Valuation API

Building a lightweight vehicle valuation API over a weekend that cut costs from £3 to £0.0025 per lookup, boosting margins by 74,900%. Now used by dealers, it delivers instant, accurate valuations for cars and bikes, without relying on expensive providers.

Highlights

Simple & Valuable

An easy way for dealers to test my product and get valuations fast and free.

The Problem

Valuation providers like Experian, Cap HPI, and Brego charge up to £3 per valuation, with even higher rates for motorcycles. These valuations were essential to the Motorcart experience, yet were being handed out to users for free, these costs are unsustainable.

£3 per valuation was incompatible with a free-to-use product

£3 per valuation was incompatible with a free-to-use product

Motorcycles lacked public variant data (unlike cars via DVSA)

Motorcycles lacked public variant data (unlike cars via DVSA)

Motorcart needed instant, accurate valuations to drive conversions

Motorcart needed instant, accurate valuations to drive conversions

No scalable or affordable alternative existed

No scalable or affordable alternative existed

Discovery

To understand how valuation companies operated, I dove into research, reverse engineering, and competitor analysis. I broke down the essential components of a valuation engine and explored how data was being sourced, cleaned, and interpreted.

Investigated how third-party providers price vehicles

Investigated how third-party providers price vehicles

Identified use of datasets + rules-based/AI pricing logic

Identified use of datasets + rules-based/AI pricing logic

Mapped out how variants, body styles, and mileage impact prices

Mapped out how variants, body styles, and mileage impact prices

Strategy

The strategy was simple: build fast, validate fast. The goal was not to build the perfect system but to prove that accurate valuations could be delivered at near-zero cost.

Prioritized speed over perfection

Prioritized speed over perfection

Built a stripped-back prototype in Python

Built a stripped-back prototype in Python

Hardcoded assumptions: excellent condition, full service history

Hardcoded assumptions: excellent condition, full service history

Focused only on the most impactful inputs (year, make, model, mileage)

Focused only on the most impactful inputs (year, make, model, mileage)

The MVP

The first version was ultra-lean. Users entered year, make, model and mileage, and the api returned an average valuation assuming the vehicle was in excellent condition. It worked well enough to validate the core idea but exposed some key issues.

Features

The first MVP before vehicle lookup allowed users to just enter thier vehicle

The first MVP before vehicle lookup allowed users to just enter thier vehicle

Version 2 of the mvp had a vrm lookup

Version 2 of the mvp had a vrm lookup

Problems

No variant control (e.g. Carrera vs GT3 RS) returned averaged prices

No variant control (e.g. Carrera vs GT3 RS) returned averaged prices

Body styles like coupe vs cabriolet were not factored in

Body styles like coupe vs cabriolet were not factored in

The Solution

The final solution is a fast, scalable valuation API that delivers near-instant results at negligible cost. It now supports granular parameters with optional condition, service history, and feature inputs that adjust the valuation in real time.

Features

Supports both cars and motorcycles

Supports both cars and motorcycles

All you need is your VRM and mileage

All you need is your VRM and mileage

Dealers let users enter thier condition, service history, optional extras

Dealers let users enter thier condition, service history, optional extras

Built in logic to enhance accuracy instantly

Built in logic to enhance accuracy instantly

Outcomes

The valuation API has shifted from a weekend hack to a core platform asset, enabling dealers to offer high-volume, low-cost valuations that rival industry leaders, without the dependency or overhead.

Results

Reduced from £3 to £0.0025 (74,900% margin gain)

Reduced from £3 to £0.0025 (74,900% margin gain)

Currently 88% of valuations rated accurate by testers

Currently 88% of valuations rated accurate by testers

Gives price in seconds

Gives price in seconds

Currently in pilot use with select dealers

Currently in pilot use with select dealers