Within 5 weeks of launch, inquiries about already-sold cars dropped by 95%, and managers recovered 12–15 hours per week thanks to a synchronized catalog and automated valuation and financing integrations.
The task was to build an online storefront for a network of offline car lots so the site would reflect the actual inventory across the lots. Beyond that, the platform had to support two customer flows. First: buying a car with a clear description, photo gallery, and financing options. Second: a sell-your-car flow with a preliminary valuation and a legal check on the vehicle.
Vehicle data is stored in a third-party SaaS dealer management system, photos arrive as large, unoptimized files, and the React frontend was built by another team. Our job was to give the client a stable API that would hold up for filters, galleries, availability statuses, financing parameters, and buyout requests — across a working catalog of roughly 800–1,200 active listings spread over several lots.
Â
Noah runs Alvergans, a network of used-car sales lots. Most of his customers come to the lot after already choosing a vehicle on the site. A recurring issue: people would drive to a lot to see a car that had already been sold. He wanted a site that keeps the catalog in order, gives a fast preliminary buyout valuation, and shows all the financing options available.
Outdated listings
Â
Because of manual updates and multiple data sources, the catalog kept showing cars that had already been sold.
Â
Heavy image galleries
Â
Galleries with dozens of photos per car loaded too slowly — a buyer could leave for another dealer before the page even finished loading.
Â
Unclear sell-your-car flow
Â
If a user submitted their details but got no clear sense of the valuation or the legal status of the vehicle, they were unlikely to return.
Â
Incorrect financing options
Â
Credit and leasing depended on price, down payment, term, currency, and provider-specific programs. Mistakes in those numbers damage trust before the first call.
Â
Frontend built by another team
Â
The project needed clear API contracts, stable data formats, and predictable error handling so the integration wouldn’t turn into an endless chain of clarifications.
Â
We proposed a separate backend layer on Directus with integration services — turning a messy stream of data from the dealer SaaS into a clean, fast, predictable catalog for the frontend. The external system stays the source of truth; the site only sees normalized data.
The business-side integrations mattered as much as the catalog sync: a third-party vehicle valuation service, a legal-check API, and two consumer-credit providers (a bank and a leasing company), all sitting behind the same API layer, so the frontend could work with one consistent interface instead of depending on each external system separately.
Â
Vehicle inventory sync
Â
Regular scheduled imports so new cars appear quickly and sold ones disappear from the site without a manager touching anything. We also handled potential edge cases such as price changes, listing removals, incomplete records, and duplicates.
Â
Unified vehicle data model
Entities for vehicles, trims, lots, media, and statuses designed so the frontend could work with them without being coupled to the dealer system’s internal structure. Technical fields for integrations and logging were kept in the model so problem records could be handled correctly.
Â
Media pipeline for galleries
Image processing was designed as a separate pipeline — multiple sizes for listing, card, and full-gallery views, size control, and predictable URLs. This reduced photo weight and made loading faster.
Â
Sell-your-car flow with an answer, not just a form
We proposed turning the vehicle-sale form into a scenario that responds. The user enters the details and gets a valuation range and a baseline legal-check result; the manager gets a structured lead that can be worked without back-and-forth.
Â
Financing on the vehicle page
A module that prepares credit and leasing parameters for the specific car and stores the user’s selected options alongside the lead — so the manager starts the call from the terms the customer already liked.
Â
API contract as a delivery agreement
Versioned endpoints, consistent error formats, clear filtering rules — a stable release the frontend team could build against without surprises.
Â
We started by asking Noah what the new storefront had to do and what was breaking sales right now. The problems were typical for this niche: sold cars lingering on the site, prices out of sync with the dealer database, photos getting mixed up between listings. From there, we locked in the requirements for each flow — buying, trade-in, the sell-your-car form with preliminary valuation, legal check, financing options — around a single rule the site had to follow: only show what’s actually on the lot.
Â
In the first 5 weeks after the backend layer went live, inquiries about already-sold cars dropped sharply, and managers stopped manually pruning stale listings from the catalog.
Â
inquiries for cars that were already sold
freed up across the manager team
Â
first contact with leads
Â
inbound requests
40+ freshly arrived cars synced to the site in under 15 minutes — work that would previously have taken the team the better part of a day
Or send us a message and we'll get back to you within 15 minutes during business hours