What changes about custom software when you run multiple locations?

The same word means different things per location

One branch calls it a maintenance agreement, another a service plan. One tracks a lead source as "Google," another splits paid and organic. One marks a job complete at the truck, another after invoicing. Sum those and you get a number that is arithmetically correct and operationally meaningless.

Nothing about this is anyone's fault. Locations grow through acquisition, inherit their own habits, and are usually right about their own operation. The mistake is assuming a rollup can happen without translation.

Build a mapping layer and give it an owner

The durable pattern is a canonical dictionary maintained centrally: the corporate definitions of job type, lead source, revenue category, and job completion. Underneath it sits a mapping table per location, so local naming survives and rollups still mean something.

The critical part is that this list is a living asset with a person responsible for it. New lead sources appear weekly. If nothing routes unmapped values to that person, they silently fall into "other" and the rollup slowly decays. Make the unmapped queue visible in the tool itself — that single feature keeps cross-location reporting honest over years.

Decide what is central and what is local

Standardization is a political question before it is a technical one, and the useful move is to be explicit about which category each thing falls into rather than fighting every case individually.

• Central by necessity. Metric definitions, the location list, the roles model, and anything appearing in a comparison.
• Local by right. Price books, scheduling rules, staffing, and the day-to-day process. Standardizing these by software mandate generates resentment and workarounds.
• Negotiated. Lead source taxonomy and job type. Locations need enough granularity to run; corporate needs enough consistency to compare. A two-level scheme usually resolves this.

Rollups need rules, not just addition

Averaging location averages is not a company average unless volumes are equal, and they never are. Weight by the underlying counts. Decide how a location that opened in March is treated in a year-over-year view. Decide whether a customer served by two locations is one customer or two.

Write those rules down where readers can see them, because in multi-location operations a number is only useful if two general managers can agree on how it was produced. That is the whole reason our briefs are written per level — ownership, region, location — instead of sending one report to everyone and letting each reader interpret it differently.