Back to workCase study
VIN Decode Without API Latency
Built a client-side VIN decoding system from historical vehicle records, enabling instant make/model/year suggestions for most entries.
- JavaScript
- Backbone.js
Context
Users were entering VIN, make, model, and year manually, which introduced friction and input errors.
The product needed fast, reliable prefill behavior without hard dependency on external lookup services.
Historical production data already contained the labels needed to bootstrap an internal decode reference.
Approach
- Grouped VINs by prefix and retained mappings with strong sample size and agreement thresholds.
- Separated deterministic year decoding from make/model lookups to keep mapping logic clean.
- Loaded manufacturer-specific shards on demand so decode stayed fast without bloating the main bundle.
Outcomes
- Enabled instant client-side decode for the majority of VIN submissions.
- Reduced manual entry errors and form completion time.
- Created a data asset that improved organically as new vehicles were entered.