The first video is an interview with the fascinating Dr Carl Marci – the founder of neuroscience company Innerscope.  I wrote about him in a previous post but nothing beats listening to the man himself!

I mapped the companies on a scatterplot using Many Eyes (please be patient as this page can take time to load).  There were a few outliers like Starbucks so I took the log of both the mentions and followers to make the map legible.

The dots on the horizontal axis represent brands that do not officially participate in the conversations on Twitter (they have no followers).  The ones on the right hand side of the graph (Apple, Microsoft and AIG for example) are mentioned frequently in conversations so they might want to consider taking a more active role.  

For the brands that do have followers there seems to be a correlation between followers and mentions.  That makes sense intuitively.  Brands in the top right quadrant are very active and the ones on the bottom left are not.

If you draw an imaginary trend line through the scattercloud you would find brands on that line to have a number of followers that is in proportion to the volume of mentions they get.  Brands above the line however get followed more than they are talked about.  It’s the other way around for brands below the line.  If you mouse over the dots on the graph you see which brands fall where.

I think we need more data to really interpret the graph.  Being followed more than being talked about could be good or bad depending on the engagement levels of the followers.  Being talked about more than being followed could be good or bad depending on the sentiment of the conversations about the brand.  Even if the interpretation of this graph isn’t 100% clear, I thought I’d post it anyway as I think it is interesting to compare brands visually.  Would love to hear your thoughts.

Yesterday was Memorial Day in the US  - the official start of the summer BBQ season and everyone seemed to have their grill out in the city.  I had a few friends over for a BBQ and decided to prepare a smoked brisket.  I followed the instructions of this enthusiastic grill master and slowly cooked an 8 pound brisket for 8 hrs on the charcoal grill.  8hrs … that’s right – a lot of time to spend on your own by the BBQ.  My mind started wandering after a few hours and I caught myself thinking about all sorts of things, the analogy between slow cooked meat and the long term effects of advertising being one of them.  It’s not a great analogy – I admit.  It takes time to slow cook a brisket just like it takes time for advertising to truly affect behavior.  That’s about where the comparison ends.

Paul Feldwick¹, former global planning director of DDB, used a much better metaphor.  He once said brands need to go to the gym if they want to be big strong and healthy and he described how brands are built through repetition over a longer period of time – very much in the same way people build muscles.  This metaphor seems to be supported by empirical evidence.

Fact 1 : Advertising efforts rarely pay themselves back in the short term. 

Dyson (2008)² gives an overview of 5 recent studies that show that every 1$ spent on advertising on average only generates 50c in payback in the short term.

Most advertising efforts don’t pay back in the short term.  Going to the gym doesn’t pay off initially either.  It cost a lot of money and effort and you get can see very little results in the short run.

Fact 2 : Advertising’s long term effects are substantial

Plenty of evidence exists that the long term effects of advertising can be substantial.  The most famous and most often quoted piece of research was published by IRI⁶ in 1992.  They selected 44 advertising tests that generated significant results in the 1^st year and that received no additional advertising support in the 2 subsequent years.  The average incremental sales of these tests in year one were 22%.  The incremental effects in year two were 14% and in year three 7%.  This means that the long term effects (year 2 + year 3) doubled the overall impact of the results.  IRI also found that long term resulted from existing buyers buying more or more often rather than new buyers coming into the franchise.  This suggests that the advertising caused trial which resulted in short term effects and that some of the customers who tried would continue to buy as a result of a positive experience, resulting in longer term effects.

The following table gives an overview of some of the evidence for the existence of long term advertising effects :

The examples above all demonstrate that the sales effects of advertising can carry on in the long run.  Advertising can also impact long term price elasticity.  This means that brands can maintain a higher price level as a result of advertising.  A good example of this is the work done by OHAL¹⁰ in the UK sanitary protection market where prior to 1990 no TV advertising was allowed.  They found that customers were les price sensitive after advertising was allowed than before.  The price elasticity dropped from 2.5 to 2.

Other evidence for long term advertising effects comes from cases studies of individual brands such as Andrex¹¹, BMW¹², PG Tips¹³ and Stella Artois¹⁴.

It’s clear that the long term effects of advertising can be just as substantial as the transformation your body will go through over time if you start going to the gym 4 times a week.

Long term effects are rarely measured

Because the long term advertising effects can be so large, omitting them from measurement can easily lead to wrong decision making.  Broadbent (1999)¹⁵ called quantifying the long term effects of advertising the most important task facing advertising researchers.  He said that a lot of work still needs to be done in the area and that this work shouldn’t just be done in tough times.  In spite of the obvious importance of quantifying long term advertising effects very few companies are doing this.  This is for the following reasons : 

Measuring long term effects is hard : In the long term, many factors other than advertising impact consumer behavior, not the least the consumer’s experience with the brand itself.  This makes it very hard to isolate the long term effects of advertising.  It usually requires advanced econometric modeling and long term in market tests set up with “clean” experimental design.

Long term insights are not always immediately actionable : Marketers cannot wait one to three years for results on long term effects to become available.  They need to make decisions almost instantaneously.  This means that in order for long term measurement to affect decision making today it should be focused on finding predictors of long term performance that can be observed in the short run.  This adds to the complexity.  Measuring alone is not enough, prediction is required for the work to be actionable.

The abundance of short term data : The advent of digital has made a lot of data available that is enthusiastically being used by marketers to better understand performance.  Most of this data is available with the speed of light which means it can be used to get a very quick read on a campaign’s short term results.  This however has shifted the focus of most marketing analytics efforts to the short term.

Short term pressures : There is currently little appetite for the long term view on marketing performance.  Marketing budgets are under pressure and marketers are forced to demonstrate pay back – preferably in the short term.

Shareholders care about long term effects

One common misconception is that shareholders don’t care about the long term.  Barwise¹⁶ gives seven proof points of how shareholders are most interested in maximizing long term shareholder value.  Perhaps the most compelling proof point is the substantial value they attach to intangible assets and brand value as illustrated by the graph¹⁷ below.

So we know that advertising hardly pays back in the short term but that it can generate substantial returns in the long run.  We also know that, contrary to common belief, shareholders care about long term effects of advertising on business performance.  Advertising, as an industry, would therefore probably do well with creating a much better understanding the long term effects of their efforts. 

REFERENCES

1.P Feldwick, Building Brand Muscle, Admap, July 2002

2. P Dyson, Cutting adspend in a recession delays recovery, WARC Online, March 2008

3. M Campbell: Is ROI dead?, Admap, March 2005 

4. Deutsche Bank: Commercial noise – why TV advertising does not work for mature brands, Ad Age, May 2004

5. A Ruffle, ROI: A passing fad or enduring trend? Admap (China Supplement), February 2007

6. E Ephron & G Pollack, The curse of Lord Leverhulme, Admap, July/August 2003

7. L Lodish & B Lubetkin, General truths? Nine key findings from IRI test data, Admap, February 1992

8. M Hess & G Ambach, Short and Long-term Effects of Advertising and Promotion, American Association of Advertising Agencies, 2002.

9. S Broadbent, What Do Advertisements Really Do for brands?, International Journal of Advertising, Vol. 19, No. 2, 2000

10. P Baker, Surely there are lasting effects of advertising?, Admap July 1999

11. Andrex sold on a pup, M Stow, IPA Awards, 1992.

12. T Broadbent, BMW how 15 years of consistent advertising helped BMW treble sales without losing prestige, IPA Awards, 1994.

13. C Cooper, L Cook and N Jones, PG Tips how the chimps have kept PG Tips brand leaders through 35 years of intense competition, IPA Awards, 1990.

14. P Brown, Stella Artois reassuringly profitable, IPA Awards, 2000.

15. S Broadbent, Tough Times, Admap April 1999

16. P Barwise, Advertising for Long Term Shareholder Value, Admap 1999

17. P Doyle, How Shareholder Value Analysis Re-defines Marketing, Market Leader 2000

A couple of weeks ago, McKinsey came out with the results of their study on “Measuring Marketing”.  In one of their questions they asked whether companies planned to change their marketing budget in the next 12 months.  The results are displayed below.

While the overall picture is negative, I was surprised by the spread of these answers.  45% of the companies were planning on reducing their budgets, 27% had no changes planned, 20% of companies were planning on increasing their budget and 9% didn’t know.  What is driving these widely different investment strategies?  I would like to think that they are the result of thorough analysis and deep understanding of what the effect is of different marketing investment levels on business performance.  However, something tells me this is probably not the case. 

The truth is that very few companies understand the business impact of spending more or less on marketing.  When it comes to setting the marketing budget, most of them use very arcane and rudimentary methods to make multi-million dollar investment decisions.  Below is an overview of some of the most popular methods.  It’s a long list but it’s by no means exhaustive (if you know of any other articles please post them below).  I do think it captures most of the methods companies use today.

1979 – Harry’s Rules

Exactly 30 years ago Harry Henry published an article in the Cranfield Broadsheet that described 15 approaches to determine how much to spend on advertising.  They are :

1.      Intuitive / rule of thumb – “enough to do the job” based on experience

2.      Maintaining previous spend, sometimes inflation adjusted

3.      Percent of previous sales – backward looking, compounds failure (or rewards success)

4.      “Affordable” – what’s left after cost and profit requirements are met

5.      Residue of last years profits – focuses on source of funds, not their use

6.      Percent of gross margin – begs question of cost efficiency

7.      Percent of forecast sales – most common method used

8.      Fixed cost per unit of sales – like % of turnover

9.      Cost per customer/capita – mostly business to business

10. Match competitors – assumes they are right

11. Match share of voice to brand share – like the above

12. Marginal return – direct response approach

13. Task approach : define objectives and cost out how to reach them – best in theory but may require modeling

14. Modeling – the most sophisticated approach: not easy

15. Media weight tests – looks empirical but usually difficult to evaluate or replicate.

Ten years later, Simon Broadbent included a very similar list in his book The Advertising Budget (1989). 

2003 – Harry’s Rules Revisited

In 2003 Gullen wrote an article in Admap where he introduced 5 steps to effective budget setting.  He categorized the budget setting approaches in 5 independent schemes or “budget bearings” as he calls them :

1.      Inertia bearings based on usual practices (ie last year’s budget maybe adjusted for inflation)

2.      Business bearings based on sensible ratios within the business plan (eg affordability, advertising to sales/volume/margin ratios)

3.      Media bearings based on the cost of fulfilling a sensible media plan (eg. Cost of media plan to achieve goals)

4.      Competitive bearings based on the spend of competitors (eg match share of voice to share of market)

5.      Dynamic bearings based on the observations of apparent effects of previous advertising activity (eg. in market tests, econometric modeling)

2004 – Harry’s Rules Revisited again

In 2004 Dyson wrote another Admap article on “How to budget better”.  He outlined 7 categories of methods :

1.      percentage of sales : the ad budget is a proportion of last year’s actual sales, or next year’s forecast sales

2.      objective and task : objectives are set (turnover, profit, growth) and the budget required to meet these objectives is estimated

3.      competitor : the amount spent by competitors is used as a yardstick; a version of this is the well-known rule of thumb that share of voice should be at least equal to share of market

4.      affordability : the budget is the amount left after everything else has been accounted for

5.      historical : do the same as last year, with an adjustment for inflation

6.      executive judgment : basically guesswork, but probably an informal use of one or more of the above.

7.      brand led : a more scientific approach that uses research data and econometric modeling.

2006 – Harry’s Rules Revisited one more time

Finally in 2006 Green published an article on “How much should I spend on advertising?” on WARC.  He said that there are roughly 2 categories:

1.      Task based approaches that start with the financial goals of the brand (eg rules 1-11 in Harry’s list).

2.      Resource based approaches that start with what resources are available or assume that what a company has been doing in the past will work in the future brand (eg rules 12 and 13).

As you can see, the methods for setting the marketing budget haven’t really changed in the last 30 years.  Dyson quotes research that shows that “advertising as a percentage of sales” is by far the most popular methodology (53% of companies seem to use this approach).  That is scary!  But it confirms my personal experience.  Scientific methods, some of which have been around for a long time and are tried and tested, often seem to play only a very small part in the budget setting process.  Marketers tend to stick to the methods they know, even if they know they might be over-simplistic or even wrong.  

Will the current recession force them to adopt more scientific methods?  I think it will.  It has put the spotlight on the marketing budget and there is no place to hide.  But over the last 30 years marketers have created some bad habits that might prove to be hard to break.

NBC started this three years ago when it guaranteed engagement scores for Toyota.  Then just before the upfront in 2007 NBC extended this offer to all advertisers.  Now Jim Farley, the Toyota CMO who brokered the original deal, has taken it to his new employer Ford.

Used on their own, traditional Nielsen measurements have always seemed inadequate to me as a way of assessing media value.  We all know from our own experience that simply having the TV on means nothing.  Your interaction with it can range from active avoidance to rapt attention, yet up until recently these would have been measured as the same impression. 

About fifteen years ago I undertook a qualitative research project for Barclaycard, the UK’s biggest credit card, to look at different types of TV watching behaviors.  We identified five different types:

1. Golden Hour:  Your favorite show of the week.  You’ll stay in to watch it. 
2. Appointment:   You plan to watch these shows  (in the days before DVR’s)
3. Regular:  You click around to find something to watch
4. Fodder:  You’re barely watching
5. Radio:  You’re listening to it in the background, using it as a radio

This is pretty basic stuff, but the principles hold true:

1)     All TV programming is not created equal.  

2)     Impact = numbers of people watching X engagement. 

Now marketers have really embraced this notion.  An article in this week’s Business Week estimates that, “more than half the top 100 US advertisers are now pegging prices to viewer engagement”.  

The current leader in engagement measurement is IAG, which Nielsen bought last year.  It uses a simple method: sampling from a panel and asking them ten questions about a show.  The more correct answers, the better the engagement. 

IAG uses the same system to measure engagement in the advertising as in the TV programming, ensuring a like-for-like comparison.

“It found, for example, that while the Discovery Channel’s (DISCA) Dirty Jobs series, starring Mike Rowe, delivered puny ratings points, the engagement level of the show’s viewers is off the charts — and proved to be a ripe demographic for Ford trucks. That led Ford not only to advertise on the show, but to strike a deal with Rowe, who appears in Web videos for Ford showing how tough the F-Series pickup is. “It’s a hugely efficient buy for us, but none of that happens without the kind of data we can get now on what people are tuned into, for how long and how closely they are watching our ads, because the Nielsen ratings alone wouldn’t have led us there,” says James Farley, Ford’s chief marketing executive.”

This sort of deal takes us one step closer to the future of communications planning, where the objective is not maximum tonnage but an environment that becomes part of the message itself.  

As you will have surely heard, Facebook recently acquired its 200 millionth user.  It has achieved this in under five years, doubling in size in the last 8 months alone. 

This has produced a lot of commentary.  Since there is so much data surrounding people’s interactions with Facebook, it can be examined in all sorts of social studies.

One in particular caught my eye last week.  The New York Times article had a sidebar produced by Facebook’s analysts, which is a wonderful piece of data visualization. 

There are two parts.  The first shows the global expansion, in a series of step changes precipitated by changes in the rules of who could join.  It plots this against the average age distribution of the members.  Facebook famously started at Harvard in February 2004 and quickly expanded to other universities, and you can see that initially all the members are 18-24 in the first graph.  In 2005, it expanded to include most American colleges and high schools.  In February 2006 it allowed anyone to join, and the age profile increased.  By February 2007, it had reached 50 million members.  By February 2008 Facebook is translated into more than 40 languages.  Growth is fastest among those over 35. 

For me, the second set of diagrams is the most fascinating.  This is a real life example of what I’ve been calling ‘The Ecosystem of Influence’.  It illustrates the Facebook network of one employee.  The friends are dots, and the interactions are lines between the dots.  It looks like an airline route map. 

The point of the diagrams is that this individual has four networks of different sizes, defined by different levels of social intimacy.  He has 178 friends in total.  But he reads the postings of a much smaller group, and sends messages to a much smaller group yet again (“one way pings”).  And of these only some reply.  Without counting the lines, this looks like maybe 15