(First published in the WAO/FACTOR Newsletter in February 2012)
It is almost a cliché to say that metrics should be actionable. No, it is an actual cliché. Books and consultants like myself usually say that so people will avoid measuring, and reporting, on stuff that has either very little value to the business (think hits), or that is very hard to impossible to influence (the weather if you sell umbrellas). So, we all agree that analytics should deliver actionable insights, i.e. new knowledge susceptible to generate action.
What does that really mean?
I identify two types of action a digital marketer ought to master: the capacity to influence a metric, and the capacity to have something done about it in the business, the former being subsumed to the latter, I strongly believe. They are probably rather two stages of the same thing.
As a marketer, what can I do to make a number change in the direction I want? For example, if I need to increase the conversion rate, what can I work on? Many things, really. I can optimize the check out process (rather mechanical), offer a discount (but compromise profitability), better target by traffic sources, or through an email program that allows me to segment on multiple variables. The idea here is that actionable makes it possible for me to undergo clear, concrete, definite actions that will have an impact on the needle I want to move.
Let’s take a KPI that would be “Percentage of Web Sales” which would be a part of my company’s total sales. Is it really actionable? In some ways yes; I can definitely do things (campaigns, UI optimization of all sorts, etc.) that will impact that percentage. However, since a good part of the equation comes from other activities I don’t control, namely non-Web sales, I can see my percentage go down just because my colleagues decided to get busy. In short, actionable needs some element of control.
However, here’s where things get more complicated. Anyone living the corporate life (and maybe even more so in government) knows that “actions” are undertaken within a complex environment of departments, divisions, teams, diverging goals, agendas, all fighting for good grades at the annual performance review.
By that, I mean how people are themselves actually measured in their organizations. It is a fact of work life that people tend to behave the way they are compensated.
I think we too easily forget that measurement is not isolated from the vagaries of organizational life. Metrics, KPIs, optimization processes, etc., although rightly fact-based, are also used within the environment of internal politics, agenda, priorities, and execution contingencies. Should we actually be surprised? Never think that measurement is all about objectivity: it evolves in a socio-political environment that necessarily pre-determines what will be adopted, rejected, or plainly ignored; what will be “actioned”.
We have all seen situations, too often unfortunately, where measurement served an agenda, either of a manager or a team. Most Marketing Analysts, with even only a few years experience, will have at least one anecdote of people trying hard to give a positive spin to results that very strongly suggested otherwise.
In short, metrics are politics, as Pat Lapointe said so well.
Jacques — Sorry to nitpick, but when you say “metrics should be actionable” I must state my objection. You do go on to say that “analytics should deliver actionable insights” which is something I’d wholeheartedly agree with.
But a metric, in and of itself, is not actionable, nor does it need to be.
In an organization, a metric may be established simply as for compensation incentive. In this regard, the metric isn’t designed to “actionable”, which implies an operational connection.
Hi Ron
Well, I see your point, but I wonder if this is not simply a difference in definitions. When you say “a metric may be established simply as for compensation incentive” it seems to me that here the metric in question would be a KPI. But this is not important, because I think your statement somehow proves my point: if metrics are useful as compensation incentive, how can I be compensated on something I don’t control? How can I be responsible for something I don’t have any authority on? And what is that control/authority if not the capacity to act in a manner to influence the course of those metrics?
I was implying operational connection, and I also do with KPI when I identify which drivers, or operational KPI (if one can use such expression), have direct impact on the strategic, or high-level, indicators.
So, here I guess we should discuss what insights are. To keep this short, I would offer that insights are pieces of knowledge about how facts evolved, what their causes are, and which buttons I should press in order to influence the future course of those facts. Here, let’s play metaphors, metrics are the buttons. They help me build those insights, tell me what factors were most important, and which actions I should undertake.