The gap nobody schedules

Most dealerships are decent at training day and terrible at day 12.

You bring in a trainer, or you run a Saturday session yourself, and the floor learns a new way to open the phone. Everyone nods. Two reps even use it that afternoon. Then a heavy weekend hits, a manager takes vacation, the trainer's follow-up webinar gets skipped, and by the end of the month the new opening exists only in the notebook it was written in.

Then somebody buys another tool to fix it.

The loop below is the thing that goes between training day and the next purchase. It has four parts — observe, correct, verify, then decide. It runs on a calendar, not on willpower. And its most important output is a yes/no answer to one question: did this behavior actually get adopted, or did we just talk about it?

Because until you can answer that, you have no business adding another tool. You'll be layering a new system on top of an unmeasured one and calling it progress.

Step 0: Turn the training into one observable behavior

You cannot reinforce "better rapport." You can reinforce a sentence.

Before the loop starts, write the behavior down in the form a manager can hear on a recording in under five seconds. Compare:

  • Weak: "Reps should build urgency."
  • Observable: "Before the call ends, the rep names a specific vehicle in stock and a specific time window: 'I've got the Silverado with the tow package sitting here — can you be here today at 5:30 or tomorrow at 10?'"

Same for a service-drive upsell, an insurance cross-sell, or an inbound internet lead. One behavior. One sentence pattern. One place in the call where it belongs.

If your dealership sales training produced fourteen new behaviors, pick one for this cycle and put the other thirteen in a queue. A floor can absorb one change at a time. Two changes means neither gets measured, and unmeasured changes are how process drift starts.

Step 1: Observe — 10 calls per rep, scored the same way

Reinforcement without observation is just repeating yourself louder.

Set a fixed sample. Ten calls per rep, per cycle, pulled from real inbound and outbound activity — not the ones the rep hands you. If you're pulling manually, take two per day across a week so you don't accidentally sample only Monday energy.

Score each call on exactly two things:

  1. Did the behavior appear? Yes / no. Not "sort of."
  2. Was it in the right place? A trade-in question asked after the customer already gave a walk-away date doesn't count.

That's it. Do not turn this into a 22-point rubric this cycle. You are measuring one change, and a narrow scorecard is what makes the number believable when you show it to the rep.

Write the tally somewhere visible to managers:

Marcus — 7 of 10 Dana — 2 of 10 Priya — 9 of 10 Sam — 4 of 10, but 3 of the misses were on service transfers, not sales calls

That last note matters more than the numbers. It's the beginning of a diagnosis.

Step 2: Correct — separate the four kinds of miss

A 4 of 10 is not one problem. Reps skip steps for different reasons, and each reason needs a different correction. Sorting the misses is the actual work of training reinforcement.

Didn't know it applied here. Sam used the new close on sales calls and dropped it on transfers because nobody said transfers counted. Correction: clarify scope in one sentence and re-sample next week. This is a manager error, not a rep error.

Knows it, can't say it under pressure. The rep starts the sentence, the customer talks over them, and they abandon it. Correction: two minutes of live reps at the desk. You play the customer, you interrupt them on purpose, they finish the sentence anyway. Do it three times.

Tried it, got burned, quietly stopped. This is the sneaky one. A rep asked for a same-day appointment, got "wow, pushy," and never asked again. Correction: pull that specific call, listen to it together, and find the actual failure point — usually the ask landed before the rep gave the customer a reason to come in. Rewrite it with them.

Doesn't buy it. Rare, but real. The veteran who thinks the new opening sounds like a script. Correction: a direct, unpleasant conversation, plus a measurement bet — "Run it exactly as written for 20 calls. If your set rate doesn't move, I'll drop it for you."

Notice that only one of those four is a compliance conversation. Managers who treat every miss as attitude burn trust fast and get better at hiding, not selling.

Step 3: Verify — the second sample is the whole point

Here's where most reinforcement programs quit. They observe, they coach, and then they move on to the next initiative and assume the coaching took.

Verification is a second sample, same size, same scoring, 7 to 14 days later. Nothing else. But you have to actually run it.

Three outcomes:

  • Adopted. 8+ of 10 across the floor, held for two consecutive samples. The behavior is now part of your process. Add it to your standard scorecard, stop sampling it separately, and take the next item out of the queue.
  • Partial. Two or three reps up, one or two flat. Individual coaching problem. The behavior is fine; those reps need reps.
  • Not adopted. Everyone flat or down. Something about the behavior itself is wrong — it's too long, it doesn't fit your inventory reality, or it contradicts something else you told them. Rewrite it before you blame anyone.

That last outcome saves you months. A behavior that nobody adopts after real coaching is usually a bad behavior, and finding that out in three weeks is cheap.

Step 4: Decide — the gate before you buy anything

Now the tool question.

When a vendor demo lands on your desk — a new dialer, a new lead router, a new AI thing — run it through this gate:

  1. Name the behavior it's supposed to change. If you can't state it as a sentence a rep would say, or a step in the CRM they'd take, the tool is a hope, not a fix.
  2. Show the last verified sample for that behavior. If you never measured it, you don't know whether you have a tool problem or a coaching problem. Measure first, one cycle.
  3. Confirm the last thing you bought got adopted. If your text-messaging platform sits at 30% rep usage, adding a second channel does not get you to 100%. It gets you two half-used systems and one more login for a new hire to ignore.
  4. Name who owns the rollout and what number they'll report in 30 days. No owner, no purchase. Sales manager accountability is not a poster; it's a name next to a number on a specific date.

Most dealerships that feel like they need a new tool actually need one behavior verified. The loop tells you which situation you're in.

What this looks like on a calendar

  • Monday, 20 min: managers pull and score the week's sample. Post the tally.
  • Tuesday, 15 min at the desk: each rep with a miss gets their category and one correction. Not a lecture — a category and a correction.
  • Following Monday: verification sample. Same 10, same scoring.
  • End of cycle, 10 min in the manager meeting: adopted, partial, or not adopted. Next behavior out of the queue, or same behavior for one more round.

Four touches. Under an hour of manager time per week. That's the whole system, and it's the difference between process consistency you can prove and process consistency you assume.

The unglamorous truth about drift is that it's not a discipline problem. It's a measurement problem. Behaviors you sample survive; behaviors you announce don't. If pulling and scoring the sample by hand is the part that keeps falling off your calendar, that's exactly the piece worth automating — so the observe step happens whether or not you had a good week.

Then, and only then, look at the next tool.