Fixing headcount rollforwards

A rollforward is a common way of presenting the change in any quantity from one period to another. Headcount changes from one period to another are often presented like this:



The columns are the rollforward itself. The rows indicate either points in time (e.g. January 1 or December 31) or activities (hire or remove people).

Presented like this, headcount rollforwards work pretty well. Depending on how you're measuring headcount, however, they quickly run into issues.



The question I always get from the businesses goes something like the following:

If I rehire someone into the same position that was vacated, do I put it in as an add or a negative removal?

It's not immediately clear. Also note the fact that two very interesting "leading" indicators of Heads (i.e. the additions of role A and role B) appear nowhere in the walk. That's a lot of interesting data being lost.

The root of this confusion is the walk itself, which tries to measure two abstract concepts with one walk:

• Roles
• Heads

These are related but not identical concepts. In general, walks should only measure the changing quantity in one abstract concept (Roles or Heads). That said, you can get creative about how you combine the two concepts in a way that's both consistent with this principle and informative.

Take a look at this example:



Now there's no confusion about how transactions are coded.

• Any time a Role is added or removed, it results in a change in the "Roles" metric (i.e. how big your org chart is). There is an offsetting move in the "Open Roles" metric, resulting in no net change in Heads.
• Any time a Head fills or is removed from a Role, the "Open Roles" metric changes. No need to change the Roles walk.

This scheme has a few benefits:

• It maps more closely to the underlying data model (which usually has one table for Roles and another for Heads). 
• It lines up more closely with how operational people think about their organizations. If you talk to ops people, they tend to think in terms of org chart boxes, not number of heads filling those boxes at any point in time.
• It's more clear how each operating activity (hiring/firing, adding/removing roles) impacts the walk.
• It allows you to get a leading indicator of your Head movements by capturing the changes in your org chart (i.e. Role count changes).
• It allows you to have interesting conversations with your HR staff about "Open Roles" (pipelines, aging, etc.).

The story gets a bit more complicated when you tackle multiple-division reporting (e.g. elimination of intracompany activity in the consolidated rollforward), but the underlying model is cleaner than the existing methodology of smashing Roles and Heads together into the same walk.