Never let a vendor grade its own homework

A vendor dashboard can be accurate and still answer the wrong question.

The CRM reports sold leads based on its statuses and matching logic. The advertising platform reports conversions based on its attribution window. The call vendor reports appointments based on its classifier. Each view may be internally consistent. None should be the final operating truth.

Start with the event that settles the argument

For a delivered vehicle, the DMS normally settles the transaction. For a paid repair order, the service ledger settles it. For an answered call, the telephony record may settle it. The source depends on the outcome, but it must be named before the dashboard is built.

The FTC describes the DMS as managing accounting, inventory, service and vehicle financing across the dealership. That is why it belongs at the center of commercial reconciliation even when another system deserves credit for creating the opportunity.

Write the metric contract

A metric contract is a plain-language definition with enough detail to reproduce the number. For every important measure, write:

• the business event being counted;
• the source system and date field;
• included and excluded transaction types;
• identity matching rules;
• the attribution window;
• how cancellations, unwinds and duplicates are handled;
• who owns changes to the definition.

"Sales from CRM leads" is not a metric contract. "Retail and lease deliveries by delivery date, excluding wholesale and unwinds, matched to the earliest qualifying lead within 120 days" is close enough to test.

Reconcile before you optimize

Put the store's outcome next to the vendor's claim at record level. Sample mismatches. Most gaps fall into a few buckets: date windows, duplicates, status errors, unmatched identity, excluded transaction types or post-sale changes.

Do not turn the first mismatch into a vendor accusation. The point is to understand the rule. Sometimes the vendor found a real outcome the store missed. Sometimes the vendor definition is too generous. Either way, the group learns something only if both records can be inspected.

Build one scorecard for every provider

A durable vendor scorecard should separate:

1. Activity - messages, calls, impressions, tasks.
2. Store response - worked records, appointments, show, manager action.
3. Business outcome - delivery, repair order, gross category or another agreed result.
4. Incrementality - what likely would not have happened without the program.
5. Data quality - unmatched records, late feeds and disputed outcomes.

The first three can be automated. Incrementality often needs a test design, a holdout or at minimum restraint in the claim.

Independent measurement does not make the vendor the enemy. It makes the relationship better. A provider that performs should want the store's real outcomes. A provider that does not should not control the report card.

Let each dashboard keep its job

The independent scorecard does not replace every vendor dashboard. A platform's own view is often the fastest place to troubleshoot delivery, creative, routing or user behavior. Keep it for operating the platform.

Use the warehouse view for comparison and commercial decisions. That separation removes an impossible demand: no vendor has to reproduce the entire dealership ledger, and no dealership has to accept a platform-specific metric as the final business outcome.