Skip to content

A Portfolio Approach to Marketing ROI

2010 June 30

The media and marketing industry, much like the financial industry before it, has become blinded by metrics that they don’t understand.  Somehow, the notion of accountability has morphed into a misguided quest for the sure thing in the form of ROI metrics for just about everything.

This is silly. In actuality, when we talk about marketing ROI, individual actions don’t account for much.  What really matters is portfolio performance and that’s what both buyers and sellers need to focus on.

To see why, we only have to look at the mess the financial industry made when they ignored the simple fact that every investment involves risks that can’t be quantified.

Portfolio Selection

In 1962, Harry Markowitz published his paper on portfolio selection, which was to have great influence on the development of modern finance.  The basic idea was that investors should not only seek returns, but look to minimize risk.  This was a much different way of looking at finance.

He also showed that by diversifying, better returns could be gained with lower risk. In fact, buying an uncertain security could be a very good idea if the risks are uncorrelated with the rest of the portfolio. In effect, sometimes an investment which is volatile by itself can increase overall portfolio stability.

Although this is standard thinking today, at the time it was revolutionary.  In 1990, Markowitz was awarded the Nobel Prize for his work on portfolio selection.  It was a true achievement, but it didn’t stop there, it helped form the foundation for much of modern finance.

Valuing Investments

Markowitz’s colleague at the RAND Corporation, William F. Sharpe, took the reasoning a step further.  He came up with an ingenius way to value investments using the portfolio selection model as a basis.  He developed a measurement of risk, called beta.  (The higher an investment’s beta, the more volatile it is compared to the market).

Sharpe then realized that companies could use beta to not only value securities, but also to evaluate operational investments such as plants and equipment through a purely quantitative process.  The result was the Capital Asset Pricing Model (CAPM).

A CAPM example:

Riskless Rate of Return: 2%  (i.e. Treasury bonds):

Market Risk Premium: 5% (i.e. difference between S&P 500 and Treasury bonds)

Company Beta: 2

From there, you can just “plug and play.”  You multiply the market risk premium and the beta, then add the riskless rate of return:

5% (Market Risk Premium) x 2 (Company Beta) = 10% + 2% = 12%

Therefore any investment within these parameters should have an expected rate of return of at least 12% to be an attractive investment.

The result was a new era of financial engineering. The idea was that through computing power and powerful mathematical models, you could not only improve return on investment, you could eliminate risk through quantifying and then controlling it through hedging techniques.

It seems that a similar overconfidence in numbers is now infecting the marketing industry.

Quantifying Risk

The financial revolution then progressed even further.  The development of the Black-Scholes model generated metrics that quantified not just investments, but risk itself.  This made possible a large market for derivatives; securities that weren’t actually based on assets, but rather derived from their price movements.

Soon, traders were not just trading based on prices, but volatility (a surrogate for risk).  No longer was the object to “buy low and sell high,” but to use complex hedging strategies to engineer enormous returns safely.

The old clubby world of finance became a relic and was replaced by “quants” and massive computing power.  Intuition was superseded by mathematical models, relationships by high speed internet connections and judgment by computing power.  Fear was for wimps and dullards who didn’t understand the wonders of risk management.

The result: disaster.

The Problem with the Engineering Approach

All of this was completely logical and widely accepted.  The problem is that they were using the wrong math.  As Mandelbrot and others have shown, people can’t be described by the same principles as atoms and celestial bodies (where Gaussian based models work well).

The problem with statistical models is that they are often based on an independence assumption, which is not valid when markets full of interdependent people are involved.

This should have been clear back in 1998, when the collapse of LTCM nearly took the world financial system along with it.  The firm was led by the Nobel prizewinning team of Myron Scholes and Robert Merton along with an impressive cadre of PhD level econometricians. No one knew the models better and no one failed more spectacularly.

In their quest to optimize returns, they became all too enamored of their mathematical certainty.  Instead of diversifying their risk, they made highly leveraged bets that were in fact highly dependent on the Russian economic crises.  Within a few months, the firm had lost $4.6 billion.

Today, you can find new marketing wizards who believe they can optimize marketing returns for specific media, just as LTCM thought they could calculate risk so accurately that leverage didn’t matter. They haven’t won any Nobels, but their ideas stem from the same flawed reasoning.

Media Portfolios

One of the most emphatic claims of the new media gurus is that if marketers knew what they were doing, they would act much differently.  A good way to test that notion is to look at some media market portfolios and see if they make any sense.

The above charts show the biggest developed markets – the US and Western Europe – side by side.  What should be immediately apparent is how similar they look, even given vastly different market structures and cultures.

The other thing that stands out, given all the hype about budgets moving to digital, is how small the internet share is.  About 85% of budgets still go to traditional media.  Although the data shown is for entire markets, not individual advertisers, these baskets make perfect sense according to a portfolio approach to ROI.

Firstly, people tend to be varied in their activities, so it wouldn’t be smart to bet all of your marketing money on just one channel.  Moreover, because new media is untested and therefore more risky, new media would have to far outperform traditional media in order to get the same budget allocation.

When we look at two sample developing markets, we can immediately see that TV gets a much higher share (mainly because of cost) and digital is significantly lower (no surprise there).  Yet what’s really interesting is that the baskets are much less diversified generally.

Moreover, not only do they differ from the developed markets, they differ from each other.  As Tolstoy wrote, “All happy families are alike. Unhappy families are unhappy in their own way.”

The specifics of why they look this way have to do with internal market characteristics (which I won’t go into).  However, the reality is that if they could diversify more, they would. It’s just that some areas of the market haven’t developed yet.

The data leads to a clear conclusion:  the real trend is towards media diversification.  If you think that one marketing channel is taking over, you are surely wrong.  No amount of bluster or fancy figurin’ will change that simple fact.

What Marketing Portfolios Mean for Buyers and Sellers

The implications are clear.  Effective marketing is not about picking winners among channels, but by forging effective combinations focused on achieving specific goals.

Here are some things to keep in mind.

Forget about ROI for Individual Actions: Unless you have a communications planning contract and access to confidential client data, forget about ROI entirely.  You simply won’t have any idea what your talking about.

And as Wittgenstein famously pointed out, if you can’t say anything meaningful it’s better to keep quiet than to talk nonsense.  False metrics might make people feel smart, but lead to stupid actions.

Time and energy wasted on trying to calculate the ROI of individual channels would be much better spent on devising an effective portfolio.  The whole is far more significant than the sum of the parts.

Understand the portfolio: It’s is natural for people to want to advocate something.  Sellers, of course, have a fiduciary interest, but marketers often get in the game as well.  It feels good to say that you “believe” in social media, or ambient or whatever.  Nobody goes out and brags that they bought 1000 GRP’s.

However, treating these things separately completely misses the point.  Direct response metrics improve when mass media is on air and social media can vastly extend the reach and involvement of a media campaign through what network theory pioneer Duncan Watts calls Big Seed Marketing (pdf)

Championing a particular marketing channel is a complete waste of time.

Integrate: The biggest problem marketers have today is integrating their marketing programs.  Media has fragmented and agencies have metastasized.  With so many moving parts, getting it all coordinated is a key success factor.

The real benefit to new media isn’t that it has higher ROI, but that it creates options.  Learning how to combine digital with more traditional channels is the fundamental planning challenge today.  Mindless talk about competing ROI metrics for various media only serves to muddy the waters.

The marketing world shouldn’t repeat the mistakes of the financial industry.  ROI for individual investments is misleading and can even be dangerous.  What really matters is portfolio performance and that’s where the focus needs to be, not on devising new and complex ROI metrics.

Information is a poor substitute for understanding.

– Greg

10 Responses leave one →
  1. StuartNicholson permalink
    June 30, 2010


    I agree with most of what you have said.However there is a need to an extent to continue to make comparisons of the performance of one platform with another as the reality is that marketeers continue to have spearate budget “siloes” that they need to make choices about in terms o where to spend the proportion of their budgets.

    The siloes however are more likely to be between “media” and other expnditure areas such as In-store, events, other forms of btl etc.
    While there still seems to be a division made by marketeers between digital communcation inestment and offline, that is being eroded very quickly through the interchangeability of platform (nternet on TV, TV on computer etc) to the extent that while its true that at the moment many look for ROI comparisons btween say internet and TV inorder to justify their investment decisions, that will quickly become irrelevant as platforms will be so mixed up that noone will care too much any more.The common bond is going to be the content.

    As all media becomes digital in nature it will in any event be possible to measure ROI more easily as the range of metrics available are greater.These metrics will allow us to make cross-platform analysis more easily.The skill will be in selection of the most important metrics to use from the many available to use as the benchmark of comunications success.

  2. June 30, 2010


    Your points are well taken. I think it’s a matter of how seriously you take the numbers. Even digital metrics are open to quite a bit of interpretation. So I don\t think it’s a matter of using metrics or not (they don’t give out Nobel’s for nothing), but understanding that they will only take you so far.

    – Greg

  3. June 30, 2010


    Outstanding piece. Your intellect and insight are spot on. I applaud your insight into the analogous relationship between finance and marketing. I would offer one thought: Don’t hold your supposition as a “law” because, like finance, if one were to “own” a particular media through a saturation campaign, one can have very strong returns. Yet, as you point out, the risk is the the key variable. I hope to read more of your thoughts, and welcome your comments on my blog:

  4. Armand permalink
    July 1, 2010


    Very interesting topic. Indeed, information is a poor substitute for understanding, but It’s powerful tool for understanding and can lead to an easier and clear action.

    The most important is how people deal with information they have and how they are able to transform this information into intelligence and knowledge and this clearly involves processes, policies and well developed human capital.

    Information never aimed to be a substitute for understanding, but a support tool and an accelerator for it.


  5. July 1, 2010

    Thanks for sharing your perspective, Armand.

    – Greg

  6. July 2, 2010

    Thanks Peter,

    Good luck with

    – Greg

  7. Alex Bakun permalink
    July 28, 2010

    Nice point on diversification. Very seldom even MNL clients even try to understand this in Ukraine. I guess it’s not only our local failure. As for ROI, for me it’s accountable in modeling or whatever only if it is multimedia (better multi-contact including BTL). Otherwise it doesn’t make a model of reality at all )
    It’s like doing a correllation instead of multiple regression. Though I agree Stewart just like to propose a workaround to return to single-metrics.

  8. July 28, 2010

    Thx, Alex. How are things back in Kiev? Clients spending again?

    – Greg

  9. Mats permalink
    August 4, 2010

    Excellent post – only discovered your blog recently from a link on LinkedIn but will be following you closely from now on!

  10. August 4, 2010

    Thanks, Matt

Leave a Reply

Note: You can use basic XHTML in your comments. Your email address will never be published.

Subscribe to this comment feed via RSS