Auto Parts Fitment Data: What the RC Omits
A part fits a specific make, engine and year of manufacture, and a catalogue that cannot express that generates returns no photograph prevents. India has no shared vehicle table to draw from, so every seller's application list is homemade.
- Fitment is an application list, not an attribute. Amazon's Part Finder resolves year, make and model and may need trim or engine type, and its category style guide tells sellers to keep vehicle names out of titles and bullets for fitment specific parts.
- The Auto Care Association's Vehicle Configuration Database, which underpins ACES, covers vehicles sold in the USA, Canada, Mexico and Latin America. There is no Indian equivalent to subscribe to, so every seller's vehicle spine is built in house.
- Form 23A under rule 48 of the CMVR carries maker, fuel, cubic capacity, cylinders, date of manufacture and chassis number, but no model year and no variant field. Ask the buyer for what is printed, and use the chassis number to break ties.
- Rule 6(3) of the Consumer Protection (E-Commerce) Rules, 2020 requires a seller to take back goods that are not of the characteristics as advertised. A fitment claim is such a characteristic, and Amazon.in disclaims responsibility for suggested fit accuracy.
A shoe that does not fit comes back folded and gets resold. A wheel hub that does not fit comes back scratched, after a mechanic has had it in his hands and charged for the hour. Same return reason on the dashboard, different economics, different root cause.
Fitment belongs to this category and no other. It is a catalogue problem before it is a returns problem, and the two get confused constantly.
Fitment is a relationship, not an attribute
Most categories describe a product with attributes: size, colour, material, wattage. An auto part needs one more thing, a list of the vehicles it applies to. Amazon’s category style guide for Automotive and Powersports calls this fitment or application data, the brand and part number combinations that apply to particular vehicle configurations. Its Part Finder asks the shopper for year, make and model, and notes that trim or engine type may also be needed.
Two consequences follow. Universal articles are deliberately kept out of the tool, with floor mats, seat covers and scan tools given as examples. And the same guide instructs sellers not to put year, make and model into titles, bullets or descriptions for fitment specific parts, because the finder carries that load. Most Indian auto listings do the opposite and stack four vehicle names into a title that then truncates on mobile before it reaches the part type.
The reference table India does not have
In the United States the application list is not improvised. ACES, the aftermarket catalog exchange standard, is underpinned by the Auto Care Association’s Vehicle Configuration Database, so competing catalogues draw vehicle rows from one shared table. The Auto Care Association describes that database as covering vehicles, powersports, off highway and equipment sold in the USA, Canada, Mexico and Latin America. India does not appear in its coverage.
There is no Indian equivalent to subscribe to. The Vahan registration system run by the Ministry of Road Transport and Highways holds registrations, not applications: it indicates what is on the road, not which bearing belongs in it. So whatever vehicle spine your catalogue uses, you built it, your competitor built a different one, and the marketplace is matching free text across the two. That is why fitment quality is a durable advantage here rather than table stakes.
What the buyer’s own document actually says
Ask an Indian buyer for year, make, model and variant and you are asking for something they cannot read off their registration certificate. Form 23A under rule 48 of the Central Motor Vehicles Rules, 1989, the smart card certificate of registration, prints in its visual inspection zone the name of the manufacturer with make, colour, fuel, vehicle class, body type, seating capacity, date of manufacture as month and year, unladen weight, cubic capacity, wheel base, number of cylinders, chassis number and engine number. The older paper Form 23 adds a maker’s classification field.
Read that list for what is missing. There is no model year. There is a date of manufacture in mm-yyyy, a different fact, often a year behind the registration. There is no trim or variant field of the kind a part finder interrogates. What is there, and is exact, is the chassis number and the engine number, which rule 122 requires to be embossed, etched or punched on the vehicle with the month and year of manufacture.
Design the question around what the buyer can see. Maker, fuel, cubic capacity, cylinders and month of manufacture separate most applications on their own. Where they do not, take the chassis number and resolve it at your end, rather than making a customer guess a variant name off a brochure.
How a loose mapping fails
Four patterns account for most of the damage. A nameplate that ran nine years through two facelifts, carried as one vehicle. One model sold with two engines, where the part differs only on the diesel. A superseded OE part number, with the cross reference still pointing at the old one. And the word universal, doing the work of a mapping nobody performed.
The property that matters operationally is that fitment errors cluster by application, not by SKU. A wrong row does not lift your return rate by a fraction across the range. It returns every order placed for that one vehicle. The signal arrives as a lump: a quiet part number takes four returns in a week and every buyer drives the same model. Group returns by vehicle before you group them by SKU, or you will read a data defect as a quality problem.
Why this is not a sizing return
Three things separate a misfit from the sizing returns that dominate apparel. The first is where discovery happens. The customer finds out at a workshop, having already paid for a slot and sometimes for removal of the old part, so the complaint carries a cost you never charged. The second is condition on arrival. A garment comes back wearable. A part comes back handled, often fitted and removed, which pushes it down the ladder before it reaches your grading bay.
The third is the legal position. Rule 6(3) of the Consumer Protection (E-Commerce) Rules, 2020, notified as G.S.R. 462(E) on 23 July 2020, provides that a seller on a marketplace shall not refuse to take back goods or refuse to refund where the goods are not of the characteristics or features as advertised or as agreed. A fitment statement is a characteristic as advertised. Your returns window is not the operative document here.
Nor does the platform stand behind the mapping. The suggested fit note on Amazon.in says the indication rests largely on details provided by sellers and the shopper, and adds that “Amazon is not responsible for any inaccuracy in this regard”. The tool renders your data. The accuracy is yours.
Run the application list like inventory
One row per part number and vehicle configuration, dated and sourced, and only three sources count: an OE cross reference, a physical measurement against a known vehicle, or a workshop confirmation. Retire rows on evidence and record why, because a silently deleted row is how the same error returns next quarter. Give a confirmed misfit an owner in the catalogue team rather than the support queue, and count it alongside the other catalogue decay you already audit.
Handled that way the application list becomes the most valuable thing in the catalogue, because in this market nobody can buy one.