Ohio and Michigan share only 22 tire sizes.

Two adjacent Midwestern states. Similar climate, similar income, similar driving. Build a stocking list for each from the vehicles actually registered there and the lists barely overlap.

Most distributors stock from a national list. Some version of a top 100, top 150, or top 200 by size, adjusted at the margins by regional judgment and whatever the branch manager has been complaining about. It is a reasonable heuristic and it has run this industry for decades.

We tested it. Take two neighboring states, pull the vehicles actually in operation in each, map them to their fitments, and ask a simple question: how many distinct sizes does it take to cover 80% of replacement demand?

The result

Thirty-three sizes covers 80% of Ohio. It takes 231 to cover the same 80% of Michigan. And only 22 sizes appear on both lists.

A single national stocking list is simultaneously too big for one of these states and far too small for the other.

Why the gap is this large

The explanation is in the registered fleet, not in consumer preference.

Ohio’s vehicle population skews heavily toward light trucks, work vehicles, and commercial applications. Light-truck fitments are standardized. A relatively small number of sizes serve an enormous number of vehicles, because pickups and work vans across manufacturers and model years converge on the same handful of dimensions. Demand concentrates.

Michigan’s fleet is passenger and crossover heavy, with the trim-level diversity that comes from being the state where the domestic manufacturers sell hardest to their own workforce. Passenger and CUV fitments proliferate. Wheel diameter creeps upward every model year, trim packages fragment sizes within a single nameplate, and demand spreads across a long, thin tail.

Same region. Same weather. Same roads. Completely different demand structures, because a demand structure is a function of what is parked there, not of where it is parked.

What a national list actually costs you

Apply a national top-100 to both states and you make two different errors at once.

In Ohio, you overstock

You are carrying roughly 67 sizes beyond what the fleet requires to hit 80% coverage. That is capital sitting in a rack, aging, occupying slots, and generating carrying cost against demand that is not there. It does not look like a problem on a report, because the sizes do eventually turn — slowly, at a discount, sometimes transferred to another branch.

In Michigan, you understock

You are missing 131 of the sizes required to reach the same coverage level. Those become special orders, lost sales, or substitutions into whatever is on hand. Substitution is the invisible one: the customer drives away on something, the ticket closes, and nothing in the system records that you sold your second choice at your second-best margin.

Why this hides so well

Here is the part that makes this durable. Roll both branches into a regional P&L and the errors partially cancel. Ohio’s excess inventory and Michigan’s lost sales average into something that looks like ordinary performance. Turn rate looks acceptable. Fill rate looks acceptable. Nothing triggers an investigation.

The signal only appears when you stop measuring against your own sales history and start measuring against the vehicles that are actually there. Sales history is downstream of stocking decisions — it tells you what you sold, which is bounded by what you chose to carry. Ask it whether your stocking was right and it will confirm whatever you did last year.

The unit of analysis is wrong almost everywhere

The state is not the right boundary. Neither is the region, the DMA, or the sales territory. Those are administrative lines drawn for other purposes, and a fleet does not respect any of them.

The right boundary is the geography a given location actually serves, and the vehicles registered inside it. Two facilities four hours apart can require materially different lists. That is not a failure of standardization — it is the correct answer, and treating it as noise is what turns working capital into a rack of sizes nobody in that market drives.

Why counting vehicles is not enough

Here is the part that trips up most attempts at this. Pulling registration counts and mapping them to fitments is the easy half. It produces a ranked list that feels authoritative and is quietly wrong, because vehicles in operation do not convert to replacement demand at a uniform rate.

• A vehicle's age tells you very little about where it sits in its tire cycle
• Replacement intervals diverge widely across the fleet in ways a blended rate erases
• Fleet exit rates vary by segment, so raw counts overweight vehicles on their way out
• Sizes on current OE that have not reached first replacement rank near zero today and near the top within a few years

Get the conversion layer wrong and you have built a more expensive version of the sales-history model you were trying to escape: confident about the past, blind to what is arriving.

How that conversion is modeled is where TreadSignal's approach differs from every other read on this market, and it is the part we keep. What we publish is the output.