In my last post, we took a look at a framework for optimization that creates results that will inform future programs and tactics. Using Scientific method, we created control and test groups to prove out our hypotheses. We also looked at ways to increase our testing flexibility using test matrices allowing us to test multiple factors simultaneously to get the best combination of variables. Now let’s look at a real-world example that might challenge our assumptions.


Deodorant goes Viral!
Let’s say our good friends at Acme, a Consume Packaged Goods company, is releasing a new mens’ deodorant brand. The marketing brain trust has spent plenty of time, resources and money testing and validating the brand positioning and customer value proposition. Acme has settled in on a particular brand voice that is best presented in a humorous light.

Testing on several control groups have settled on a particular series of video clips that have shown the potential to “go viral” and spread the new brand’s message over the Internet for very little money. The demand gen team packages and presents the idea to brand leadership and everyone agrees the videos are engaging and hilarious and will definitely spread like wildfire over the Internet, so the project gets the green light. Time to light ‘em up! The Demand Gen team posts the videos on You Tube and eagerly waits for the results.

Within hours the video does, indeed, go viral as expected. Views are in the millions, and rapidly increasing. Within days the videos are at the very top of the view lists and are making the rounds via Social Media channels. All is good, right?

And now, a word from our sponsor…

Expectations are high, but word from the distribution team is the product is staying on the shelves in droves. Expected backfill orders are non-existent. The videos are obviously a smash hit, spreading the new brand across the Internet at record speed. Why no product sales as a result? Quick, somebody check their Twitter account to see what’s going on!

Houston, we have a problem. While everyone seems to absolutely love the humor in the video, the reaction is not exactly what product planners have in mind.






Yes, the videos are hilarious, but nobody wants to be associated with a product whose promotional videos depict its users as complete idiots. And it’s too late for those tens of millions of viewers to “un-see” those videos.

What should Acme have done?

This is a clear example of where spending a little time and money on a limited pilot could have saved a lot of embarrassment later. Executing tests in a controlled environment with limited, but representative audiences can provide insight into those unintended consequences. These pilots should concentrate on understanding not only what will happen, but why it happens.

Change the conversation.

Your pilot should be conducted with the end objective in mind. These pilots are good at bridging assumptions you may have made in causal relationships. In this case, ACME assumed that millions of Internet views would cause millions of product sales. Instead, those views led to brand ridicule and corporate embarrassment.

Pilots can also be used to test specific audience segment reactions against the whole. It is a great way to determine how message/offer variants track in one segment vs. another. Do customized messages really improve responses? Better to pilot and know than spend additional time and money only to find out after the fact you generated no additional lift for your efforts.


Pilot before committing to an untested plan.

Pilot with the end objective in mind

Pilot to correct causal assumptions.

We’re now piloting testing and optimizing every bit of success out of every program we execute. It’s time to tell your C-Suite about the fantastic work you are doing. Unfortunately, Marketing historically has done a less-than-stellar job of reporting on its success. In our next installment, we’re going to begin a series on reporting to overcome that legacy: Create Marketing Reports That Will Thrill Your C-Suite!