“What did it mean?”
“People who come in to look.”
“Why trash?”
“Bad joke.”
“Did customer appearance affect your decision about whether to approach?”
“Sometimes.”
“Why?”
“Efficiency.”
There.
Derek became more confident.
“Sales is triage.”
“You have six customers, you prioritize likely buyers.”
Rachel asked:
“Based on what evidence?”
“Experience.”
Walter almost smiled bitterly.
Again.
Derek continued:
“You can act shocked, but every luxury dealership does this.”
Walter spoke for first time.
“Does that make it right?”
Derek turned.
“No.”
“Then why say it?”
“Because you're treating me like I invented reality.”
Walter nodded.
Good point.
He had not.
Derek said:
“You built luxury dealership.”
“You want premium brand.”
“You reward high gross.”
“You rank us every month.”
“You put plaques on wall.”
“You pay for closing efficiency.”
“Then you're surprised people prioritize rich customers?”
Silence.
Michael looked toward Walter.
Derek leaned forward.
“I was rude to you.”
“Yes.”
“I shouldn't have been.”
“No.”
“But if you're going to fire me and call this fixed, you're lying.”
Walter stared at him.
There it was.
A man defending himself with truth.
Walter did not like him more.
But truth did not require likability.
Derek continued:
“You created scoreboard.”
“Michael enforced.”
“Brandon controlled leads.”
“Everybody learned.”
“I'm just stupid enough to say it out loud.”
Walter sat back.
“You are responsible for your behavior.”
“I know.”
“Do not confuse system responsibility with personal absolution.”
Derek nodded slowly.
“Fair.”
Walter looked toward the glass wall.
Top Producer plaques.
Commission board.
Premium-delivery photos.
Numbers everywhere.
Complaints in folders.
He had come expecting to evaluate customer service.
Now he was looking at a business model that taught employees which customers mattered most before any employee consciously chose cruelty.
Walter turned back.
“The review continues.”
Derek asked:
“Am I fired?”
“Not today.”
“You really don't like fast decisions.”
Walter looked at him.
“I spent fifty years watching people make fast decisions about me.”
“So no.”
“Not when evidence is still moving.”
Derek said nothing.
By sunset, Walter had learned enough to know the dealership had a problem.
He had not yet learned how large.
That answer arrived next morning when Rachel Monroe found an internal spreadsheet labeled:
LEAD QUALITY NOTES
One column was named:
Visual Probability.
And suddenly the jokes in the group chat no longer looked like jokes.
They looked like informal policy that someone had tried to turn into data.
The spreadsheet contained 1,842 customer entries.
Walter stared at the number.
“Who made this?”
Nobody answered immediately.
Rachel Monroe had discovered it on a shared sales drive.
File owner:
Brandon Cole.
Floor coordinator.
Modified repeatedly over eighteen months.
Columns:
Date.
Customer first name.
Vehicle interest.
Source.
Salesperson.
Financing likelihood.
Trade-in.
Follow-up status.
Reasonable.
Then:
Visual Probability.
Scores one through five.
Walter scrolled.
Five:
High-end clothing.
Luxury vehicle arrival.
Known business owner.
Referral from premium client.
Four:
Professional appearance.
Late-model vehicle.
Specific model knowledge.
Three:
Unknown.
Two:
Casual low-cost appearance.
Old vehicle.
Financing questions early.
One:
Work clothes.
“Unclear employment.”
No vehicle.
Young with companions.
Heavy price questioning.
Walter stopped.
Age.
Clothes.
Car.
Financing questions.
Not financial qualification.
Appearance.
“What was score used for?”
Michael looked sick.
“I did not know this existed.”
Rachel said:
“We need establish.”
Brandon was called.
He arrived twenty minutes later.
No attorney.
Rachel advised he could have one.
He declined.
Fine.
Walter did not speak at first.
Rachel showed spreadsheet.
Brandon’s face changed.
“That is not official.”
“Did you create it?”
“Yes.”
“Why?”
“Lead management.”
“Explain visual probability.”
“It wasn't literal probability.”
“Then why name it that?”
Brandon struggled.
“We needed way to prioritize walk-ins.”
“Based on appearance?”
“Not only.”
“Columns say otherwise.”
Brandon looked toward Walter.
“This sounds worse than it was.”
Walter replied:
“Most bad systems do when written down.”
Brandon flinched.
Rachel continued.
“Were salespeople instructed to use scores?”
“Not formally.”
“How were scores assigned?”
“I entered.”
“Based on salesperson notes?”
“Sometimes.”
“Did score affect lead assignment?”
Silence.
There.
“Yes.”
“How?”
“Higher-probability buyers got senior salespeople.”
“Lower got whoever was free.”
“Or?”
“Sometimes junior staff.”
“Sometimes no immediate assignment if floor busy.”
Walter felt anger rising.
“Meaning a one could wait while a five who arrived later got served.”
Brandon looked down.
“Yes.”
“How often?”
“I don't know.”
Rachel asked:
“Did Derek know?”
“Yes.”
“Michael?”
“No.”
Michael said:
“Absolutely not.”
Walter looked at him.
“Not knowing may become its own finding.”
Michael’s face hardened.
“I cannot know every spreadsheet.”
“No.”
“But you can know how leads are assigned.”
Silence.
Walter continued:
“You are general manager.”
“If an unofficial scoring system can determine customer attention for eighteen months without you noticing, that matters.”
Michael nodded reluctantly.
Good.
No defensiveness.
Mostly.
Rachel asked Brandon:
“Where did idea come from?”
Brandon hesitated.
“Another dealership.”
“Which?”
He gave name.
Former employer.
They used “purchase likelihood” based on engagement signals.
Brandon adapted.
“Did original use clothing?”
“No.”
“Vehicle arrival?”
“Sometimes.”
“Race?”
“No.”
“Did yours?”
“No.”
Walter examined rows.
No race field.
Good.
But proxies can still produce disparate treatment.
Need analysis.
External counsel recommended data review.
Walter approved.
Customer demographics incomplete.
Could not reliably assess protected-class impact from spreadsheet alone.
Important not to invent.
But appearance-based service discrimination existed regardless.
Maybe not illegal in every instance.
Still unacceptable.
They sampled security footage.
That became painful.
Example one.
Man in construction clothes enters.
Waits nine minutes.
Two salespeople available intermittently.
One greets couple arriving after him.
Man leaves.
Spreadsheet score:
No follow-up.
Example two.
Woman in athletic clothes arrives in older sedan.
Looks at touring model.
Waits six minutes.
Emily eventually approaches.
Sale completed four days later.
$41,000.
Visual score:
Walter stared.
Bias disproved by own data.
Example three.
Young man in designer clothes arrives in Mercedes.
Immediate Derek greeting.
Visual:
No purchase.
Three test rides.
Two hours salesperson time.
The system claimed efficiency.
Evidence said assumption often failed.
Rachel requested statistical sample.