The hidden cost of building an in-house auto parts catalog and why CFOs are picking API-as-a-Product

The global online auto parts market is valued at roughly $135B USD in 2026 and projected to reach $440B by 2034, growing 16% annually. European aftermarket e-commerce will pass $38.5B USD by 2028. The opportunity is real and demand exists — but the technical barrier to entry is the same for everyone: guaranteeing the shipped part actually fits.

Five years ago, building a vehicle catalog in-house was a reasonable call: OEM data was scattered, commercial APIs were expensive and rigid, and a 2-3 engineer team could maintain decent coverage. In 2026 that math inverted. Specialized APIs are faster, more complete and cheaper than any reasonable in-house effort — and constant model changes, SSPL supersessions and local plants make maintenance more expensive than initial development.

If you're CEO or CFO of a parts platform, a marketplace or an automotive ERP, this is probably the question on the next product meeting agenda. This article gives you the real numbers to answer it.

When a tech team estimates 'building a VIN decoder', they typically picture consuming the free NHTSA API, mapping it against a couple of supplier catalogs and wrapping it in a REST endpoint. That's 15% of the actual work.

The other 85% is what you find six months in: NHTSA doesn't return part numbers, supplier catalogs cover ~60% of the fleet, Latin American plants aren't mapped, SSPL supersessions don't exist in public sources, and each OEM has its own schema requiring its own normalization module.

• Sprint 0-3 months: research, schema design, NHTSA integration — 1 senior + 1 mid · ~$45k USD.
• Sprint 3-6 months: scraping/parsing supplier catalogs, first fitment pass — 2 mids · ~$65k USD.
• Sprint 6-9 months: LatAm coverage, local plants, cross-OEM normalization — 1 senior + 1 mid · ~$70k USD.
• Sprint 9-12 months: maintenance, bugs, sales team requests, manual supersessions — full team · ~$100k USD.
• Year-1 total (conservative): $280k-$450k USD depending on senior/mid mix and team geography.

Initial budget looks at development. Real cost lives in maintenance. A vehicle catalog isn't a project you finish — it's a live database needing constant updates: model facelifts, new engines, plant changes, SSPL supersessions, discontinued parts, new OEMs in the portfolio.

Teams that built in-house report on average 40-80 hours/month of maintenance just to not lose coverage. At loaded mid rate of $60-80/hour, that's $2,400-$6,400 USD/month recurring — without counting the senior who has to jump in when something breaks in production at 3 AM.

Then there's the return cost. A wrong fitment shipped to the customer costs twice: reverse logistics + the customer who doesn't come back. In parts e-commerce, an 8% wrong-fitment return rate isn't unusual when the base is in-house with partial coverage. Each percentage point of returns costs 1.5%-3% of net margin.

While your team spends 12 months building a VIN-to-OEM catalog, what aren't they building? Competing platforms that integrated an API in 2 weeks used those extra 11.5 months to ship:

• Improved cart with oil/filter upsell on related parts (+8-12% AOV typical).
• Loyalty program for workshops with purchase tracking and volume discounts.
• Mobile app for sales reps on the road with offline quoting.
• WhatsApp Business integration for conversational quotes (where the LatAm customer lives).
• Aftermarket marketplace compatible with automatic cross-selling on top of OEM.

A modern VIN Decoder API isn't just an endpoint. It's a managed service that includes: official EPC connections to OEMs (with signed legal agreements), automatic schema normalization across brands, live SSPL supersessions, uptime and latency SLA, dedicated technical support, OpenAPI docs, TS/Python SDKs, free sandbox and coverage that updates at the manufacturer's pace — all without your team writing a single line of maintenance code.

Typical pricing for a specialized API runs between $18,000 and $60,000 USD per year depending on volume. Compared to $280k-$450k USD in year 1 in-house + $30k-$80k USD/year recurring maintenance, net savings are 5-8x in year 1 and hold at 3-5x in following years.

The following numbers are conservative estimates based on real projects we've seen across LatAm, US and Europe. They include direct costs (development, licenses, infra) and indirect costs (maintenance, wrong-fitment returns, opportunity cost measured as features not shipped).

Build isn't always the worst option. It makes sense in three specific scenarios: (1) you're a vehicle manufacturer or a dealership group with direct access to your own proprietary EPC and enough internal volume to amortize development; (2) your core business IS the vehicle catalog and you'll compete selling it as a product to third parties (you're a competitor of ours, not a customer); (3) you have extreme data sovereignty requirements (defense, government) where no external vendor is acceptable.

Outside those three cases — which cover <5% of companies asking the question — the math always favors Buy. And even in those three cases, most end up doing Buy + their own layer on top, not pure Build.

If you're taking this decision to your board, skip the technical debate ("we have the talent to build it") and frame it financially. The three questions your board should answer: what's a 2-week vs 12-month Time-to-Market worth to us? what does each month of delayed launch cost in lost opportunity? who owns catalog coverage maintenance in year 3, 5 and 7?

When those three questions are answered with numbers, the decision stops being technical and becomes obvious. The engineering team almost always prefers to build (interesting work); the finance team almost always prefers to buy (clear math). The CEO has to break the tie by looking at market speed.