Marketing: alchemy, magic or science?

Marketing is a profession often viewed with suspicion. Some see it as no better than medieval alchemy: a profession which attempts to turn worthless base metals into glittering items of value.
There are others who see it as only a creative calling in which a spark of magic is required to bring brilliant unique ideas to life.
Most do not recognise that it is also a science. Data, logical conclusions and insights are required before we even attempt to turn base materials into something of greater value or before the creative forces can be unleashed.
The study of the science of marketing has been going on for a long time. Professor Andrew Ehrenberg (and many illustrious others) had identified law-like patterns in brand metrics many decades ago. The largest revelation from all of this has been in realising that what I had previously taken to be unique patterns in my brand’s data were not unique and, often, that the implications are not what I had predicted. In short, I was wrong. Science was right.
To those who read my website regularly, I should warn you that this article is much longer than normal as it goes much deeper into insights and their implications. I hope you enjoy it though. (If you want to read more, then there are many references at the end of the article. If you want to read less, search here for edited highlights)

As economist John Maynard Kenyes highlighted early in the last century, unlearning what we already thought we knew is the really hard bit!
Three solutions for better FMCG marketing
Target penetration growth, not loyalty growth
Strive to be distinctive, not different
Be memorable, not persuasive
These three statements are so easy to write, but so at odds with most perceptions of how to grow FMCG brands.
Hence the title of the next section:
The dangers of common sense
Before diving into the theory, the data and the logic, let’s start with a simple formula for a brand’s Total Revenue:
Total revenue
= Number of Buyers * Frequency of Purchase * Spend per purchase occasion
Seemingly nothing radical in that: if the goal of a business is to grow revenue, then logically any of the three factors can be influenced.
However this perfectly logical assumption is a trap. You may believe other factors are changing, and that you have strategies to drive loyalty and to encourage frequency. However, these are largely a fallacy.
Practically only numbers of buyers alter significantly between brands in a category. Put very simply, big brands are big because lots of people buy them.
The rest of this essay reviews the data and empirical proofs that underpin this and, importantly, reviews the implications for marketers in moving beyond this empirical reality.
Penetration – big is beautiful
Loyalty, share of requirements, or purchase frequency do not move significantly.
“When packaged-goods brands grow, both penetration and loyalty increase, but penetration grows two or three times more than loyalty” (Romaniuk, 2013).
Big brands have more buyers. They do also have a slight advantage in that the largest brands not only have higher penetration, but also have slightly higher loyalty levels. This “double jeopardy” phenomenon was formalised for FMCG brands in 1990 by Ehrenberg, Goodhardt, and Barwise. IThe double jeopardy reality had already been “revisited, again” by Ehrenberg and Goodhardt in 2002.
This is well proven data but it is amazing how many marketers (this author included a while ago) are not aware of this. And it is robust data, confirmed in 2012 by Uncles, Kennedy et al. using data spanning 25 years, across 50+ categories and 60 data sets. It is so important that, whilst I will not use data to confirm academic knowledge throughout this article, I will illustrate it here. This example (using data in the public domain) is a shampoo category in the UK. (Source: TNS)
| Brands |
Market share (%) |
Annual penetration % |
Purchase frequency (avg.) |
| Average |
|
|
1.9 |
| Head & Shoulders |
11 |
13 |
2.3 |
| Pantene |
9 |
11 |
2.3 |
| Herbal Essences |
5 |
8 |
1.8 |
| L’Oreal Elvive |
5 |
8 |
1.9 |
| Dove |
5 |
9 |
1.6 |
| Sunsik |
5 |
8 |
1.7 |
| Vosene |
2 |
3 |
1.7 |
We can see that the leading brand has a share 5 times larger than the smallest. However, its purchase frequency is only 35% higher. The penetration (i.e. Head & Shoulders’ much higher number of buyers) is what is driving performance. It is double jeopardy in action.
Knowing this pattern exists in consumer markets makes it easy to spot it when you look. (Find some panel data now and try. Contact me and let me know how it looks!)
For me, this was a significant learning. I have previously created superb, seemingly compelling strategies to grow loyalty. However, academia shows that any minor gains in loyalty were more likely a statistical fluctuation, or a feature of successfully growing penetration instead.
Examining this further I discovered that a study of 880 entries in the Institute of Practitioners in Advertising annual effectiveness awards confirmed the finding. “Focusing on penetration, not loyalty” was the key to success (even in strategies that had actually set out to grow a brand through increasing loyalty!).
In a related article one of those authors titled the piece “the dangers of common sense”, so apparently logical is the Revenue equation at the start of the previous section. Binet was “shocked to discover that what most marketing people do and what actually works are really quite different.”
To check this in my brand world, I looked back on past data for brands I have worked with. My own data conforms to the Double Jeopardy pattern. My conclusions often assumed something else. In a related point, the agencies that we as a client paid money to, were also often unaware of the Double Jeopardy phenomenon. I have uncovered a wonderful example from 2004. A well-known global research agency highlighted that our brand was 2nd in the market with smaller penetration than the leader. A further slide proclaimed the seemingly interesting “almost identical frequency of purchase between brands.” However, if that agency or I had been aware of Double Jeopardy this headline would have been unsurprising! It is the nature of FMCG markets as shown by Ehrenberg and others decades ago!
Implications
If brands are big primarily because of penetration (i.e. number of buyers) then it is important to attract new users.
Perhaps no surprise, but in the next section I’ve reviewed more of what it means to grow through acquiring new users, specifically that they are going to be light buyers (not average, medium …or loyal).
Growth comes from light buyers
To understand buyer behaviour, I’ve been using data such as that below for many years. (This is actual data, but anonymised by the removal of category/country titles!)
|
% Buyers |
| 1 time |
43 % |
| 2 times |
17 |
| 3 times |
9 |
| 4 times |
7 |
| 5 times |
5 |
| 6 times |
5 |
| 7 times |
4 |
| 8 times |
2 |
| 9 times |
1 |
| 10 times |
1 |
| >=11 times |
7 |
It shows, apparently, that we have a massive opportunity to encourage people to buy the category more often. 60% of them buy it only once or twice a year.
In fact, what I only realise – and realise very clearly – from reviewing the academic research, is that frequency/loyalty charts typically have this shape, with most buyers being very occasional! This shape – the “negative binomial distribution” effect on brands (captured as “NBD Dirichlet” by Goodhardt, Ehrenberg and Chatfield in 1984) typically shows the same pattern where most buyers are “light”, and a brand relies for its sales on the occasional purchases by lots of such buyers.
The reality of what happens is that the users who are added are “light” users. In any time period, a new user who buys once is going to be light (vs, for example, the rare 7% who buy monthly in the annual data above).
This is a key learning! It is easy for marketers to make false assumptions if they only look at their current data (e.g. the chart above). I have used data such as that above to focus on building loyalty and repeat rates. (The same global research agency presentation I referred to above went on to make a strategic recommendation to invest in frequency-building activities). However, realising that all categories look like this means we need to understand brand and category performance in a different way.
This does not necessarily mean that there is no such thing as a customer who appears loyal. Reichheld & Teal argue for such customers’ importance in their 1996 book, The Loyalty Effect. However, I have come to realize that the promotion of loyalty as a key driving force is not borne out by FMCG data. In services (admittedly not FMCG) a 1990 HBR article by Reichheld and Sasser suggested that “Companies can boost profits by almost 100% by retaining just 5% more of their customers.” When a trusted source such as the Harvard Business Review makes such a bold claim, the world pays attention. (For an alternative view it is interesting to read Professor Sharp’s dissection of the “loyalty myth” of the HBR article here)
However, as we’ve now seen, the FMCG data highlights that even if loyalty is important, penetration is much more significant.
Implications
Double Jeopardy highlights that Penetration matters, and NBD Dirichlet shows that brands sell mostly to large numbers of light buyers.
Those two thoughts require an adjustment for many marketers.
Loyalty programs, CRM, tight targeting on social media: all such activities make a lot less sense and need to be seen in a different light (with recommendations to follow in a future article. Contact me if you’d like a preview of my advice on that).
The challenge then is how to attract trial from the all-important light buyers.
Salience matters
To answer that challenge I have looked back on Ehrenhberg’s papers from the 1980s and beyond and also Professor Sharp’s research. If growth depends on light buyers (who by their very definition do not purchase you very often) a brand owner needs to do everything to make the brand stand out. This, so far, does not present anything radical. Marketing strategies the world over will talk about raising awareness!
However, to grow a brand, Sharp talks about a brand needing both mental and physical availability.
In a buying situation the brand must be salient. At its simplest that means being visible on the shelf in the supermarket from which consumers buy this category, or being in the stores where such goods are bought. (This is distribution and merchandising, creating “physical availability”).
In that situation the potential consumer needs to recall the brand when faced with it on shelf. This is “mental availability”. The implication from this is that brands need to do everything possible to maximise linkages between communication outside the store (e.g. advertising) and communication inside the store (which would always be packaging, and could also be point of sale materials). A brand needs constantly to build salience.
Implications
All of the above has a significant media planning implication. I have tackled media planning in multiple markets. There is a constant tension between having strong “bursts” (heavy investment levels in a short period) vs weaker “maintenance” / “sustaining” investment. The argument for strong bursts is that it “cuts through” and gets noticed.
However, the danger of a “burst” strategy is that it could be 2-3 months since a consumer was last exposed to my brand messages. A lower level, but more continuous media strategy makes sense. “Mental Availability” will be stronger – i.e. the chance of being thought of in the buying situation. (This, of course, does not remove the need to create strong creative work that gets noticed, but I will come back to that below).
There are other implications about measuring the impact of advertising, especially vs competitors. A move to seemingly weaker advertising needs a more refined research methodology to capture its impact on long term memories of target consumers. We need to be mindful of research effects from brand scale. Romaniuk highlights that the classic Top-of-mind awareness indicator “is highly biased to brand users (and particularly heavy users) and will favor larger share brands. For the same reason, it is a poor metric choice for tracking a small (potentially growing) brand.”
The differentiation myth
To build salience in the form of “mental availability” large FMCG brands typically communicate through advertising. Hours (and hours!) are spent refining advertising messages and campaigns. Many marketers are brought up on the idea of Unique Selling Points. Reeves “created” the concept in the 1960s so it has had a long time to be established. Trout and Rivkin reminded everyone to “Differentiate or Die” in 2000.
However, whilst it seems conceptually appealing, a review of (again!) the data shows that all is not as we might expect.
What the data highlight is that most consumers buy from a repertoire of brands. In Sharp’s words they are “polygamously loyal” to a range of brands not an individual one. Brands gain most of their purchases from people who also buy other brands in the category. This is also consistent with the NBD Dirichlet fact that most buyers of a brand are “light”.
This means of course that advertising doesn’t convert or persuade! (What it does instead is remind, and build salient memories that connect the product to the brand and the category. It can then be recalled in buying situations).
In their 2012 article, It’s a Dirichlet World, Sharp et al confirm that “loyalties are distributed across consumers with little differentiation between brands, such that each brand in effect sells to all category buyers rather than a particular segment”. Other studies (e.g. Uncles et al 2012) explain this further. “From a consumer viewpoint, many people see competing brands as substitutable”. (The data in these studies is remarkable. It really is worth a read!)
For more confirmation, in an article by research agency, Millward Brown, reviewing their own BrandZ database of 6000 brand and category buyers, they confirm that “on average, the proportion of people willing to endorse any brand as “different from other brands of (a specific category) is low… Among those that consider a brand acceptable, an average of 18 percent agreed that it was different from others”
Put another way, people happily buy a brand even though in 82% of cases they do not think it is different from others in the category.
Implications
This means that advertising does not “persuade” people of the merits of one brand over the other. Consumers remain somewhat indifferent.
What becomes important is to ensure consumers think of the brand as a suitable substitute in the category. This means truly understanding how the consumer thinks of the category and how your brand can fit into that. It is very different from trying to persuade them of a unique feature of your brand. It is actually about showing them that you are a good example of your category: a very different challenge from the USP way of thinking.
Time for magic
The title of this article raised the idea that marketing may contain a magical creative spark.
In a “Dirichlet World” where growth depends on new buyers and where loyalty is not a likely reality, creativity matters.
The largest brands in a category score highest on generic attribute measures for that category (e.g. refreshing, washes whiter, tastes good). Consumers see them as a good example of their category.
In the section above we reviewed that consumers see competing brands as largely interchangeable.
What matters is to be distinctive. This is the magic of marketing: to turn the generic into the memorable. Unilever’s washing brand, Persil/Omo, embraces the idea that “dirt is good”. The benefit is still that the product washes whiter, but the story is distinctive. The benefit is not unique. The ingredients are not unique (or are unlikely to offer any sustainable advantage). However the story becomes memorable for the way it is dramatised.
This, for me, is powerful. In a previous role I was presented with a marketing concept from our global team to launch a new product. Striving to be different it avoided the core category attributes that the market leaders already “owned” and defined a range of unique selling propositions to stand out from the crowd.
However, what we needed was not a new set of rational benefits (which were of low interest to the consumer as the top ones were already “owned”). What we needed was a creative way of branding our version of the category core benefits. The sad fact is that we already had it – a creative world in which our brand already lived, distinctively! Our creative challenge should have been to say that we now also had a product from that same distinctive world that did everything the category leaders said their products did. We needed to brand the generic, not try to attach our brand to meaningless USPs.
It’s also important to watch out for the realities of researching such brand attributes. Scriven and Goodhardt reviewing “The Ehrenberg Legacy” in the Journal of Advertising Research in 2012 highlighted that “Users of a brand tend to say positive things about it (e.g., “It tastes nice”; “I like it”), whereas non-users are less likely to comment”. Usage drives attitude! “Whether people say that a brand has a given attribute largely depends on whether or not they buy the brand” (Romaniuk, Sharp, 2000)
So, brands with the largest penetration (user base) score highest on category attributes.
Consumers are not really “persuaded” of a brand’s unique attributes. Ehrenberg’s view of advertising was that it works by publicising a brand not by persuasion. If consumers were “persuaded” of rational brand benefits they would a) likely be loyal, b) stay consistent in their beliefs [but only about 50% express the same view in subsequent research interviews (e.g. Romaniuk et al. 2012)]. In Ehrenberg’s words in a 1997 article, the role of advertising is more as a “nudge” (a weak, reminding force) than as a strong persuader.
More bluntly an Admap 2009 article confirms that “Ads must be memorable, not persuasive, to influence choice.” This makes sense. If FMCG brands grow mostly through attracting new buyers or reminding light ones, then a brand’s communication needs to be remembered in store to “nudge” purchase behavior. Salience matters (and continuous media strategy reaching large numbers of people is more important than shorter bursts).
Implications
From an advertising and creativity perspective this all shows that we should not be afraid of category generics.
It is important to understand better what data actually proves. It is remarkably liberating to realise that the largest brand will always appear to “own” the category generics. Causality is not clear in a snapshot of the data. Respondents agree with the statements because they have experience of (i.e. buy) the brand. They do not necessarily buy the brand because they see the attributes as unique.
The largest brand appears to “own” attributes but that is simply the way consumers respond (saying positive things more about brands they buy). More buyers means high attribute agreement.
Seeking a “USP” can be a dangerous trap. Consumers buy a category for its fundamental attributes. The task for a brand is to turn those attributes into something memorable for that brand. The magic in marketing is in branding the generic.
Dumb data and alchemy

At the start of this article I posed the observation that marketing is perceived as little better than alchemy. However, I don’t believe it is possible to persuade someone to keep spending money buying a “bad” product. It is not possible to turn lead into gold.
However, I do think that many marketers (this one included) have been guilty for many years of trying a similar trick, of trying to convert flawed knowledge of base data patterns into valuable strategic recommendations.
If we look at our brand’s data and conclude that we will grow best by increasing loyalty (rather than penetration), we fail to understand the science.
However, if we recognise that loyalty is a statistical artefact related to scale and penetration, then our media strategies need to act as constant reminders to the uninterested potential consumer. Strategies that “burst” to better “convert” competitor users are likely to fail. We need constant efforts to stay salient, to nudge and to top up mental availability.
If advertising does not really “persuade” because no sizeable FMCG brands consist of entirely 100% loyal converts, then its role is to be memorable. As described above we need to be distinctive and embrace the generic motivations of the category.
The science of marketing
To return to the title of this essay, marketing is not alchemy – though many alchemists still practice here.
It can be magical (and it can be a wonderful thing when brilliant creatives are given a tight brief and turn the generic into the memorable).
What I hope I have shown through this article is that it is certainly much more of a science than many (including this writer) have ever realised. Key fundamentals were the Double Jeopardy law and the NBD Dirichlet model – which you should read if you haven’t already!
It’s important to humbly recognise that, whatever efforts brilliant marketers make, actually consumers don’t see brands in a category as very different from each other.
We need to recognise that research data appears to paint a certain picture (e.g. of the market leader “owning” a set of category attributes), but is often misleading.
All of this understanding drives a recognition of how advertising and media planning should work (through increasing salience for lots of non/light buyers) – which challenges a huge range of other theories (e.g. of advertising as persuasion).
If you would like to know more and think I could help you or your business, do get in contact.
In the meantime, happy discovering…
Sources
Some sources have been directly cited above. Others were a means to an end: they informed my thinking and pointed me off in new directions to explore even if they are not specifically referenced in the essay. I am particularly indebted to Professor Byron Sharp and to Martin Weigel for their excellent guides to these topics.
- N Barnard and Prof A S C Ehrenberg, Advertising: Strongly Persuasive or Nudging? Journal of Advertising Research
Vol. 37, No. 1, January/February 1997
- Binet L,
The dangers of common sense, Market Leader Quarter 3, June 2009
- Binet, L and Field, P; Empirical Generalizations about Advertising Campaign Success. Journal of Advertising Research, June 2009
- H Bloom, R Kennedy, A Ehrenberg and N Barnard, Brand advertising as creative publicity, Journal of Advertising Research
Vol. 42, No. 4, July/August 2002
- M Bongers and J Hofmeyr, Why Modeling Averages Is not Good Enough: A Critique of the Law of Double Jeopardy,
Journal of Advertising Research, Vol. 50, No. 3, 2010
- P Buckley,
Exploiting the implicit: I believe it’s what brands don’t say that matters,
WPP Atticus Awards 2012
- Bullmore: Advertising and its Audience. Advertising Association lecture 1998
- Ehrenberg, A; Goodhardt, G, Double Jeopardy Revisited, Again, March 2002, Marketing Research; Spring2002, Vol. 14 Issue 1, p40
- Ehrenberg, A S C, G G Goodhardt, and T P Barwise (1990), Double Jeopardy revisited Journal of Marketing, 54 (3), 82-91
- T Ewing, B Pankauskas, R Brown and J Chen, Research in a world without questions, ESOMAR
Congress, Atlanta, September 2012
- Goodhardt, G J, A S C Ehrenberg, and C Chatfield, “The Dirichlet: A comprehensive model of buying behaviour,” Journal of the Royal Statistical Society, 1984, 147 (5), 621-55.
- R Heath and P Feldwick, Fifty years using the wrong model of advertising, International Journal of Market Research Vol. 50, No. 1, 2008
- N Hollis, It Is Not a Choice: Brands Should Seek Differentiation and Distinctiveness,
Millward Brown Points of View, April 2011
- R Kennedy, S Long and A Ehrenberg, Competitive Brands’ User–Profiles Hardly Differ, Market Research Society Annual Conference, 2000
- Kluger, Why You’re Pretty Much Unconscious All the Time, Time Magazine, June 2015
- J Leighton Ads must be memorable, not persuasive, to influence choice,
Admap
February 2009, Issue 502
- Sharp, B. How Brands Grow, 2010
- B Sharp, The only way to grow your brand, Admap
April 2003, Issue 438
- B Sharp, Point of View: The power of branding, Admap
June 2014
- B Sharp and J Romaniuk, Using Known Patterns in Image Data to Determine Brand Positioning, International Journal of Market Research, Vol. 42, No. 2, 2000
- B Sharp, M Wright, J Dawes, C Driesener, L Meyer-Waarden, L Stocci, P Stern, It’s a Dirichlet World: Modeling Individuals’ Loyalties Reveals How Brands Compete, Grow, and Decline.
Journal of Advertising Research. Vol. 52, No. 2, 2012
- G Precourt, EffectiveBrands: Best practices move beyond data to build a brand vision, ANA Brand Masters, February 2014
- Reeves, R (1961). Reality in Advertising. Macgibbon and Kee. pp. 46–48
- Reichheld, W. E. Sasser, Jr Zero Defections: Quality Comes to Services,.HBR September–October 1990
- Reichheld F & Teal, T The Loyalty Effect, 1996
- J Romaniuk, How Healthy is Your Brand-Health Tracker? A Five-Point Checklist to Build Returns on a Critical Research Investment Journal of Advertising Research, Vol. 53, No. 1, 2013
- J Romaniuk, De-Clutter Your Mind: Seven Worries You Can Drop from Your Research Agenda,
Journal of Advertising Research, Vol. 53, No. 2, 2013
- Romanuik, J, Byron S, and A Ehrenberg (2007), Evidence concerning the importance of perceived brand differentiation, Australasian Marketing Journal, Vol.15 (2), pages 42-54.
- J Romaniuk and E Nicholls, Evaluating advertising effects on brand perceptions: incorporating prior knowledge,
International Journal of Market Research, Vol. 48, No. 2, 2006
- J Scriven and G Goodhardt,
The Ehrenberg Legacy: Lessons in Buying Behavior, Television, Brand Perception, Advertising, and Pricing, Journal of Advertising Research, Vol. 52, No. 2, 2012
- Trout, J and Rivkin S, Differentiate or Die: Survival in Our Era of Killer Competition, 2000
- M Uncles, R Kennedy, MagMda Nenycz-Thiel, J Singh and S Kwok, In 25 Years, Across 50 Categories, User Profiles for Directly Competing Brands Seldom Differ: Affirming Andrew Ehrenberg’s Principles, Journal of Advertising Research Vol. 52, No. 2, 2012
- M Weigel The liberation of magic: How marketing science opens up creative opportunity, June 2013