04
A fair price for a car part
Two of the 32 researched parts never shipped, because the audit could not make their price bands defensible. The gaps are the proof that the rest is real.
- Python
- public price research
- adversarial re-check
- PostgreSQL
Thirty price bands for common parts, in zloty, quality aftermarket rather than the cheapest thing on the shelf, each one independently re-verified against public Polish retailers before it was allowed to load. Where a genuine Toyota part has a published price, the row carries that too, so an owner can see the actual gap rather than being told which to buy.
The two that did not make it are the interesting ones. One sensor had a band that would have quietly priced the wrong variant of the part, and one exhaust component simply does not exist as a specific aftermarket item for this body shape, so any band would have been a guess dressed as a fact. The decisions are written into the generator rather than applied by hand, which means a re-run reproduces the same result, refusals included.
The AUDIT RESOLUTIONS the auditor demanded are encoded here, not hand-applied,
so a re-run reproduces the exact same seed (D008 discipline):
* DROP lambda_sensor - its 130-400 band prices universal/post-cat sensors
while the genuine UPSTREAM wideband sensor (the part an honest garage quote
is usually for) costs 680-1450 zl; one row cannot carry both without
misleading the owner. Confidence was 'low'. Held back, not shipped.
* DROP exhaust_silencer - no E170-SEDAN-specific aftermarket rear box exists
at any major Polish retailer (finder AND verifier confirmed independently);
the band was a platform-twin proxy. Confidence 'low'. Held back.
Aftermarket band against the genuine part
Amber is the quality aftermarket range. The dark tick, where one exists, is the published genuine Toyota price for the same item.
Source: 32 part curated research, 30 shipped. Part only, fitting not included. PLN.
How it was done
Research wide, then verify independently
Each part was researched against public retailers and then re checked by someone who had not done the original research, before any of it was allowed near the database.
Compare like with like
Bands describe quality aftermarket parts, not the cheapest item on the shelf, and they cover the part only, never fitting. Where a genuine manufacturer part has a published price, the row carries it as an anchor so an owner can see the real gap instead of being told which to buy.
Encode the refusals in the generator
The audit decisions live in the script that builds the data, not in someone’s memory, so a re run reproduces the same result including the two parts that were dropped.
Design for being superseded
Prices load as the freshest row per part, so a future automated feed can supersede the hand curated rows without touching the application.
What this does not tell you
Every dataset has an edge. Here is where this one stops, said plainly, because a number you have to qualify is worth more than one you do not.
- Two of the 32 researched parts are missing on purpose. One sensor had a band that would have priced the wrong variant of the part, and one exhaust component does not exist as a specific aftermarket item for this body shape. A guess dressed as a fact is worse than a gap.
- Part prices move. These are a point in time, and the schema is built so they can be replaced wholesale rather than patched.
- Fitting is not included, and labour rates vary far more between garages than part prices vary between retailers.