Lead and lag measures
You know I’m all about metrics and measures. One of the things that made entering the CX profession so attractive to me was that this is a field of study that’s not only based in numbers, it’s starving for people who have an affinity for measuring. CX is a study that’s founded on measuring…from survey results to Customer habits and attitudes to top-level improvements in your CX KPIs, numbers are all around us.
One topic of confusion I’ve seen a lot over my time is in regard to lead versus lag measures. Everybody’s got their own opinions and there doesn’t seem to be a textbook answer to what’s what, so take this as simply my theory and how I approach what’s meant when we speak of such things.
Lag measures are just another term for KPIs or output measures. These are those top-line, on-the-dashboard, am-I-getting-fired-or-promoted figures. Think revenues or market share, sales, etc. In the CX world, we’ll discuss things like Net Promoter System (NPS) score, Customer satisfaction (C-SAT), and perhaps Customer Effort Score (CES). These are the numbers that, at least within a discipline, are the end-goals; what we’re shooting for (even if they’re not necessarily considered the Über alles metrics for your larger organization). Usually they come straight from your Customers. I think this concept is the most straightforward.
Complementarily, in my opinion, a lead measure requires three characteristics:
One, it has to be up-stream from the lag measure. Simply put, it can’t be the end-all top-line metric. Sure, you can have goals for your lead measures, and definitely you want to see them go in a particular direction, but it’s not likely the stuff you’ll show to your CEO or shareholders, other than in the context of, and supporting a deeper-dive into, those actual lag, or output measures and KPIs. Again, this is simply definitional: of course, lead measures come first, which is to say, before the lag measures. But there’s more to it:
Secondly, these lead measures have to have some reliably predictive characteristic. They needn’t be perfectly correlated (and of course, we have to avoid the conflation of correlation with causation), but they really don’t serve much good in their purpose of leading if we can’t look at a set of lead measures and get some reliable idea of what our lag measure will look like. It’s even better if such lead measures can be seen to move temporally before the lag measure does (this is where the terms ‘lead’ and ‘lag’ come from anyway), which is to say, there’s still an opportunity to course-correct when necessary before bad lead measures turn into bad lag measures. Which leads us to the third, most controversial — for some reason — aspect:
Finally, a true lead metric is one that we can directly impact. You have to be able to explicitly make your lead measure something, not that its level is something that comes as a result of something else (that ‘something else’ is the actual lead measure). I have seen many clients (and bosses, in fact!) completely vexed when I challenge their alleged “lead” measures by inquiring of the measure, “so what is that set to?” The response is often an output: “Well, we’re seeing 84% on that survey result”…as if a survey result is a leading measure. But it lacks this specific and vital quality of being controllable.
Take this example: I have a programmable thermostat in my house. My lead measure is where I deliberately set the thermostat’s gauge (I like it to be at 70°F in the summer). The lag measure is my comfort level (I’m too hot or too cold). The actual temperature isn’t a lag measure, because it’s not my goal (which is to feel comfortable), but rather an aspect of my comfort level. The overall output of my HVAC system is how comfortable I feel in the environment of my house. The temperature (an arbitrarily man-made number affixed to a physical condition) will surely impact that comfort, but it isn’t, in-and-of itself, what or how I feel. Nor is the temperature in my house something I can directly set (it’s a result of my furnace or A/C interacting with the ambient air). So, lead is where I program my thermostat, lag is how I feel. The temperature of the room is…something else, something in between. It has a predictive quality (if it’s 95°F, I’m not going to be very comfortable), but it’s not something I can simply make happen on my own.
Closer to home (or rather, farther from home, but closer to work): In your contact center, your lead measure isn’t average hold time, because you can’t control it directly (as with a knob). The lead instead is, say, how many agents you have working on the line at any one point. While the average hold time can definitely be an indicator and a predictor of your Customers’ experience and satisfaction (whether measured in NPS, C-SAT, or whatever…your KPI, or lag measure), and precedes it, it’s not truly a lead measure in my book because you can’t directly affect it. You can hire more people (well, maybe you can) or shift schedules around to put more people in the contact center during certain times (and you can measure that you’ve been successful in doing so by counting heads), but you can’t simply dictate that the average hold times simply be a certain value (and let’s not get into the variation in that average) by adjusting a setting to “30 seconds each.” In fact, that average is impacted by other variables as well (many, like call volume, that you have no control over either). That’s what keeps it from being a true lead measure.
So this middle space, these predictor or indicator measures are what trip people up…and sometimes cause frustration because people are trying to control them as if they’re knobs you can spin around and set to a particular level. Mistaking indicators for lead measures is like saying, “I don’t know why I’m so hot in here; I just this second set the thermostat down to 65°F.” Sure, but that setting was just your lead indicator. Factors such as how warm it currently is, the size of the room, the capacity of your air conditioner, etc., will have impacts on how long it’ll take you to feel more comfortable. Some of these other factors are true lead indicators (you can buy a bigger A/C unit, for example), and some are exogeneous indicators that are baked into the cake (if it’s warm enough, ha ha).
You can dismiss this as all just quibbles about how we characterize or categorize different types of measures (“I just want my C-SAT to be 90!”), but if we don’t take the time to develop shared definitions of terms, we’ll likely not get very far and instead spend our time discussing things that are of no impact on getting to where we need to go. You can argue now and come to an agreement about which measures we have control over and which are interim or output results of the system; or you can spin your wheels with your analysts afterwards wondering why you can’t get your lag measures to move the way you want.
(Originally Published 20201015)
- LtCol Nicholas Zeisler, CCXP, LSSBB, CSM
- Principal, Zeisler Consulting