Metrics are always the wrong place to start

HR has an unhealthy appetite for metrics. It worries about which ones it should be using, or how they should be defined or how they are going to be produced.  Somehow it feels that only if it owned the golden-metric-key then a wonderful world of corporate credibility and importance could be unlocked.

Business leaders never really need metrics.  What they do need is to make decisions to improve their businesses (and further their own careers). Decisions only need to be made if there’s uncertainty and the right information can help reduce that uncertainty resulting in better decisions.  We can put a cost on the uncertainty, and we can put a value on the information that creates better decisions.  There’s the universal business case for better information.  Simple?

In business, anything of value can be observed, either directly or indirectly.  If something can be observed it can be measured.  HR and employees are no different, but might need different measurement techniques than some other areas.  Some forms of measurement might not be cost-effective but that is a secondary question.

So why does much of HR insist on avoiding why business leaders need to make decisions and insist on jumping straight into choosing some pre-packed metrics?  Without understanding the problem how do we identify which information is needed and the value of that information against other forms?

What is a metric?

Metrics (and the related KPIs) are just simplified ways of presenting information.  They’re created using some form of calculation from observations.  There are some universal truths for metrics:

  • As numbers, on their own, they are abstract.  Their worth in isolation is close to zero
  • To create a metric you always need to throw away information. That information had a value
  • Much of the time they’re some sort of mean.  We often use means because they’re easy to calculate
  • The variability of the metric is much more important than the level
  • Insight usually comes from spotting patterns between multiple metrics in combination
  • Just by measuring and reporting metrics you change employees’ behaviours.

Metrics are just the ingredients you need to create insight, to reduce the uncertainty that leads to inefficient decisions.  A good chef combines and prepares ingredients to create a dish much more enjoyable than the base ingredients.  A good analyst does the same with metrics.  The current buzzword for the process they use is ‘analytics’.

Decisions involve trade-offs

All decisions involve trade-offs and the choice involves what combination of the options is possible and the preferences of the decision maker.  The sponsor of a recruitment function needs to decide on the optimal combination of cost and service.   They can always reduce or increase cost or service but are they better off?  They need to understand the various alternatives they can choose between.

If we only present the total cost, or the cost per hire, or, as I prefer, the marginal cost of hiring, the decision maker can’t make the optimal decision.  They need to understand the level of service, both in terms of absolute levels (quality of hire, time taken) but also perceived quality.  Presenting just the cost metrics might indeed lead to worse decisions than no quantitative information at all.

Furthermore, if benchmarking their cost metrics with others they need to also to understand their comparators service level.  Someone with lower cost might be operating on a lower frontier curve and therefore be less effective.

Technology can’t be the full solution

If we look to technology to solve our problems in this area we are setting ourselves up to fail.  Technology can make the analytic process easier, it can improve the reliability of results, it can communicate to a diverse group quickly, but decision making is inherently human.  This is not saying there is no use in analytic technology but instead it should be seen as a tool to enable us to meet our aims.  As a function we’re not short on technology, we’re short on expertise.

If we think to our recruitment function optimisation decision we need data on hiring performance which is likely to be in our ATS, there will be cost data, probably from multiple sources, there will be new-hire performance data from other HR systems, there needs to be perception data from key stakeholders.  This list can probably be extended. Can we really expect the reporting functions of any one of the systems to be able to give us a comprehensive answer?

Always start with the decision to be made, and how it is made

The one thing I look for in a good analyst is the ability to really understand the problem.  This can’t be done just by asking the decision maker; we often need to use other techniques (There’s an iphone app from IDEO to give you some ideas here.)  This, the most important part in the whole process can’t be done by technology.

Only with this understanding can we start to explore what information would help drive the decision, where that information is (or whether we need new measurement), how to transform it to new, more valuable variables and which analytic techniques to use.  Which combination of metrics to use comes quite far down the process and is usually obvious if you’ve done the preceding activities.

Finally we need to understand how to communicate information.  We need to understand how our reports are perceived and interpreted, how the decision maker uses them, where they use them, why they looked at them and what they did after viewing.  By using usability techniques to study users doing this this my work and thinking is being driven in new directions.  How we use information in the future will be very different from today.  Metrics will still be there but their importance is certainly on a downward curve.