How to compare two org structures side by side

Comparing two org structures means answering four questions about each option against the other: how many positions it has (and how many are filled), what it costs, which positions it adds, removes or changes, and who ends up reporting to whom. Get those four answers on one screen, from the same starting data, and most reorg debates get much shorter. Here is how to set the comparison up so the answers are honest.

Decide what you are comparing

"Compare org structures" can mean two different jobs. One is comparing structure types: functional against divisional, a flat org against a layered one, a matrix against a single line of command. That's a design question, and the textbooks are right that no type wins everywhere. The other job, the one a reorg actually needs, is comparing two concrete options for your own team: "Platform team" against "Two divisions", or either of them against today. This post is about the second job. Choosing a type is only useful once you can see what it does to your positions, reporting lines and budget.

Be clear which pair you're comparing, too. Each option against the current organization tells you what it would take to get there. The two options against each other tell you what you give up by picking one. You usually want both readings.

Build both options from the same starting point

A comparison is only as fair as its inputs. If one option was sketched from last quarter's headcount and the other from an HR export, every difference you find is partly noise. Start both from one faithful picture of today: every team, every position (filled or open) and every reporting line.

In Resorcerer each option is a model: a full copy of the organization, with its own units, positions, reporting lines and assignments, that you can rearrange without touching any other model. Copy the real organization once per option. Copy it with assignments when the question is "next quarter, starting from where we are", or without assignments when the shape is the question and the names aren't yet. A copy remembers where it came from, so its comparison defaults to the model it was copied from.

Keep the options private while they name people. A private model is visible only to organization admins and the people you grant access to.

Price every position by the same rules

The fastest way to rig a comparison, by accident or not, is to cost the two options differently. One option gets the real salary of each named person, the other gets a round number per new hire, and the cheaper option is simply the one with the friendlier assumptions.

Price both from one set of rules instead. A cost profile gives a price to a shape of position (an Engineer, at Senior, in London) and every matching position is costed automatically, with the most specific rule winning when several match. Cost profiles live at the organization level and are shared by every model, so both options are priced by exactly the same rules, and the cost follows as you keep editing.

Two checks before you trust a total:

• Unpriced positions. A position no cost profile covers contributes nothing, so the total is understated. The model's Summary reports these as unmatched; get the count to zero in both options.
• Filled against vacant. The Summary splits cost into filled and vacant. An option that looks cheap because half its new positions are open will cost the full amount once they're hired.

Compare the totals first

Open one option's Compare view and pick the other model. The top of the comparison is the number the conversation usually turns on: total positions, filled positions and total cost for each model, and the delta between them.

Read those first, against the goal you wrote down for the reorg (see How to plan a reorg for setting one). If an option misses the budget by a wide margin, the detail below won't save it. If two options land within a rounding error of each other, the decision is about structure, not money, and you know where to spend the meeting.

Then read the differences, position by position

Totals hide trade-offs. Two options can add the same number of positions and cost the same, yet one moves a dozen people and the other moves three. Below the totals, Resorcerer classifies every difference as added, removed, changed or unchanged, for units and positions alike. Filter to one change type, or to one part of the tree, to answer questions like "what happens to Engineering?". Open a changed position and you see the before and after of each attribute that differs: role, level, location, company, occupant, units and reporting line.

That last one is where structure comparisons usually go wrong. A new reporting line is a real change for the person on the end of it, even when their title, level and pay stay put. Reading the reporting-line changes per manager is the quickest way to see who gains a team, who loses one, and who ends up with a much wider or narrower span than before.

Know how positions are matched

To say a position changed rather than was removed and another added, a comparison has to recognize the same position in both structures. Resorcerer matches on the strongest signal available, in this order: the same person assigned (high confidence), the same attributes and the same reporting line (high), the same attributes and the same unit (medium), and the same attributes alone (low). Anything left unmatched shows as an addition or a removal.

This matters when you compare two heavily rewritten options. The more you change at once, the less certain the matching gets, so a comparison that looks stranger than expected is usually two big rewrites being matched on weak signals. Copying each option from the model you mean to compare it against keeps the matching on its strongest footing. The table is in Compare and merge.

Look past the totals: breakdowns, charts and problems

Each option is a full model, so every analysis view works on it. Open the same view on both options and you get the comparison at a level the totals can't:

• Counts and Costs break headcount and spend down by role, role category, level, location, company or vacancy against the unit hierarchy. Put the same breakdown up for both options to see where each one concentrates people and money.
• Org Chart, Diagram and Matrix show the reporting structure, the shape of the units, and which teams hold which roles. Flicking between the two org charts is still the fastest read of layers.
• Problems reviews each model for a person in more than one position, a person whose attributes differ from their position's, a reporting cycle, and a position that belongs to no unit. An option with a reporting cycle in it isn't finished, however good its totals look.
• Slots lists every position as a table you can filter and export to CSV, for anyone who wants to check the numbers their own way.

The views are described in Analyse a model.

Compare over time, not just at one moment

Few reorgs land in one day, so a single snapshot can flatter an option that front-loads its hires. Give each option's positions start and end dates, or lay its phases out as dated points on a timeline, and the Counts and Costs views plot headcount and cost as a trend. Two options with the same end state can look very different in March.

The same Compare view also works backwards. Pick one of a model's own snapshots as the other side and you see exactly what has changed since it was taken, which is how you check the reorg you shipped against the one you approved.

Choose one, and keep the comparison

When an option wins, the comparison you used to choose it becomes the change list. Merge the winning model into the real organization and review each change on its way in: accept the new team, reject the hire that isn't funded yet. The losing option stays as it was, so if the decision is revisited next quarter, the alternative is still fully worked out rather than a half-remembered slide. That whole loop, from copy to merge, is what org scenario planning in Resorcerer is built around.