Emotional Branding Pays Off illusion

Behavioural loyalty is strongly correlated with propensity to agree to ‘brand love’ survey questions but…… most lovers still buy other brands, and most of a brand’s buyers don’t love it.

John Rossiter & Steve Bellman (2012) “Emotional Branding Pays Off – how brands meet share of requirements through bonding, commitment and love”, Journal of Advertising Research, Vol.52, No.3, pages 291-296.

Rossiter and Bellman (2012) purport to show how consumers’ attachment of “strong usage relevant emotions” to a brand affects behavioural loyalty. All they actually show is that if you buy a brand more then you are more likely to agree (on a market research survey) to positive statements about that brand. We’ve known for 50 or so years that people do this – that stated attitudes reflect past behaviour. Or more succinctly: attitudes reflect loyalty.

Specifically Rossiter & Bellman showed that people who ticked “I regard it as ‘my’ brand” tended to report that this brand made up more of their category buying (than for buyers who didn’t (regard it as their brand)). What an amazing discovery!

“I regard it as ‘my’ brand” was, by far, the most common of the ’emotional attachments’ they measured – with about 20% of the buyer bases of particular brands of beer, instant coffee, gasoline, and laundry detergent ticking this box. It was also most associated with higher share of requirements (behavioural loyalty). I’m not surprised because it is most like a direct measure of behavioural loyalty. If I mostly buy this brand of coffee then I’m much more likely to tick “I regard it as ‘my’ brand”. If I buy another brand(s) more then I’m hardly likely to tick that I regard this one as my special brand.

So reasonably we’d call this question (“I regard it as ‘my’ brand”) a measure of reported behavioural loyalty, and so it would have to be highly associated with any other measure of reported behavioural loyalty. But Rossiter & Bellman in classic sleight-of-hand call this question a measure of “bonding”, which they say is a measure of an emotion (not a self-report of behaviour)! Naughty naughty.

On safer ground their measure of “brand love” was if brand buyers agreed “I would say that I feel deep affection for this brand, like ‘love’, and would be really upset if I couldn’t have it”. Interestingly, hardly any of any brand’s buyers ticked this box. Just 4% of the average beer brand’s (male) buyers, just 4% of the average laundry detergent’s (female) buyers, 8% of the average instant coffee brand’s (female) buyers, and a mere 0.5% of the average gasoline brand’s (male) buyers. Restricting the samples to the specific gender that represents the main weight of buyers reduced the proportion of light and lower involvement category buyers. This would have increased the incidence of brand love yet it was still about as low as is possible. Rossiter & Bellman wrote that these results “reveal the difficulty of attaining strong attachment-like emotions”. Hmmm, well yes and these results also reveal how successful brands largely do without brand love.

With so very few of any brand’s buyers agreeing that they feel deep affection for the brand we would expect the few that did would be quite different from the average. We’d expect that they would be the heaviest, most loyal in the buyer base. And these lovers did report higher behavioural loyalty though it was far from absolute (100% share of category buying). In fact, ‘lovers’ only reported buying the brand about half the time (50% SoR). Behavioural loyalty is strongly correlated with propensity to agree to ‘brand love’ questions but…… most lovers still buy other brands, and most of a brand’s buyers don’t love it.

Rossiter & Bellman interpret their results differently. Their article title says emotional branding pays off, even if the article does nothing to investigate marketing practices. They act as if they are unaware of the research going back decades that shows, over and over, that usage affects propensity to react to attitudinal type survey questions (see Romaniuk & Sharp 2000). Instead, this single cross-sectional survey data is supposed to show that if marketers (somehow) run advertising that presents attachment emotions, then consumers will link these to the brand, and then change their behaviour to buy that brand more often than they buy rival brands. Rossiter and Bellman’s results show nothing of the sort, their clearly written article turns out to be highly misleading. Yet I fear that this will not stop many unscholarly academics citing the article, and many believers in this discredited theory citing it as evidence to support their blind faith. Beware of such nonsense.

Refreshing brand memories after a gap

I haven’t seen a Life Savers ad in ages, most probably because they haven’t been advertising – a lot of brands go “off air” for long periods.

Yet when I saw the ad (see below) the jingle bounced back into my consciousness – thankfully they are still using the slogan.

Advertising exposures that follow a long gap can be particularly powerful memory refreshers.

Unfortunately this is another factor that tempts marketers to go off-air for periods, when the real lesson is don’t bunch your exposures together (burst).

Think how much better effect Life Savers would get if instead of bursts followed by long gaps they just kept on advertising at very low levels. If they did Most of us wouldn’t see a Life Saver ad very often, but we would see them regularly if infrequently and each time they would have a tremendous refreshing effect.

With the burst and silence pattern we seldom ever see Life Saver advertising but when we do we see it several times close together when the 2nd, 3rd and 4th exposures don’t have anywhere near the refreshing effect as the first. That’s wasted advertising money that could have been used to reduce the long silence between bursts.

Get a hole lot more out of your advertising, don’t burst, don’t go off-air. Spend less, for longer.

20121216-105651.jpg

What’s wrong in the house of academia, and a suggestion how to fix it

Presented to the Australia & NZ Marketing Academy Conference December 2012.

Marketing has a small ‘crisis literature’ where academics themselves bemoan the lack of real-world importance of academic research into marketing. For example, “Is Marketing Academia Losing Its Way”, Journal of Marketing, 2009.

Back in 1989 John Rossiter documented the growing gap between marketing scientists and consumer researchers even though they were supposed to be studying similar things. He warned the consumer researchers in particular that they were in danger of retreating into an Ivory Tower detached from the empirical findings regarding mass buying behaviour. Yet the trend continued unabated.

I myself, and colleagues, have had articles rejected from good journals when they chiefly documented a substantive finding about the world. My ‘favourite’ was when the editor of Marketing Science wrote to Jenni Romaniuk and myself about our work documenting the 60/20 law (ie it’s not 80/20). Effectively he said, “great stuff, I’m going to use this in my teaching, but we can’t publish it in Marketing Science because the journal tries to feature leading edge analysis whereas what you did was simple and transparent”. An open admission that Marketing Science is really a journal about engineering above science.

Yet around the same time the Nobel Prize for Physics was awarded to two Russian scientists who found a way of producing graphene, a single atom thick layer of carbon, this potentially extremely useful material had once been thought unlikely to exist in the real-world. They isolated graphene with a simple technique using common household sticky tape. Placing bulk graphite between two sheets of Scotch Tape they simply repeatedly pulled the tape apart removing layers of atoms until they achieved graphene. A colleague remarked that it showed you could still win a Nobel Prize “for mucking around in the lab”. In physics there is still respect for substantive discoveries.

The defence or excuse from marketing academia is that we have been placing our emphasis on rigour over relevance. But recent shocking findings in marketing academia, other social sciences, and even medical research have exposed the myth of improved quality. There have been been some high profile examples of scientists disgraced for falsifying results (including in marketing), while 10% of psychologists admit to falsifying data (but they presumably evaded discovery), and most admitted to sometimes practicing dubious practices like selectively reporting the studies “that worked” (and hiding those that did not support their hypotheses. Relatively higher rates of dubious practice were found among neuroscientists and cognitive & social psychologists. What do you think the rates would be in marketing?

A recent analysis (Wilhite and Fong 2012) of the dubious practice of journals encouraging (or bullying) authors to cite other articles from the same journal reported that the Journal of Retailing, Journal of Business Research, and Marketing Science were stand-outs at the very top of the suspect list – and that’s not a list of only marketing journals. Indeed marketing journals stood out from other disciplines as being into coercive citation to try to manipulate their citation impact scores.

In medical research standards are undoubtably higher. Yet when pharmaceutical companies seek to replicate findings reported in medical journals in most cases it can’t be done – even though they try hard, after all they are hoping to make money from the discovery. Many of the findings for cancer drugs are highly specific to particular circumstances (e.g. patients with particular genetic profiles) but the researchers didn’t explore these conditions, they just got lucky with their so far unrepeatable finding.

In marketing Hubbard & Armstrong (1994) documented that academics hardly ever try to replicate findings. We simply assume they are true (or perhaps not worth bothering with). Not surprising perhaps, when replications are done they usually are unable to repeat the original result. The same sort of scandal has hit a number of famous psychology experiments. “The conduct of subtle experiments has much in common with the direction of a theatre performance,” says Daniel Kahneman, a Nobel-prizewinning psychologist at Princeton University. Trivial details such as the day of the week or the colour of a room could affect the results, and these subtleties never make it into methods sections of research articles. Hmm, what’s the difference between a result that is so sensitive to many trivial, unknown and unpredictable details and no result at all? Why should we care about a finding that only occurs in high particular circumstances?

This is a bigger problem than fraud and dubious research practice. We need to stop publishing one-off flukes and explore the generalizability of findings – where and when does a result hold? How does it vary across product categories, brand size, brand age, different types of consumers, at different times, and so on.

Even large and varied data sets are being wasted in marketing when results presented as an average across many different conditions eg “marketing orientation is associated with higher financial performance r=0.28”. This tells us little about the real world; the average may even not actually apply in any of the major conditions.

We need to explore generalizability or otherwise our ‘discoveries’ tell us very little about the marketing world that we are supposed to be studying.

And we need to stop prematurely building shaky prescriptive theoretical edifices upon these doubtful, poorly documented findings.

If we don’t carefully and thoughtfully (call it ‘theory driven’ if you wish) examine a finding and how it varies (or not) across conditions then we are stuck with findings that probably were one-off events – with no way of telling. Currently we have to treat our findings as either applying to one historic data set covering one particular set of conditions that may never be seen again OR a result that generalises to all product categories, all countries, all seasons. Both views are preposterous, something in between is far more likely but there is a lot of land “in between”, it needs to be explored.

The dubious research practices discussed above come partly from ‘confirmation bias’, the fact that (marketing) scientists want to find evidence to support their hypotheses – and they want “a positive result” otherwise they lack the motivation to publish, or the belief that they will be accepted by any decent journal. Brodie and Armstrong (2001) suggested researchers adopt multiple competing hypotheses as a way of overcoming this bias. A worthy suggestion, but those implementing it tend to simply have their favoured hypothesis and the opposite – and they still obviously want to see their favoured hypothesis supported. So I would like to make a different suggestion. Let’s use research questions with the words “when”, “where” and “under what conditions”. Rather than black and white “does X cause Y” type hypotheses let’s ask “when does X cause Y?”, “does X cause Y in highly advertised categories?”, “is X more a cause of Y in developing economies?”. This is the basic work of science, documenting patterns in the real world. When do things vary and when to they not.

If researchers use “when”, “where” and “under what conditions” research questions they are aren’t trying to prove a proposition, so they don’t have to worry about failure, so they should hopefully be less likely to tweak data and pick findings. Also, very importantly, researchers will be documenting something useful about the world because they will be exploring generalizability.

PS The Nobel Prizes for Physics are awarded in line with Alfred Nobel’s criteria “to those who, during the preceding year, shall have conferred the greatest benefit on mankind” which explains the worthy emphasis on substantive fundings. Alexander Fleming’s accidental discovery of penicillin is another example of the Nobel prize committee valuing important discovery over display of academic prowess.

REFERENCES

ARMSTRONG, J. S., BRODIE, R. J. & PARSONS, A. G. “Hypotheses in marketing science: Literature review and publication audit.” Marketing Letters 12, 2 (2001): 171-187.

HUBBARD, R. & ARMSTRONG, J. S. “Replications and Extensions in Marketing: Rarely Published but Quite Contrary.” International Journal of Research in Marketing 11, (1994): 233-248.

REIBSTEIN, D. J., DAY, G. & WIND, J. “Guest editorial: is marketing academia losing its way?” Journal of Marketing 73, 4 (2009): 1-3.

ROSSITER, J. R. “Consumer Research and Marketing Science.” Advances in Consumer Research 16, (1989): 407-413.

WILHITE, A.W & FONG, E.A. “Coercive citation in academic publishing”. Science 335, (Feb 2012): 542-543.

Stores compete for shopping trips

One of the fallacies of retailing is that stores compete in terms of selling items.  Of course they need to sell items to make money but they do that by attracting customers (or rather, shopping trips).

The more attractive a store is, i.e. the greater the share of shopping trips it wins, the more it sells.  And this is the real retail battle.

The more shoppers a store attracts the more brands will compete to buy their shelf space. In a way a store is like a TV station, it needs to attract viewers so that advertisers will pay a lot for the little bit of advertising space it has to sell.  Stores work to attract shoppers so that they can take a bigger slice of brand owner’s sales to consumers.

Store owners can easily lose sight of this.  They do strange things like try to trap consumers in store, making it harder for them to find the things they buy regularly in the vain hope that they will spend more in the store if they are trapped there for longer.  This is not a good way to earn repeat shopping trips.

Category management systems can send stores astray.  Each category manager wants only to increase sales of their category, and loses sight of the bigger picture, which is for the store to win a greater share of all the shoppers.

It’s the brand marketers (the store’s suppliers) who want to sell specific items.  If they want the store to stock their items, and to give better display space then they need to show that doing so will make the store more attractive to shoppers.  That their brand will help the store win shopping trips from the other retailers.  When you think about it this way one realises that price specials are just one very small part of making a store attract more shoppers.

Why stores stock many items that hardly sell

One line take-out: Each of us has a very different opinion on what the store should stock.  To win us all stores need a wide range.

The top selling 1000 items in a supermarket generate about half of its sales revenue. Which means that it’s vital that store managers make these items easy to see and buy – but that’s another story.

What I’d like to highlight today is that the other 30,000 or so items they stock sell very little volume.  This is what is sometimes called “the long tail”.

Stores try hard to weed out items that don’t sell.  So the typical store item does sell, but rarely. Stores are full of stock that barely moves while a tiny percentage of the items fly off the shelf.

This can lead marketing consultants to advise retailers to pare back their range to concentrate on the items that deliver most of their revenue and profits. Yet this range (and cost) cutting strategy often fails.  Unfortunately, it’s been encouraged by recent research (some of it flawed) on consumer confusion – research that mistakenly suggested that smaller ranges will increase sales.

It’s true that stores look cluttered and complicated.  The average household only buys a few hundred different items from a supermarket in a year. That is, they do a lot of repeat buying of some items over and over.  So each buyer is looking for a few things out of the 30-50,000 on offer in the store.  That makes shopping sound like a horribly complicated task.

So why on earth would consumers be attracted to stores that stock so many items – most of which they don’t buy?  One notion is that consumers like the IDEA of choice, that they are attracted to variety but once they actually arrive in-store they fall back on their habitual nature and existing loyalties.

There may be a little truth in this explanation but the real reason is that consumers are very heterogeneous in the items they buy.  Remember that all those items in the store do sell, each item has its buyers.  So given that each of us is buying only a tiny proportion of the items in store the odds that my shopping basket will share anything in common with the person in front of me in the queue (or anyone else for that matter) is very low.  As I often point out, if you look at what’s in the shopping trolleys of fellow shoppers you see that “other people buy weird stuff”, or at least that they buy different items from you.

The few items in common in any two trolleys are, of course, most likely to be those items that sell in large volumes.  These will appear in many more people’s trolleys.  Even so most of the items in our trolley will not be from the ‘top 1000’ and so hardly anyone else will buy them.

The Double Jeopardy Law tells us that an item with low market share will be repeat-bought less often than its rivals, but not dramatically less often, the main reason that it sells so little is that few people ever buy it.  Which means that many of the many low selling items in a supermarket are, in effect, being stocked for just a few consumers.  Some may even be stocked for a single household.  But for these few buyers these items are important, they buy them, maybe not that often (but that’s true of most things we buy), they know them, they are in their heads and their pantries – but not many other people’s.

Because we buy these items we like stores that stock them.  We each enter a store looking for “our stuff”. If the store doesn’t stock the things we buy we can sometimes find ourselves inconvenienced.  We want to see, and be able to find, the items of interest to us.  That makes a store attractive to us.  Fortunately for store managers consumers are extraordinarily good at filtering out all the brands and SKUs that aren’t in their personal repertoire and finding their brands.  Successful stores make this even easier for consumers.

So my point is don’t make the mistake of thinking that a store can do without 90%+ of its range.  Stores compete for shoppers, and shoppers vary enormously in what they look for, in what mental structures are in their head, in what they see.  Each of us has a very different opinion on what the store should stock.  To win us all stores need a wide range.

What’s Not to “Like”? Can a Facebook Fan Base Give a Brand the Advertising Reach it Needs?

Our earned media research article has now been published in the Journal of Advertising Research:

What’s Not to “Like?” Can a Facebook Fan Base Give a Brand the Advertising Reach it Needs?
Karen Nelson-Field, Erica Riebe, and Byron Sharp

Journal of Advertising Research, 2012, June, Volume 52, No.2
A marketer with a Facebook Fan base has at least some ability to advertise to that audience. What quality of reach, however, does this sort of “earned media” deliver? The landmark discovery by Andrew Ehrenberg of the negative binomial distribution (NBD) implies that the most effective advertising requires media that reach across both heavy and light buyers of the brand. This article investigates the buying concentration of the Facebook Fan base of two different brands (both Fast Moving Consumer Goods (FMCG) categories) and compares it to the brands’ actual buying bases. The buyer base of each of
the brands is distributed in the typical NBD, whereas the Fan base delivered by Facebook is skewed in an opposite pattern—skewed toward the heaviest of the brands’ buyers— making the quality of Facebook’s reach appear rather unappealing.

The danger of chasing market share and trying to harm competitors

It might seem odd for the author of “How Brands Grow” to warn against aiming to grow market share, but here I’m offering a reminder that growth should be an outcome of a strategy to grow profits – profits should not be sacrificed for growth, and especially not for the goal of harming competitors.

We all know that market share growth can deliver increased profits.  But we also know that it can also decimate profits.

It’s fine to aim for share gains, so long as the strategy is carefully developed so that the share gains really do deliver profits.  Because research shows that companies that focus on profits are more profitable, while companies that aim for winning market share from competitors are LESS profitable, and more likely to go broke.

Here is a short essay by Wharton and Ehrenberg-Bass Institute Professor Scott Armstrong which does a pretty good job of summarising his extensive research on this topic.

The Dangers of a Competitor Orientation

Question: Do profits improve when firms attempt to gain market share?

If you believe in the common wisdom of students, managers, and professors of marketing, the answer would be “yes.” However, the evidence tells a different story.

Fred Collopy and I summarized prior research consisting of nearly 30 previously published empirical studies. Twenty-three different laboratory experiments were conducted with 43 groups spaced over many years and countries. In addition, we analyzed 54 years of field data for 20 companies to compare companies that used market share as an objective versus those that focused only on profits. Our research extended over a decade. The results from all approaches showed that market-share objectives harmed profits and put the survival of firms at risk (Armstrong and Collopy 1996).

The paper was difficult to publish. Reviewers disagreed with our findings and seemed intent on blocking publication. They kept finding what they thought to be serious problems with the research. When we would respond to their criticisms with additional experiments, they became incensed. In all, it took about five years to get through the review process. In the end, an editor over-ruled the reviewers.

In a follow-up paper, Kesten Green and I described new evidence from 12 studies that were conducted since the 1996 publication. The new evidence provided further support for the conclusion that competitor-oriented objectives are harmful. In fact, there has been no empirical evidence to date to challenge this conclusion.

While our research has received much attention (e.g., 167 citations for the 1996 paper), it seems to have had little effect on what is learned in business schools.

In teaching the introductory marketing class to Wharton MBAs, I would present the results of this research. This proved to be upsetting to many students as it conflicted with their beliefs and with what they said they were learning in other courses. After one session in which I described this research, an MBA class representative came to me with the “friendly advice” that the students did not appreciate hearing about my research; they would prefer to know what is going on in the real world.

To illustrate the dangers of a competitor-orientation, I also used an experiential exercise known as the “dollar auction” (Shubik 1971). In this exercise, the top two bidders pay, but only the top bidder wins the dollar. Typically the bidding would start at a penny, then move up at an increasing rate. I always made money on the dollar auction. But in 1982, I had my most successful session when I received over $20 for my dollar. I have kept in touch with the 2nd highest bidder, Ravi Kumar, over the years. On a recent trip to India, Ravi reminded me of the name of the winning bidder – Raj Rajaratnam, a hedge-fund manger who was found guilty of insider trading in May 2011, and who is suspected of funding suicide bombers in Sri Lanka (New York Times May 12 stories starting on the front page). Apparently I failed to convince Mr. Rajaratnam that a competitor orientation is harmful to oneself as well as to others.

Professor J Scott Armstrong

References

Armstrong, J.S and K. C. Green, “Competitor-oriented Objectives: The Myth of Market Share,” International Journal of Business, 12 (2007), 117-136.

Armstrong, J.S. and F. Collopy (1996), “Competitor Orientation: Effects of Objectives and Information on Managerial Decisions and Profitability,” Journal of Marketing Research, 33 (1996), 188-199.

Shubik, M. (1971), “The Dollar Auction game: A paradox in noncooperative behavior and escalation,” Journal of Conflict Resolution, 15, 109-111.

Getting emotional about brands – the real New Coke story

I was recently asked a question by an astute member of the audience: “what about New Coke?”

The implication being, that if consumers rarely feel deep emotional bonds towards the brands they buy then why the rejection of New Coke?

It was a very good question.

On the spot I answered that one of the few times that people do get emotional about brands is when you take something away from them, particularly when you do it in a very public noticeable way. Psychologists talk about the endowment effect, how people tend to place a much higher value on something they already have than something they are offered. Also people are much more adverse to loss than they are attracted to gains.

I think this is a pretty reasonable answer but it turns out that there is more to the New Coke story…., read on…

The story of New Coke is now a marketing legend. Covered by many a textbook and still regularly referred to by marketing consultants. The story goes like this….. worried by ‘The Pepsi Challange’ taste test Coca-Cola executives decided to change Coke to make it sweeter and more likely to win in blind taste tests with consumers. So a new formula was devised and in a bold move Coca-Cola’s time-honoured taste was changed and launched with national advertising and publicity. Cans carried the message “New”. But the American public hated the idea of a classic like Coke being tampered with. They complained and refused to buy. Coca-Cola realised their mistake and brought back “Classic Coke” which quickly became the norm and “New Coke” was phased out of production. It was considered a huge marketing mistake, although some conspiracy theorists proposed that it was all part of a master plan by Coca-Cola corporation to get publicity.

So goes the legend but the real story doesn’t fit marketing theory quite so neat and tidily.

Coke’s share price actually went up when it introduced the flavour change. Earlier Coke bottlers gave the CEO of Coke a standing ovation when he announced their plans. Sales did not dive when the new formula was introduced, they rose. So the degree of consumer backlash has been exaggerated, and wasn’t immediate. Resistance came from the South, around Atlanta Coca-Cola Corporation’s home. Here were people who felt they owned part of Coke, who felt current management didn’t have the right to change things. And here consumer backlash was highly visible to Coke executives many of whom lived in the South.

The legend that developed later served Coke well. The idea that Americans loved Coke so dearly that they demanded its return and were reminded of their love.  If I worked for Coke it would be a myth I’d be happy with.

Fast advertising is good advertising

What I mean is that the faster a consumer can understand your advertising the better. They should not only be willing to watch it (over and over) but it should also be EASY for them to realise that it’s you advertising. If they only see a portion of it, they should still understand who is selling and what they are selling.

Take care in straying from this advice.

Can you use facebook to stimulate your fans to talk about you?

Since the Advertising Age covered the Ehrenberg-Bass Institute’s analysis of facebook’s ‘talking about’ metric there has been a flurry of internet coverage.

The findings got reduced to a sound bite of “only 1% of facebook fans engage with brands”. Which could easily be misinterpreted. Dr Karen Nelson-Field’s result is actually that around 0.4% (ie less than one percent) of the fans of a brand actually interact with it on facebook in a typical wek.

The interaction is what facebook report as “Talking about’, and includes activity such as to like, comment on or share a Brand Page post (or other content on a page, like photos, videos or albums), post to a Page’s Wall, answer a posted question, liking or sharing a check-in deal, RSVP to an event, mention a Page in a post, phototag a Brand Page…all the activity that facebook measure.

Now 0.4% in a week doesn’t sound so bad. It sounds like it might cumulate to near 25% in a year, but this would be a heroic assumption. In these sorts of social phenomenon we usually see highly skewed distributions. There will be a small percent of fans who do most of the talking every week. So this probably cumulates to something much less than 10% in a year. Karen is investigating.

Even facebook’s own fans don’t talk much about facebook (on facebook)

One of the questions asked of Dr Karen Nelson-Field’s analysis of facebook fans engagement with their brands on facebook is whether the result is simply due to slack social marketing by the brands in question.

Given that Karen analysed the 200 brands with the most facebook fans it seems a bit of a stretch to say that these brands “don’t understand facebook”.

Some have speculated that brands that understand passionate loyalty probably do much better.  But Karen’s analysis included brands such as Old Spice, Harley-Davidson, Ferrari, and Tiffany & co.

Finally, Karen’s analysis included facebook’s own facebook fans.  In a typical week only 0.28% ‘talk about’ facebook on facebook.  Maybe facebook itself doesn’t care much about fan engagement, after all they are clever marketers.

The flawed Stengel Study of Business Growth

Here I describe the ‘Stengel Study of Business Growth’ using quotes from “Grow: How Ideals Power Growth and Profit at the World’s Greatest Companies” by Jim Stengel, published by Crown Business 2011. Along the way I point out the fatal flaws in the research design.

The ‘Stengel Study of Business Growth’ started in 2007 when Procter & Gamble’s CEO A.G. Lafley endorsed Jim’s idea to “commission a study to identify and learn from businesses that were growing even faster than we were, in whatever industry” (p. 24).

Initially the P&G team studied “the fastest growing brands over the previous five years” (p.24) identified in collaboration with market research agency Millward Brown Optimor using their BrandZ database. The team “assembled five-year financial trends on twenty-five businesses that had grown faster than P&G over that period. The teams then dug behind the numbers with additional research, including interviewing business executives, agency leaders, brand experts, and academics at Harvard, Duke and Columbia”. (p.25)

“We went in looking for superior financial growth, and only after that for whatever the top-ranked businesses were doing differently from the competition” (p26). Professor Philip Rosenzweig explains this classic sampling mistake as being like trying to learn about blood pressure by only looking at a small group of patients who all have high blood pressure.

Another very important mistake, that we have learnt about as various strategy researchers have made it over the years, is to look for causes of success by interviewing managers and ‘experts’ for their opinions on firms that have been doing well. Known as “the Halo effect” people tend to say that firms they know are performing well possess all sorts of desirable characteristics in terms of culture, leadership, values and more. No one describes a known winner as having “unfocused strategy”, or “weak leadership”, or “lack of customer focus”, or “lack of ideals”, or whatever the researchers choose to decide to ask opinions about. As Philip Rosensweig shows clearly in his book “many things we commonly claim drive business performance are simply attributions based on past performance”.

“Successful companies will almost always be described in terms of clear strategy, good organization strong corporate culture, and customer focus” (p.87). Rosenzweig dramatically shows how when successful companies falter experts abruptly change their assessment. Suddenly the previously described “strong culture” is now described as “rigid”, their previously declared “promising new initiatives” are now described as “straying”, their “careful planning” now in hindsight turns out to be “slow bureaucracy” and so on. In reality large businesses change very slowly, but opinions about them change quickly and are largely based on current financial performance (which is itself is largely due to environmental and competitor effects).

The Halo Effect is particularly strong for subjective, nebulous concepts such as ‘values’ and ‘ideals’. The ‘Stengel Study’ made no attempt to supplement their judgements with ‘hard’ objective measures.

In the Stengel study they ‘discovered’ that their chosen high-growth firms were ‘ideal driven’. The central finding therefore was that “businesses driven by a higher ideal, a higher purpose, outperform their competition by a wide margin”. Yet there is no mention of any systematic investigation of competitors, perhaps many of these lesser performers were also ‘ideal driven’? What we can expect is that because of the Halo Effect less successful performers would be less likely to have been described by interviewees as having a clear ideals well activated throughout the business – irrespective of reality.

Subjective concepts such as ‘ideals’ almost certainly introduce confirmation bias on the part of researchers – when there are no objective measures it’s near impossible for a researcher to stop themselves seeing what they want to see. The “unexpected discovery” of the causal effect of ideals, says Jim Stengel “corroborated what I had implicitly believed and acted upon throughout my career”. Hmm, of course it did.

With this ‘ideals’ hypothesis now firmly in place the full ‘Stengel Study’ was then done after Jim Stengel left P&G by selecting 50 brands based on their excellent recent financial performance over 10 years. As a whole this group (refered to as “The Stengel 50”) “grew three times faster over the 2000s than their competitors…individually some of the fastest-growing of the Stengal 50, such as Apple and Google, grew as much as ten times faster than their competition from 2001 to 2011.”

Promotional material for Stengel’s book says that “over the 2000s an investment in these companies—“The Stengel 50”—would have been 400 percent more profitable than an investment in the S&P 500”. The implication is that this proves Stengel’s ‘ideals’ thesis – but Stengel picked these companies for their financial growth!

If they have been picked purely based on some, ideally ‘hard’ (or intersubjectively certifiable), measure of being ‘ideals driven’ then correlations with financial performance might mean something. Especially if this were future, not past, performance. But as these companies were picked for their financial performance then their stock price performance over the same period shows nothing.

A team of four second-year MBA students being taught by Jim Stengel and Professor Sanjay Sood made the Stengal Study the subject of their required applied management research thesis; “this team crawled all over the Stengal 50 to test the role of ideals” conducting interviews with executives, academics and consultants”. No one should be surprised that they found what their instructors believed. They and the ‘Stengal Study’ both passed with flying colors reports Jim Stengel (page 34).

There is one addition to the Stengel Study which is different from previous similar (flawed) studies of business success. Stengel arranged his leading brands into “five fields of fundamental human values that improve people’s lives” by (1) eliciting joy, (2) enabling connection, (3) inspiring exploration, (4) evoking pride, or (5) impacting society (sic). Millward Brown then used implicit and explicit association measures and found that the Stengel 50 brands are perceived as more associated with their selected ideals than competitors.

Again this is a staggering piece of circular logic. First analyse a select group of brands for what particular ideals they represent, then take these ideals into market research and viola these brands turn out to be more associated with these particular ideals. This is not a test that these ideals drive performance, it is simply a test of the researchers’ judgement of brand image. It merely shows that the researchers live in the same culture as the market research respondents. Jim Stengel thinks Backberry ‘enables connection’ and so does the market, Jim Stengel thinks Mercedes Benz ‘evokes pride’ and so do normal people.

Now one might reasonably argue that there is advantage in FedEx and Blackberry being more associated with a category benefit such as “enables connection” than their competitors. However, leading brands always show higher associations because they have more users, who use them more often. Behaviour has a powerful effect on attitudes and memory, for evidence see BIRD, M. & EHRENBERG, A. 1972 “Consumer Attitudes and Brand Usage – Some Confirmations”. Journal of the Market Research Society, 14, 57. RIQUIER, C. & SHARP, B. 1997 “Image Measurement and the Problem of Usage Bias” in proceedings of 26th European Marketing Academy Conference, Warwick Business School, U.K., 1067-1083. ROMANIUK, J. & SHARP, B. 2000 “Using Known Patterns in Image Data to Determine Brand Positioning”, International Journal of Market Research, 42, 219-230.

There is no mention of controlling for this effect.

Conclusion

In summary, the ‘Stengel Study’ makes the same or similar mistakes as much earlier flawed studies that claimed to uncover the secret of sustained financial success. Jim Stengel, and none of his team appear to have read Philip Rosenzweig’s “The Halo Effect: … and the Eight Other Business Delusions That Deceive Managers” which turns out to be a great pity. As he writes “if the data aren’t of good quality, it doesn’t matter how much we have gathered or how sophisticated our research methods appear to be”. The Stengel Study is yet another study that is deeply flawed, it tries to look like science, but turns out to be merely a story, one that will appeal to many but tells us nothing reliable (or new) about the world.

A final note: Based on the track record of previous such studies I expect the financial performance of these ‘ideals driven’ companies to fall back in the near future. Some such as Blackberry, HP have already suffered very dramatic reversals of fortune.

Review of Jim Stengel’s disappointing book “Grow”

Research reveals the hidden secret to business success? No, sadly this is pseudoscience – that will only convince the most gullible of minds.

Jim Stengel seems a nice guy, he wants us to be passionate about our business and to feel that there is a greater purpose than simply making money.  Few would disagree.  But he also claims to have discovered the secret to sustained super profits – based on a flawed study dressed up as science.

Stengel is a marketing consultant, a famous one because he was formerly Chief Marketing Officer of Procter & Gamble 2001-08 until he surprisingly ‘retired’ to consult (and write this book). During the decade that Jim mostly presided over marketing at P&G the company was pretty successful, at least in comparison to the 3 year period immediately before he became CMO of unsuccessful restructuring and CEO turnover. However the success of the 2000s has been exaggerated; the reality is that during Jim’s decade P&G’s stock price doubled, though that is a misleading overstatement due to the brief dramatic dip in 2000 (the reasons why are discussed here). Without that dip the year before Jim took over as CMO the stock price only improved 20% over the full decade. That’s less impressive than the previous decade (90s) when stock price had increased 5-fold (or 3 fold when the brief dip of 2000 is considered), similar gains were also made in the prior decade (80s). So P&G’s performance during Jim’s tenure should perhaps more accurately described as a mild turnaround, or partial restoration. This chart shows the full history of the stock price.

In all fairness though, Jim Stengel doesn’t ask us to believe his amazing discovery just because he was (like millions of others) a successful practitioner, his claims are based on what he calls an unprecedented 10-year empirical study of highly successful firms and the brands they own. But his study does have precedent, it joins a growing list of books that claim to have discovered a few simple rules for business that near guarantee profit performance that will beat all rivals. Each of these books are based on severely flawed research that ‘proves’ just what the author wanted to say in the first place (which is the opposite of a surprising discovery). “In Search of Excellence” was one of the first of these books, which was largely discredited when the excellent companies went on to make poor financial returns in the years after the book came out.

Professor Phil Rosenzweig exposes these flaws in his 2007 book “The Halo Effect: … and the Eight Other Business Delusions That Deceive Managers“.

I describe and critique “The Stengel Study”, which is the basis of Stengel’s book, here. A quick summary is that to detect factors that might cause financial success then Stengel should at least compared very carefully matched samples of both successful and unsuccessful firms, and developed hard objective measures of strategy – not relied almost entirely on interviews with experts. Also, to avoid confirmation bias, the researchers who described the firms and their strategies should not have been aware of which were the successful and unsuccessful ones. And finally, any resulting theories should be tested against the future performance of the firms. Otherwise what looks like science turns out to be simply a story.

Tellingly the ‘research’ takes up small portion of Stengel’s book, the rest is a story: anecdote and assertion. Jim tells us what to do, but experienced marketers looking for strategic advice won’t find much new or particularly helpful. It’s pretty much the standard sort of consultant fare such as “deliver a near-ideal customer experience”.

It’s well meaning though, Stengel wants us to all be passionate about our business and to feel that there is a greater purpose than simply making money (even if finding out how to make money was the motivation of his ‘research’).  This is a nice sentiment, however, the success of brands (and the large corporations behind them) is far more complex than Stengel’s book and its predecessors claim.

Zero Moment of Truth – Hype, Nonsense, and PseudoScience

Shock, how amazing – new ‘research’ from Google shows that advertisers should be spending far more of their advertising dollars online with Google.

In a report that insults the intelligence of the marketing community Google tell us that consumers are doing more on-line product research than they did in the past (when they weren’t online). Unless you have been in a coma for the past decade you didn’t need Google to tell you that. But some quantitative insight would be useful – how much are consumers using on-line sources of product information, and what sources? Unfortunately Google’s research and data presentation is so shoddy we can gleam nothing reliable.

They did an online survey (ie biased towards heavier online users) of various subsamples (eg 500 people who had bought an automobile, another 250 who had applied for a new credit card in the past 6 months, and so on).

All the data concerns claimed (recalled) behaviour and the sub-sample results are then often averaged into meaningless metrics.

The report highlights stupid meaningless quotes like “70% of Americans say they look at product reviews before making a purchase”. Is this every purchase ? Or 70% have a least once in their lives looked at a product review ? Actually this quote is sourced from an equally sloppy 2009 study but not by Google – why they chose it when they have their own “new research” puzzles me.

I could spend all day pointing out how meaningless the metrics are in the Google report, but I don’t think there is any need. Only extremely gulliable marketers would rely on such a sloppy blatent piece of self-promotion disguised as research.

In May, Professor Jerry Wind and I are hosting a conference at Wharton. If Google had some meaningful, reliable data on the value of online touchpoints we would be delighted to invite them to present.