Media, A Wharton Professor and Marketing Research - I

A couple of weeks ago I had the opportunity to actually attend a lecture by a Wharton professor - Peter Fader - at the Marketing Modelers meeting down at the ARF in NYC. The topic of conversation was "The Paradoxes of Interactive Media". Peter Fader is a professor at the Wharton school of business and is actually on the board of A-list journals such as Marketing Science and Journal of Marketing Research. He has built his reputation on his research on trial and repeat in the CPG industry, while also doing some mean research in the field of electronic commerce. Most notable was his testimony during the Napster trial. Based on that testimony it was clear that the man is a rebel and revels in lateral thinking. The talk just confirmed it.

Even though I - or most other people in the room - didn't agree with everything he said, but his approach prompted me and everyone in the room to question our beliefs and conventional thinking. In today's blog I am presenting one of the points made in the session.

Professor Fader iterated that cross-group differences across different demographics like ethnicity are often meaningless. Now this may be true from a total category point of view (he used the example of DVD purchaes by hispanic vs. non-hispanic consumers). But if you get down to the brand or attribute level, this actually leads to way different consumer behavior - and since marketing is brand-driven rather than category driven, I would have to say that distinct differences do exist and can be leveraged by marketers to drive sales differently among different demographic groups. However, his essential theory that people are the same (or distributed similarly/ normally, if you want to be statistical) everywhere is pretty solid. However, as soon as you start breaking down a category by its attributes (brand, size, flavor, feel etc.) ethnic differences come into play. I would love to hear other points of view on this. I will deal with a couple of other interesting points made by professor Fader in my next posting....

P&G Giving Up on Facebook Marketing?

I was just reading an article by Jack Neff of AdAge covering P&G "Digital Guru" Ted McConnell. Ted believes that Social Networks like Facebook may never be able to show the ROI on Ad dollars marketers spend on their websites. The article quotes Ted as saying about consumer-generated Media "Who said this is media? Media is something you can buy and sell. Media contains inventory. Media contains blank spaces. Consumers weren't trying to generate media. They were trying to talk to somebody. So it just seems a bit arrogant. ... We hijack their own conversations, their own thoughts and feelings, and try to monetize it." I am not sure if someone rewrote the English language dictionary but last time I checked Media is defined as "means of mass communication" and with a Reach of 12% of global internet users according to Alexa, Facebook certainly fits that bill. Now is it a good medium for advertising, that's a whole another thing. Ted also raises concerns about the type of targeting afforded by Facebook, but that's a fallout of the Information Age. There was a lot of hue and cry about using Credit Bureau data for marketing, a few regulations later that is still an industry. Ted makes a good point about reach fragmentation though- there's just way too much people do online to effectively reach them with any decent amount of banner ads. On the other hand as technology advances the ability of online advertisers to track a target through their internet trail and persevering until a conversion is obtained isn't that difficult. Already re-targeters are identifying unique users and repeatedly hitting them with ads- check this article.

Return on Product Innovation: Measuring your Innovation Pipeline

Innovation is a critical growth driver for most industries, but more so for industries that are mature. Growth industries are less reliant on an ongoing pipeline of innovations because the full potential of the existing portfolio hasn’t been maximized yet, penetration can be further increased and new markets can be expanded into, where success with existing products can be replicated. Products and brands in mature industries on the other hand are characterized by a lack of differentiation outside of price- barriers to entry are low, which increases the number of market players, pushing marginal profits down. In such an environment, innovation provides a strong differentiating factor, allowing a brand to lower dependency on price as a competitive lever.
So if you are responsible for the strategic planning for your firm and not in an early stage industry, you need to be thinking about your innovation pipeline and it’s not enough to say you have a department for innovation- in most industries only 1 in 10 innovations succeed. So you not only need to have a team in place that has a network reach both inside and outside the organization that allows ideas to funnel up, but you need to also have the right metrics in place to evaluate the performance of your innovation strategy vis-à-vis your industry. A study by McKinsey (McKinsey Global Survey Results: Assessing innovation metrics, October 2008) suggests that a large percentage of executives even at companies that actively pursue innovation don’t formally assess innovations at all.
One way to evaluate innovations is using Return on Product Innovation (ROPI) measured through in-market tests (in-market tests are also risky because your competitors can copy it and bring to market faster than you, stealing your thunder). For ‘breakthrough’ innovations that you are planning to take straight to the market without first testing, ROPI can be estimated as ‘one-year out ROPI’, ‘two-year out ROPI’ and so on. At the end of year 1, forecasts can be used to estimate breakeven time for ROPI to turn positive and marketing ROI can be used to evaluate opportunity to optimize marketing strategy to improve ROPI.

ROPI={[Dollar Sales-Cannibalized Sales]/ [Fixed Cost + (Variable Cost*Units Sold)]-1}*100

Fixed costs can include development or other one-time costs related to production, variable costs are usually ongoing production, marketing and distribution costs. You need to deduct cannibalized sales, because these are sales you would have gotten even without the innovation. This equation can be modified for any custom inputs particular to your industry or the nature of innovation. For instance, if estimating ROPI for an in-market test then using the full fixed cost for development is not fair and should be factored down based on the ratio of size of market tested vs. the total market-size.

Predictive Targeting In Digital Media Marketing

I was just reading an article on MarketingProfs.com that was talking of leveraging a technique called "CARVER" (reminds me of Thanksgiving!) used by the military to "identify and prioritize" targets ("How to Target Your Prospects With Military Precision" http://www.marketingprofs.com/8/target-prospects-with-military-precision-meachum.asp?sp=1). This won't be the first time ideas from the military have been leveraged in business 'warfare'. In fact Precitive Trageting is exactly the area where another concept from military surveillance has been used- Receiver Operating Characteristic Curves or 'ROC' Curves. ROC Curves are used to evaluate how well predictive models are able to identify targets that are most likely to respond to specific media tactics. The concept is based on the effectiveness of Radars to identify targets by sifting signal from noise. ROC curves are used to evaluate econometric models that identify prospects that are most likely to respond to media tactics. Wikipedia has a good explanation of this approach: http://en.wikipedia.org/wiki/Receiver_operating_characteristic_curve. Although the CARVER concept is interesting, the optimization of the weights for each factor seems a bit subjective inc comparison to the scientific precision provided by an econometric model. This link provides a fairly decent expanation of the method: http://www.ni2cie.org/targetanalysis.php.htm
Compared to the subjective approach utilized here, an econometric model not only helps identify most opportunistic targets, but also quantifies specific relationship between probability of response to marketing and amount invested in each media tactic available. Of course to do this successfully you do need a training sample based on historical programs, so if the CARVER approach doesn't need any historical data, there may be something there.