You have an alternator and no fitment data. There is an identical-looking alternator listed by somebody who clearly knows what they are doing, with a compatibility table forty rows deep.
Copying it takes ninety seconds. Building your own takes an afternoon. Everybody has done this. The question worth asking is why it breaks, because the answer is not "because it might be wrong" - it is more specific and more annoying than that.
You cannot see the provenance
Fitment data in circulation comes from roughly four places, and a compatibility table looks identical whichever one it came from.
| Source | Accuracy | Can you tell? |
|---|---|---|
| A licensed interchange catalogue | Accurate, maintained | No |
| Manufacturer catalogues | Accurate, incomplete across aftermarket | No |
| A spreadsheet somebody built once | Was accurate, years ago | No |
| Another seller's listing | Unknown - it came from one of the above | No |
So copying is not a gamble on accuracy. It is a gamble on lineage, with no way to inspect the thing you are betting on.
The error that grows
A spreadsheet written in 2021 is not wrong about 2021. It is wrong about everything after it. Every model year released since is missing. Every mid-cycle production change is missing. Every trim added or dropped is missing.
The sheet did not become inaccurate - the world moved and the sheet did not. Copy it now and you inherit five years of absence, in a category where the absence is invisible: your table looks full, and it is silently missing the newest vehicles, which are the ones with buyers who still repair rather than replace.
The mismatch that costs you an afternoon
eBay compatibility values come from dropdowns with specific text. Your source has its own text. When they differ in trivial ways the match fails - displacement in cubic inches on one side and litres on the other, a trim abbreviated here and spelled out there, an engine described by code in one and size in the other.
None of these look like errors. All of them can be enough to return an incorrect result. So you end up with a full-looking table, no error message anywhere, and no presence in filtered search. There is nothing to debug because nothing failed loudly.
What to use instead
- Start from a vehicle you actually had. The donor is a fact you own - its year, trim, engine and VIN are not inferences. Fitment reasoned outward from a known vehicle has a provenance you can state.
- Keep a record of where each row came from. Boring, and the difference between fixing a systemic error once and re-discovering it every quarter.
- Spot-check ten listings a month against eBay dropdown values. You are looking for the text-mismatch failure, and it clusters: if one engine designation is wrong it is wrong everywhere you used that source.
- Trim the rows you are not sure about. A narrower true table outperforms a wider hopeful one.
The uncomfortable trade
Building fitment from donor vehicles gives you data whose origin you can defend and coverage that is narrower than a licensed catalogue. That is a real trade, not a hidden win.
A licensed interchange catalogue knows this alternator also fits four vehicles you have never had on the lot. Your donor-derived data does not, and never will, because you only know what came through your own gate.
Which you prefer depends on whether your returns are currently driven by over-broad fitment or your missed sales by narrow fitment. Those are different problems with opposite fixes - and the method argument is here.
A copied table is not risky because it might be wrong. It is risky because you cannot tell whether it is.