When generic software costs more than bespoke

Build versus buy is usually framed as a technology decision. In a dealership, it is an operating-model decision. The real question is not whether a subscription is cheaper than a developer. It is whether the rented tool solves the whole job after data access, integration, duplicate work, adoption and contract terms are included.

Sometimes buying wins easily. Keep the vendor, use it harder and move on. Sometimes the subscription is only the first line on a much longer bill.

The invoice is not the cost

A useful tool can still create expensive seams. A lead enters through one system, gets worked in another, is delivered in the DMS and is reported back by the original vendor. If those systems disagree, somebody in the store becomes the integration layer.

The Federal Trade Commission described the DMS as software that manages nearly every aspect of a dealership, including accounting, payroll, inventory, service, financing and manufacturer data flow. In the FTC's challenge to CDK's proposed Auto/Mate acquisition, the agency also called out integration fees, long contracts and switching difficulty as competitive issues in the DMS market.

That is the arithmetic most build-versus-buy spreadsheets miss. Add these columns:

• subscription and per-rooftop fees;
• implementation and third-party access fees;
• employee time spent re-keying or reconciling;
• the cost of data that cannot leave the vendor;
• the cost of changing the store's process to fit the software;
• the exit cost if the tool stops earning its keep.

Buying is right when the problem is common

Do not build payroll. Do not build a credit bureau. Do not rebuild a mature commodity because ownership sounds good in a meeting.

Buy when the workflow is broadly standard, the vendor gives the store usable data, the integration is clean, the team adopts it and the contract leaves room to leave. A good vendor is leverage. The enemy is unaccountable spend, not vendors.

Building is right when the difference is the point

Build when the workflow expresses how the group actually operates. Examples include one group definition of a delivered unit, an inventory signal based on its own stocking rules, a service-lane action tied to its owner base, or an audience that must combine DMS, CRM and website behavior.

NADA's own data guidance now emphasizes dealer control, limited sharing and visibility into who can access dealership systems. That is also a useful commercial test. If the store cannot retrieve the data, define the metric or leave with the work, it is renting more than software.

A five-question renewal test

1. What exact job does this tool finish? Name the output, not the category.
2. What manual work remains? Ask the person doing it, not the person who signed.
3. Does the useful data come back? A CSV after cancellation is not an operating feed.
4. Can our definitions win? The store should define a sale, a shown appointment and an attributable outcome.
5. What do we keep? If the relationship ends, identify the code, data, configuration and operating knowledge that remain.

Generic software is not suddenly obsolete. It is simply no longer the automatic answer. The cost of building fell. The cost of coordination did not. Put both on the same page before the next renewal.