Rohit Bhargava’s post, OSG: The Secret Metric for Measurement Every Marketer Wants, has been bugging me ever since I first read it a couple of days ago. And it’s not just the redundancy of “metric for measurement” (as opposed to a “metric for yodeling?”). What has been sniggling away in the back of my brain is that Rohit’s point is based on two premises, both of which I wholeheartedly agree with…but then he combines them in a way that makes me uncomfortable.

Premise No. 1 — The “Paralyzing Factor” of the Status Quo

A very, very true observation: “It is no longer an issue of having the technology available to measure things online, but rather the analytical ability to hone in on the metrics that really matter.” The blessing and curse of so many of our interactions occurring electronically is how much more measurable those interactions are:

• The blessing: the potential for richer, deeper insights
• The curse: trying to garner a tablespoon of insight from a fire hydrant of data

“We” assume that more data is better. It’s a logical assumption. It just doesn’t tend to work out that way.

Premise No. 2 — The Opportunity to Visualize Data More Effectively

Rohit’s inspiration is Aaron Koblin, who makes art from data. My inspiration is Hans Rosling’s Trendalizer (Rosling is now at Google, and Trendalyzer-like functionality is now available to the commoner…for free…courtesy of Google’s Visualization API). By all means, we overuse The Basic Charts in Excel: pie charts (<shudder>), bar charts, and line graphs. This relates to Premise No. 1 — no one got fired for buying IBM, and no one will get fired for presenting data the same ol’ way that everyone else presents it. Even worse, many “advanced” Excel users simply focus on putting more data on their charts: using two Y axes with different scales, introducing a third dimension to the chart, or combining lines and bars. The result? Generally not good, as the charts are overwhelming.

There are definitely exceptions here. Rosling is one, and there was a great analysis from TechnoSocial of some Twitter activity where the visualization of the data was very effective (although this was visualization for analysis — not for metrics).

Premise No. 1 + Premise No. 2 =

I agree with the problem statement (Premise No. 1), and I agree that there is a huge need for better data visualization across the board (Premise No. 2). I don’t, however, see improved data visualization as the critical path to more actionable metrics.

Marketing is complex. The data is abundant. Searching for One Sexy Graphic is as useful is an exercise in tilting at windmills. A more actionable approach to finding actionable metrics for measurement:

1. Identify clear objectives before you even think about how you are going to measure them — get “the boss(es)” input and buy-in on what those are. This is the key for Rohit’s point that marketers’ need to “explain to their bosses what they are doing.” The objectives need to be crisp, simple, and agreed to by the stakeholders.
2. Step back and identify all of the ways you can measure progress towards those objectives — directly or indirectly; in some cases, the best you will be able to do will be to come up with a couple of reasonable proxies for progress. Again, get buy-in from the “boss(es).” Hone in on the 1-3 metrics that best reflect the objective.
3. Set targets for those metrics. Out of the gate, these may be total SWAGs, but it is imperative to set some sort of target for what success might look like (and…yes…industry benchmarks, for all of their shortcomings, can help you narrow down what you think you can achieve when establishing a new metric). Broken Record Alert: get input and buy-in from “the boss(es)” on what these targets are.
4. Visualize the data effectively (see Premise No. 2 above). The driving force behind the visualization? Tying the data back to the objectives, metrics and targets you established.

This all sounds a bit abstract, I realize (with that realization being heightened by the fact that I just finished reading Made to Stick). I’ll put it on my To Do list to get a more concrete example posted at some point. I have the examples — the trick is mucking them up sufficiently to protect the innocent!