How to safely experiment with social media marketing

  • Previous
  • 1 of 4
  • View as single page


At some point in your career, you've probably been in the position where you think you are doing a phenomenal job marketing your organization. You are proud of the bulletproof strategy you've created, pleased with the tactics you are using to achieve the strategy, and thrilled with the measurable results you are able to report. But one day, you get the message from leadership that "everyone else is using (insert latest internet fad here), why aren't we?" The short answer would be that it doesn't fit into your current strategy, but when is the right time to introduce something new, and how do you do it in a way that is not disruptive to the rest of your strategy? You shouldn't always have to wait for pressure from another person within your organization -- sometimes you just need to experiment a little bit in a safe environment without detracting from your marketing plan.

How to safely experiment with social media marketing

We all know that an Instagram profile, a crowd-sourcing contest, and a Pinterest board are not strategies. As marketers, we take care to research and analyze the best communications channels for our unique audiences and the way we talk as a brand. But what happens when you've invested time, money, and effort into a comprehensive, long-term marketing strategy only to have a phenomenon like Pinterest pop-up out of nowhere and have to answer the "why aren't we there yet" question?

Because you hold yourself to measuring ROI on your marketing initiatives, how do you guarantee the "R" before you make the "I?" Even though the higher-ups in your organization (or in the case of an agency, your clients) are asking for something new and cutting edge, you know that you will be held accountable if it fails.



Sione Palu
Sione Palu November 18, 2012 at 8:00 PM

Greg, here is an informative paper from fromer head of Amazon data-mining, (Dr) Ron Kohavi who now works at Microsoft. The paper was submitted and appeared in ACM data-mining proceeding, KDD (2007). "Practical Guide to Controlled Experiments on the Web: Listen to Your Customers not to the HiPPO". Here is the download link : []

Sione Palu
Sione Palu November 18, 2012 at 11:08 AM

Greg, I think that experimentation is the way to go. There's an article from a few years ago on ZDNet (its still there) with the title : "Microsoft looks to make product planning more science than art" where they used experimentation. The tech-lead was (Dr) Ron Kohavi who was the head of data-mining at Amazon and that's exactly of what he did at Amazon when he was there before he left to work at Microsoft. Although the experimentation that Kohavi has been doing is not exactly the same as the experimentation that you cover in your article here, but the concepts are similar.