paint a more accurate picture of the target and indentifying where we can find them if we do take out the demographic.

Regardless of age, gender and socio economic status, it's more accurate to identify what kind of spender the person is, where he/she usually hangs out and what types of activties are they into. Because identifying a combination of these (even without the demogaphics) will definitely lead you to them.

This is a great post. I'm glad I chanced upon it as I'm about to venture into a big trending & insighting exercise for our ad agency, being the lone planner =(.. This helps me in streamlining the kinds of information I'll be collecting and sifting through for this endeavor.

many thanks! this is great!!!

[mitzi]

oh, and i have a follow up question/concern/afterthought though. I think what including the demographics contribute in the process of identifying targets is in its being the most easily measureable 'variable' in identifying targets. Of course size of the target is usually important for more sales. Considering on behavior & psychographics alone may not give an accurate size of the segment, which might be a point against identifying a certain behavior based segment to focus on.

Unless there is an existing means of quantifying it other than a big quantitiative clustering study.

so my question is, how can you reconcile the need to identify the size of a segment/target with just considering psychographic/behavioral based segmentation?

[Tomi Ahonen]

Hi mitzi

And thank you for both comments.

Good stuff, clearly you're thinking about it and I'm so happy our blog has helped a fellow colleague re-examine this world of marketing and segmentation; there might be a different world out there, and even if we don't have it totally right, I hope others will join and help find that new future, ha-ha..

Now, specifically to the second comment. Very very good indeed. Yes, its easy to do ages. Its easy marketing. Its easy segmentation.

Real professional marketing, really understanding your customer base better than your competitors - that is not easy..

I would put it to you, that a good job in re-defining the segments (and yes, explicitly forbidding the use of classic demograpics) would potentially yield far more powerful segments - and best of all - will wreck the abilities of your competitors to respond..

But yes, it takes effort, a lot of work. So, for example. In my marketing segmentation workshops I like to talk of the ten changees that we as humans go through - each of which relates to how we interact with technolgoy in general, and mobile in particular. Each change results potentially a change also in our usage patterns. I'm talking of learning to write (about age 7) or reaching puberty (age 14) or moving out of the home or getting a first child, or retiring etc. (Not everyone goes through all ten, and not always in identical order, and sometimes we can do the same life situation a couple of times - some people get married and divorced many times etc)

So, a couple of those categories include a status of "single". It can be youth single ie moving away from the parents' home or it can be "again-single" ie after break-up of a live-in partner, or divorce etc. We could do a "post-modern" segmentation where one category is singles, another is adult partners (married or living together) and a third is parent with kids (again, ignoring marital status). Each of these three will yield quite dramatically different phone behaviour patterns, very similar within the segment and radically different to the two other groupings.

I don't mean that this is the ideal segmentation. I use this only to point out, that we can have the "singles" category that can very legitimately have 16 year olds who have moved away from home, and 43 year old "again singles" and anything in between, and ignore the age. Obviously if we measure the ages, we'll probably find that the majority of that group is bewteeen 18 and 25 or so, wherever the age is that people start to co-habitate and make babies etc...

Hey, one more thing. Nothing wrong, AFTER you've done your segmentation for competitive advantage, to map it against "classic" models, to help those senior managers who are not up to the latest theories. Nothing wrong with reporting internally an age-based break-down of the service usage.

Only, that the marketing should not be driven by such archaic models, ha-ha

I hope this helped. Thank you for coming back mitzi and let us know how your own thinking evolves in this space. Please also pay attention to Jonathan MacDonald's Every Single One Of Us initiative, and David Cushman's thoughts in this space (as well as our Alan Moore's obviously)

Tomi Ahonen :-)

[mitzi]

Thanks so much for your response Tomi. Truly appreciate it. I also did consider thinking about segmenting by 'life-stages' because it has definitely evolved over the years, not as clear cut as before. People can now experience and equivalent of 2 lifetimes (like one's life with a new spouse after a divorce could be characterised as a totally new lifetime altogehter -- that's just how I've observed it tho), with the kind of lifestyle one can have now.

Anyways, I am now thinking of how I can go about segmentation beyond demographics to come up with more succinct target descriptors. It's now also a matter of which kind of segmentation will i start with which will be applicable to the range of brands that we handle. Hope I can also get back to you soon with a definite feedback on where this has led me.

I am going through Every Single One Of Us now and finding it very rich in information and insights on brand building. And yes, I will continue to read you, David Cushman & of course Alan Moore ( I just hope I won't get too overwhelmed with the wealth of information & insight :)

may thanks!

[cheap jewelry]

I like what you have said,it is really helpful to me,thanks!

[top business opportunities]

Excellently written article, if only all bloggers offered the same level of content as you, the internet would be a much better place. Please keep it up!