A vehicle application arrives at the counter, not a part number. Several stocked lines can legally fill it at different costs, warranty terms and core liabilities, and the one sold is usually whichever the catalogue ranks first. Revomo binds transactions to coverage, shows what each fill route earns, and publishes a governed policy.
Demand arrives as a vehicle and a failed component, not a SKU. The catalogue turns that into qualified part numbers, several already on the shelf at different costs. Choosing between them moves hundreds of basis points, thousands of times a day.
Substitution is the structure of this market, not a grey practice, and a public filer states it plainly. LKQ describes buyers of vehicle replacement products choosing among "primarily four sources": new OEM, aftermarket, salvaged and reconditioned, the last three of which it calls "alternative parts".1
Generic distribution has a timing problem it can see. The aftermarket has an identity problem it cannot: the ERP stores the part that was sold and never the application that was asked for, so the counterfactual does not exist in the record by construction.
Margin here is analyzed by part number and by customer, and both start from the number that was sold, the one thing the decision did not choose. Which other lines could have filled the request is not in the record.
The shop’s habit of calling you first is the commercial unit, and depth of stock buys it. Availability is a ladder: the counter holds fast movers, a hub the next tier, a DC the tail. The specimen runs the same ratio, smaller.
O’Reilly reports domestic stores averaging approximately 24,000 SKUs, hub stores 63,000, and distribution centres over 156,000.2
A distributor publishes a price to a branch it owns and a program to an independent, then watches what that independent does with it. Neither instrument reaches a reseller’s own resale decision.
Genuine Parts serves 6,864 US and Canadian locations, approximately 35% company-owned and 65% independently-owned.3
At the warehouse step, vendor programs are frequently the margin: volume rebates, growth tiers, co-op, stock lift. Earned on purchases already made and settled later, so program-inclusive economics rest on money not yet arrived.
Genuine Parts recorded $160 million of non-recurring charges in the fourth quarter of 2025, primarily credit losses on volume purchase rebates due from a vendor that filed for Chapter 11.4
Two systems of record here do not join: one knows what was sold, the other what fits. The decision needs both, and the outcome lands where orders are priced.
Counter transactions, order pricing, where changes land. Examples:
Application coverage and attributes, as a source:
Transaction and coverage history, as a source:
The records with no system, that move the decision:
Data is ingested through scheduled connectors into Revomo’s governed store, and governed outputs are written back where orders are priced. Warehouses are sources, not the store of record, and so is the catalogue: Revomo does not author fitment data.
This spine starts at a vehicle, not a supplier cost file. What the counter can reach and what the account resolves to meet at one fork.
The record that exists nowhere else: the application, bound to every qualified line, level, program and core window. Coverage is an ingested source; Revomo does not author fitment data.
Explore Commercial ContextRanked by how directly the decision moves margin and how fast a distributor can act. Not the macro page’s order: cost recovery ranks last here, not second.
Why it is hardThe transaction system stores the part number sold, never the application asked for. Reconstructing the alternatives needs the catalogue, a different system on another cycle.
Explore Customer & Product ProfitabilityWhy it is hardThe buyout looks like service, the credit memo like accounting, the stale cost like data. None appears as a pricing decision, so the margin report is accurate and the conclusion wrong.
Explore Margin Leakage & Price RealizationWhy it is hardThese are not discounts off one list. They are positions in a supply chain, each a different business with its own resale margin, so moving an account changes who that account is.
Explore Price Lists & Customer-Specific PricingWhy it is hardThe return right is capped at a percentage of annual purchases and expires. The core is a deposit against a unit travelling back through three companies. Neither usually has an owner.
Why it is hardSupplier statements arrive in a different period, unit of measure and part grouping from the purchases they settle, so reconciliation is a quarterly project rather than a running balance.
Explore Rebates, SPAs & ChargebacksWhy it is hardBecause the counter price is computed from cost by a matrix band, a dollar passed through as a dollar defends profit in dollars, dilutes the rate, and moves parts across bands so the multiplier changes too.
Explore Cost, Tariff & Freight RecoveryThe quarterly coverage and supersession refresh lands. On 34,800 applications a line the group already stocks has gained qualified coverage, 4,260 numbers are superseded and 1,180 discontinued. Nobody will make 34,800 decisions, so the counter follows whatever the catalogue ranks first. Depth runs 25,900 SKUs at a store, 66,500 at a hub, 168,000 at a DC.
The file lands as a governed dataset rather than a spreadsheet in an inbox, so everything below computes on the same records the business runs on.
Costs nothing and hands the decision to a vendor’s ranking. The negative number is the point: the refresh moved 6,340 applications onto a line costing 2.6 points more, so no change is itself a change.
Takes the most margin per line and gives back what the account is buying. Fill falls 3.3 points because the cheapest qualified line is often not on the shelf, and the model prices the line lost, not the first call.
Recovers 81% of the pure-margin case while raising fill, because the buyout and the line selection are one decision made twice. Not the highest-margin answer, and the one 180 counters will execute.
| Segment | Applications | Lines / yr | Move | Recovery | Why this action |
|---|---|---|---|---|---|
| Premium duplicated by a stocked mid line | 7,920 | 116,300 | Default to mid line | +$792K | Equal coverage and warranty, 3.9 points cheaper. |
| Non-safety-critical, coverage verified | 5,480 | 73,300 | Default to private label | +$622K | Highest-margin fill, coverage confirmed at submodel. |
| Safety-critical groups | 7,070 | 97,000 | Hold incumbent line | $0 | These do not move on a margin argument. |
| Repeat buyouts against stock already held | 3,630 | 54,100 | Stock at branch | +$540K | A stocking decision priced at 9.3%, net of carrying cost. |
| Superseded numbers on a stale cost | 1,526 | 25,500 | Reprice to current cost | +$544K | One correction repairs every level under it. |
| National account files in force | 4,180 | 67,800 | Hold to term | $0 | Excluded by agreement and effective date. |
| Total | 29,806 of 34,800 · 434,000 | +$2.50M | Annualized, net of modeled volume response | ||
| Rule | When | Then | Approval |
|---|---|---|---|
LS·COV | Coverage incomplete at submodel, engine or position | Ineligible to become the default | Auto |
LS·SAFE | Application sits in a safety-critical group | Default holds on the incumbent line | Auto |
LS·WAR | Candidate warranty term is shorter than the incumbent | Sign-off with the delta | Category Manager |
LS·FILL | Modeled branch fill rate falls below 92.0% | Block until stocking is set | Auto |
LS·CORE | Change raises core-bearing share above current | Core liability review | Controller |
LS·CHAN | Application sits on a national account file | Hold the channel-level price to term | Auto |
LS·MARGIN | Margin improves by less than 150 bps | Do not re-default | Auto |
The draft was backtested against twelve months of counter transactions and shadow-run beside the live catalogue ranking in two hubs before publishing. The rule participates before the change is proposed, so a coverage-incomplete candidate never reaches an approver and "why this line" is a record rather than an argument.
The agent reasons. It does not publish. Reasoning does not confer authority: the state change is a governed transition, and policy decides if it is legal.
There is no supplier effective date here, which is what makes it easy to postpone. The clock is how fast a counter forms a habit: once 180 stores have learned a default, changing it again costs more than setting it right did. The run also claims $1.42M of discontinued stock before its window closes.
The $350K gap decomposes to three causes and no residual. $213K from adherence below plan at 22 branches where staff kept quoting the incumbent from memory, a training problem rather than a pricing one. $92K from 1,180 applications where private-label coverage proved incomplete. $45K from a core window closing on 660 units.
The next decision starts with the outcome of this one.
An adoption order, not a catalog. Each tier produces an artefact the next depends on, starting with the join nothing else performs.
Nothing above it is answerable without the join.
Transactions, applications, coverage, levels, programs and cores on one foundation, so the alternatives exist as records.
Know which fill earned what before changing a default.
Margin by fill route with the buyout isolated, realization by channel level, accrual against settlement and core liability by window.
Turn the finding into something publishable.
Every channel-level file on one spine with precedence, the matrix as a decision table, line-selection rules, backtest and shadow-run.
From what each fill earned to which it should have been.
Analysis, predictive and optimization cells with declared state, so the model and the solve are inspectable.
Line selection solved with the fill floor, coverage, safety holds and core caps as binding constraints.
Reach where programs and agreements are operated.
Volume, growth, co-op and stock-lift programs run against the purchases they attach to.
Fleet and national account files as entitlements read before a price publishes.
Relevant on the fleet lane. Counter pricing is not quoting, which is why it sits here rather than three tiers higher.
Coverage and core windows should not be watched by hand.
Why an application defaults to a line, answered as records.
Monitoring the coverage refresh, core window and allowance expiry. Live changes cross a governed transition.
Revomo enters through one decision. Here that is usually one category on one banner, because the coverage join can be proved on a single friction book. That holds on one DC or on several ERP instances: commercial transformation decomposes into decision systems, and this one has the shortest path to a published change.
Start with the economic problem. Expand on the same platform.
The aftermarket is not one commercial model. These four sit at different points of the same chain, and the first phase follows the step.
You sell to independent jobbers who set their own resale, and program income is a material share of earnings.
Price is decided at the counter under an availability constraint, and buyouts are daily.
One inventory serves two ladders, and the walk-in price is searchable while the professional price is not.
Deep coverage on a narrow parc, few competing lines, a long tail priced on scarcity.
Bring a slice of counter transactions and the coverage file you license. We will bind them, decompose margin by fill route, and turn the finding into a governed default.
Externally sourced figures on this page, in order of appearance. Every other number is specimen data inside the decision story and is labelled illustrative where it appears.