Amazon is the audit you keep postponing

The marketplace question still gets argued as a matter of principle: brand control, margin erosion, channel conflict with the wholesale book. Meanwhile the number that should settle it sits in the Auto Care Factbook. Amazon's automotive department has become a $13.73 billion business, growing nearly 10% year over year, the 12th largest category on the platform, with an average unit price of $42.26 — the third highest on Amazon.

That is not a rounding error in a retail plan. It is a large, structured, well-instrumented competitor for the same part numbers your counter sells every day.

The channel split nobody budgets for

Auto Care's 2025 Joint E-commerce Trends and Outlook Forecast puts U.S. online aftermarket parts at roughly $23 billion excluding third-party marketplaces and $44.6 billion including them. Read that again as a distribution, not a headline: marketplaces account for close to half of the online dollars in this category.

B2B tells the same story from the other side. BigCommerce's read of the market puts marketplace transactions at roughly $20.7 billion of $42.4 billion in U.S. online parts revenue in 2024, growing every year since 2020, as shops, fleets and independent distributors moved procurement online. Marketplaces are not just a consumer channel anymore. They are where your wholesale customers already buy.

What Amazon actually is

Amazon is not a shelf. It is a fitment engine with a checkout attached.

My Garage and Confirmed Fit let a shopper enter a vehicle and see only compatible parts. That filtering is powered by ACES data submitted as XML feeds. Amazon's own documentation is unglamorous about what it needs: a BrandAAIAID in the file header that matches the brand on your ASINs, a part number that matches exactly, and applications at the year, make, model, submodel and engine level. Mismatch any of it and the fitment records simply do not attach.

A separate Seller Central account is required for submitting ACES data, and the Fitment Tool needs allowlisting before you can upload. Suppressed listings — the ones missing a required attribute or an image — still take orders in your headcount plan while showing up nowhere.

That is the entire game. Amazon does not reward the brand with the best banner. It rewards the brand whose catalog can answer "does it fit" without a human.

Where the money actually leaks

• 12% referral on automotive, 10% on tires. On a $42 average unit that is roughly $5 off the top, before fulfillment.
• Fulfillment and freight. Heavy, bulky, low-density parts are where FBA math fails; oversized surcharges and dimensional weight eat the rest.
• Fitment returns. Bad applications data produces returns at your cost, plus open-box inventory and review damage that suppresses the ASIN further.
• Buy-box competition with your own customers. Your authorized resellers and your own listings can end up bidding against each other on the same part number.
• Service retention. A customer who buys the part on Amazon does not book the bay. For a dealer, that is the margin you actually protect.
• Third-party sellers on your ASINs. Counterfeit or gray-market offers on a genuine OEM listing damage both price integrity and brand trust.

Which SKUs belong on the marketplace

The mistake is treating this as a company-level decision. It is a SKU-class decision, and it changes by category, price band and return profile.

The case for doing it even if you never scale

Here is the part most dealership and distributor strategy decks miss. Amazon is the cheapest fitment audit you will ever run. Its suppression reports, fitment coverage checks and exact part-number matching will surface gaps in your catalog that your own storefront hides behind search, merchandising and a forgiving counter staff.

Those same gaps are the reason your Google Shopping feed disapproves items, your Performance Max campaigns spend on the wrong queries, and an assistant cannot verify a part on your product page. The ACES and PIES discipline Amazon demands is the same discipline answer engines consume. Fix the catalog for one machine and you have fixed it for all of them.

So the strategic frame is not "should we list." It is: list a controlled subset, use the marketplace as a data proving ground, and keep the high-margin, fitment-complex, installation-dependent business on the channel where you own the customer and the bay.

Four questions before you list anything

1. How complete is fitment on your top 200 SKUs? If you cannot state a percentage, the answer is unknown, and unknown is why the last marketplace attempt failed.
2. What is contribution margin after 12%, fulfillment and a realistic return rate? If it is negative at an 8% return rate, it is deeply negative at the 15% a bad catalog produces.
3. Does the part need a bay? If installation drives the margin, the marketplace is a lead source at best.
4. Who controls the price? If you cannot enforce MAP and identify unauthorized sellers, you are donating margin to arbitrage.

The first 30 days, if you run it

• Audit ACES coverage on the 200 SKUs that carry 80% of revenue.
• Submit the feed, then read the suppression and fitment-mismatch reports like a search console.
• Align price, images, attributes and part numbers across your .com, the marketplace and the feed. Divergence is the root cause of most returns.
• Run Sponsored Products on exact part-number and fitment queries only. Broad discovery spend on a thin catalog is a donation.
• Measure the halo: branded search volume, direct sessions and service bookings in the markets where the listing went live.

The answer-engine connection

Amazon is now a corpus too. Reviews, Q&A and structured attributes feed its own assistant, and the parts that win there are the ones described completely enough to be verified. In 2027 the same question gets asked in three places: Amazon's search box, Google's shopping surface, and an assistant. One catalog answers all three, or none of them.

That is the real audit. Amazon just makes the invoice visible.