Optimize the Work Before You Optimize the People

When a service department misses its numbers, the fastest explanation is often a people explanation. Technicians need to be more efficient. Advisors need to sell better. Dispatch needs to move faster.

Sometimes the explanation is right. Often it skips the operating conditions that shaped the result. A technician cannot flag hours on a vehicle that is waiting for authorization. An advisor cannot give a confident update without a current diagnosis and parts estimate. A dispatcher cannot make a sound assignment when the skill requirement is buried in a note. Everyone can look busy while the work itself sits still.

Before asking people for more output, a service director should ask a simpler question: did the system put the right work, information and decision in front of the right person at the right time?

The store cannot hire its way out of this

The labor market makes the question urgent. NADA Data 2025 puts the average franchised dealership at 16 technicians, a quarter of its employees, handling 16,252 repair orders a year, which works out to roughly a thousand orders per technician. The Bureau of Labor Statistics projects about 66,200 openings a year for automotive service technicians and mechanics through 2035, most of them to replace people who leave the occupation, at a median pay of $50,620. The TechForce Foundation counts 241,842 annual technician openings across all transportation sectors.

Every store is recruiting from the same thin pool, at a labor rate NADA puts at $186 an hour for customer-pay mechanical work. An hour a certified technician spends waiting for a part, an approval or an answer is an hour the store sold at $186 and paid for twice. Protecting that hour is the cheapest capacity a service director will ever buy, and it does not require a job posting.

Labor is constrained by more than headcount

Technician availability is easy to count. Productive capacity is harder. It changes with job mix, certifications, diagnostic uncertainty, parts, equipment, interruptions, and the amount of ready work in the queue.

The figure above shows why mix matters, using a fictional store's month. More than half of its repair orders carry under an hour of flagged time; one in eight carries four hours or more. The small orders set the rhythm of the lane and consume advisor and dispatch attention out of proportion to their hours. The heavy orders are where a wrong assignment, a missing part or a slow authorization costs a full day of a scarce technician's capacity. A shop that dispatches both kinds the same way is optimizing neither.

Two departments with the same number of technicians can produce very different outcomes because one protects ready-to-work time and the other lets it dissolve into waiting. The second store may respond by tightening efficiency targets, even though the constraint sits upstream.

The same is true for advisors. An advisor's performance depends on the quality and timing of the information arriving from technicians, parts, warranty and prior service history. If those inputs are incomplete, the advisor spends the day reconstructing the truth while customers wait. That is why workforce improvement should begin with the work design around the workforce.

Dispatch should make tradeoffs visible

Good dispatch is not simply giving the next repair order (RO) to the next available technician. It is a series of tradeoffs: skill fit, promised time, diagnostic risk, bay constraints, parts readiness, carryover exposure, and the need to maintain a healthy mix of work.

AI can help assemble that context and identify conflicts. It can flag that a job requires a certification the assigned technician does not have, that a waiting customer is at risk, or that a highly skilled diagnostic technician is being consumed by work another technician could perform.

The recommendation still needs human judgment. A dispatcher may know that a technician is training on a new operation or that a customer commitment just changed. A useful system exposes its reasoning so the dispatcher can approve, change or reject the recommendation and record why. Those overrides are not failures. They are operating knowledge. When captured, they reveal rules the formal process may have missed.

Give advisors a complete story, not another alert

Advisor support should work the same way. A complete, defensible customer conversation matters more than another prompt.

Before an advisor presents work, the relevant story may include vehicle history, current mileage, previous declines, technician findings, photos, warranty status, parts availability and the customer's stated needs. When those facts are spread across screens and notes, the advisor either delays the conversation or fills the gaps manually.

A prepared view can reduce that friction. It can show what is recommended, the evidence behind it, what is urgent, what can wait, and what commitment the department can realistically keep. The advisor remains responsible for the conversation. The system makes it easier to be accurate and consistent.

That distinction also protects trust. A recommendation should not appear because a model says the customer is likely to buy. It should appear because there is a documented vehicle need and a clear explanation the advisor can stand behind.

Coach from exceptions, not surveillance

Analytics can help managers coach, but only if the purpose and evidence are clear. Constant individual scoring creates noise and encourages people to manage the score. A better approach starts with operational exceptions.

Look for work that repeatedly waits after diagnosis, estimates that require avoidable rework, customer updates that occur after the promised time, or jobs routed away from the technicians best prepared to complete them. Then review the workflow and the decision, not only the person at the end of the chain.

The most useful coaching question is often not why someone was slow. It is what information or decision they lacked when the work stopped. Some patterns will point to training. Others will expose a policy, system or handoff problem. Treating both as individual performance issues prevents the department from learning, and in this labor market it sends good technicians to the shop across town.

Measure friction the team can remove

Traditional productivity measures still matter, but they need operating context. Add measures that show how much of the day was actually available for productive work:

• Time an approved job waited for assignment
• Time a technician waited for parts, authorization or clarification
• Estimate rework caused by missing information
• Promised-time changes made before versus after a customer had to ask
• Dispatch overrides and the reasons behind them
• Exceptions resolved within the department's review window

These measures lead to better actions because the team can influence them directly. They also make improvement visible. If a new parts check or dispatch rule is followed by less waiting, leadership can compare the periods and assess whether the relationship holds while accounting for other changes.

People remain the dealership's advantage. Better operating intelligence should remove avoidable searching, waiting and rework so their judgment is applied where it has the most value. It should never treat them as interchangeable units.

Questions for the next fixed-ops review

• Where does ready work stop because context, approval, parts or ownership is missing?
• Which performance problems are truly skill gaps, and which are workflow gaps appearing at the employee level?
• When a manager overrides a recommendation or dispatch decision, is the reason captured so the process can learn?