Thursday, March 25, 2010

MEDIAPOST: Building or Buying The Right Model; I Say Build For Now

Our business is maturing pretty rapidly but one of the biggest issues is that we don’t have any easy way to make an equivocal comparison to the impact we make on driving appreciable business growth.

Let me elaborate a bit.

If you exist in the world of traditional marketing you know there are many different options for forecasting the effects of media spend on sales. Any large marketer who spends multiple millions or billions of dollars per year on marketing has a media mix modeling system that they use to input a set of assumptions and projected spending and this model spits out a pretty decent forecast of what said marketer can expect to see as ROI. These models certainly have flaws and in many cases these models are tailored to be proprietary based off of an initially offered marketplace product, but they are consistently flawed and marketers tend to use them as directional data. The issue is that while digital is maturing quickly and more tactics are being considered, these tactics are not capable of being included in the models because they aren’t updated frequently enough and with an accurate level of data that would make them statistically valid.

It starts with the syndicated research tools. Many of these tools only track display and search. Most are not capable of tracking social media, CRM, or viral efforts nor are they capable of accurately tracking many ad networks, specifically those with behaviorally targeted components to them. These are tactics that are fast becoming an integral part of digital strategy. They are difficult to track because they are not standardized formats, like banners and buttons, but they have a decided impact on the ROI of a digital campaign. They are also a little difficult to track partly because they are so targeted that they are difficult to understand the duplication patterns with other media. The third reason they are difficult is because the creative impact is far too variable vs. that of a standardized TV or print ad, which has years of historical impact factored into it.

The flip side to this whole discussion is that many of these ROI predictor and media mix models are the bluntest of instruments, to say the least. They may have a plus or minus of 15% on actual sales delivery impact and may not have been robustly updated since the eighties, but marketers use them and in some cases they treat them as gospel. These tools are built for a broadcast-centric landscape and are not easily tailored to an engagement-centric landscape where the desired impact of marketing is to involve in conversation as well as create it. If I were a betting man and I had a scientific background or one in economics, that’s where I would spend my time to become a millionaire. I would try to create a new model that takes into account the ability to be involved, not just to inspire.

But I digress; a good carpenter never blames his tools.

You can go on and on all day trying to talk about and point out all the flaws in the model but it still comes down to understanding and being able to build off what has come before you by factoring in what you know to what is already understood. Many smart people and many smart companies are tackling these issues on their own and doing a great job of it. Strategists are addressing the model and building their very own proprietary models (we have our own at my company). The partners you work with should be able to explain to you what they are working on and validate their assumptions for you and your business and if they have not progressed to working on these yet, you should push them to do so. Fundamentally what works for you may be exclusive to you, but if you can consistently use your model and get a read on how it correlates to success, then you can plan for the future.

Hopefully, sometime in the not too far distant future we’ll catch up as an industry, but don’t wait around for us. Things are changing just a little too quickly.

1 comment:

Anonymous said...

If you lend someone $20 and never see that person again, it was probably worth it.............................................