Dealer Groups Should Manage Exceptions, Not Averages

Dealer groups have more data than a single rooftop and more ways to misunderstand it. A group average can make a healthy store look ordinary, a struggling store look acceptable, and a local decision look wrong because the comparison ignored context.

The group office's job is getting harder in both directions at once. The Haig Report counted dealerships acquired up 14.3 percent in the first half of 2026 while the number of transactions rose only 3.2 percent: groups are buying more stores per deal. In the same quarter the average dealership earned about $1.1 million, down 3 percent on the year, and Cox Automotive's dealer sentiment index sat at 41, below neutral, with a third of dealers naming expenses as a drag on the business. More rooftops to watch, thinner margin at each, and a regional director who cannot be in nine stores on Tuesday morning. That is the case for managing exceptions rather than averages, and the regional director owns it.

The general manager (GM) at each rooftop operates inside a particular market, franchise, inventory mix, staffing model and customer base. Group leadership needs a consistent view across stores, but consistency should not mean pretending those conditions are identical. The purpose of benchmarking is to find questions worth asking, not to issue a verdict from a leaderboard. Instead of ranking every store on every measure, the group identifies material exceptions, supplies comparable evidence, asks the local manager for context, and records the decision. Over time, those decisions become a learning system for the organization.

Two rooftops, one average, two different problems

The figure above puts two rooftops from a fictional nine-store group side by side. Rooftop A is an import franchise in a metro market; Rooftop B sells domestic trucks in a rural one. On aged used inventory the group average would flag both: 19 and 21 percent of used stock over 60 days against a group figure of 14. On declined-service follow-up it would flag neither, because 61 and 38 percent average out close enough to the group's 55.

The context column tells the real story. Rooftop A's aged share is a recon backlog after a technician left, and its pricing drifted during the backlog; the decision is about the shop, not the lot. Rooftop B's aged share is a deliberate position in work trucks with a 90-day selling cycle, and the GM has said so before; the decision is to leave it alone. Rooftop B's actual exception is the 38 percent follow-up rate, because one advisor has covered two lanes since September, and the 24 percent of internet leads getting no reply within a day, because after 6 p.m. rural leads land in a shared inbox. Neither of those shows up in a ranking of aged inventory. Both cost the store customers that quarter.

An average is a summary, not an explanation

Suppose one store has a higher share of aged inventory than the group average. That may indicate weak acquisition discipline, slow recon, pricing drift, poor merchandising, or an intentional position in a segment with longer selling cycles. The average reveals a difference. It does not explain the cause.

The same caution applies to price-to-market position, front-end gross, service labor performance, appointment show rate or negative-gross deals. Each measure can be useful, but only when the comparison set accounts for the conditions that materially affect it. A responsible group benchmark makes the comparison logic visible:

• Peer context. Compare stores or departments with relevant similarities, such as franchise, market type, volume band, inventory mix or operating model.
• Time context. Separate a persistent pattern from a short disruption caused by weather, construction, staffing, allocation or a local event.
• Process context. Identify whether stores use the same definitions, source systems and cutoff times before comparing the output.
• Decision context. Preserve prior manager actions, approved exceptions and known constraints that explain why a measure moved.

If the group cannot explain why two stores belong in the same comparison, it should not treat the difference as proof of performance.

Surface exceptions with the evidence attached

Group visibility becomes actionable when it narrows attention to a manageable set of exceptions. A regional leader should be able to see which stores moved outside an agreed band, why the movement was flagged, and which source records support the signal. The local GM should see the same evidence. That shared view prevents the group office and the rooftop from spending the first conversation debating whose report is correct.

An exception record should include the measure, comparison cohort, time window, contributing records, prior trend, named owner and response status. It should also provide a place for the local manager to add context. A store may accept the recommendation, choose a different action, request more evidence, or document why no change is warranted. The response matters as much as the flag. Without it, group reporting becomes observation without accountability. With it, leadership can distinguish an unreviewed problem from a reviewed exception with a sound local rationale, which is exactly the difference between Rooftop A's aged share and Rooftop B's.

Keep authority close to the decision

Cross-rooftop systems can easily drift toward remote control. A centralized team sees a deviation and sends an instruction without understanding the store's commitments, people or local market. That approach weakens adoption and can erase useful judgment.

The better model separates standards from decisions. Group leadership defines shared measures, materiality thresholds, evidence requirements, escalation rules and review cadence. Local managers retain authority to interpret the exception and select the response within those guardrails. This balance creates accountability in both directions. The rooftop cannot ignore a material exception without a response. The group cannot impose a conclusion without showing the comparison and source evidence. Each decision leaves a record that can be reviewed later.

Turn variation into organizational learning

Dealer groups are natural learning environments because similar decisions occur across multiple rooftops. The opportunity is not to copy the top-ranked store blindly. It is to identify a pattern, understand the operating practice behind it, test whether that practice fits another context, and observe what followed.

Artificial intelligence can help find unusual movement, assemble comparable cases, summarize manager responses, and retrieve prior decisions with similar conditions. The useful output is a better question or a prepared review, not a universal directive. When one rooftop handles an exception well, the group should capture more than the final metric. What did the manager notice? Which evidence changed the decision? What action was assigned? How quickly was it completed? What local condition made the approach suitable? Those details form a transferable playbook.

The result is a group operating model that respects rooftop expertise while making useful variation visible. The group manages the exceptions that deserve attention. Local leaders keep the judgment required to run the store. The organization learns from both the action and the reason behind it.

Questions for the next group operating review

• Which group averages are being treated as conclusions even though the comparison lacks store, market or process context?
• Can each rooftop see the evidence behind a group exception and record its decision in the same operating record?
• When one store finds a better practice, does the group capture the conditions and actions that made it transferable?