Autossimo V3 → V4Autodistribution2024 → 2026
Rebuild Autossimo, the B2B e-commerce platform of 23,000 professionals
The full rebuild of Autossimo, Autodistribution's B2B e-commerce platform: from V3 to a full-API V4, plugged into the new back office. Every module redone, while V3 keeps serving 23,000 professionals.
- Rebuild
- B2B e-commerce
- Full API
- Discovery
- AI
Impact
- 23,000professionals on the platform
- €700Min annual business, doubled over the period
- 0major service outage
- V3 journeys built on older services
- Rules scattered between interface and internal systems
- Each catalogue working its own way
- Licence plate or VIN, catalogue, basket, history redesigned
- The same APIs for every catalogue
- A journey-by-journey switch-over, with no downtime
The first V4 journeys I wrote the user stories for are live (login, search, catalogue, basket, history). Convergence continues, journey by journey.
Metrics trackedCompleted searches, from licence plate to partOrders placed and abandoned basketsMigration-related regressionsUsage per journey (Amplitude)
The engagement Autossimo, Autodistribution's B2B e-commerce platform: 23,000 automotive professionals, 3 million catalogue parts, €700M in annual business, several interconnected management systems (ERPs). Freelance Senior Product Owner from January 2021 to October 2026, with business teams, IT, the integrator SQLI and partners.
The brief
Rebuild Autossimo as V4, on new APIs and plugged into the new back office, redoing every V3 module in a new format, without ever stopping V3.
The context
Autossimo is where 23,000 automotive professionals search for and order their parts: €700M of business a year. The garage goes from a vehicle to a compatible part, checks a price and availability, then orders, across its distributor's rules and the logistics of several systems. V4 is meant to plug into the new back office: simply modernising V3 was not an option, everything had to be redone, starting from how garages use it today.
The challenge
Two versions to run at once: V3 in production, with its backlog and support, and V4 under construction. And a rebuild that goes beyond the interface: a favourite shared by a whole garage in V3 becomes per-user in V4; order history depends on matching order numbers across both versions. Those gaps had to be handled before promising the same thing as before.
What I did
Discovery: start from real usage
Audit of V3 journeys and user interviews to understand how garages use Autossimo today: search by registration plate, choosing the part, price, availability, ordering. Usage measured in Amplitude and BigQuery to confirm what really matters.
- Interviews
- V3 audit
- Amplitude
Technical and functional scoping, full API
API design workshops: contracts, endpoints, data exchanges. Rules split between the interface, the APIs and the distributors' management software (ERP), connection to the new back office. Mapping of the data that had to coexist between V3 and V4: customers, users, warehouses, favourites, orders.
- API + OpenAPI
- ERP
- Okta
Roadmap and backlog, with the Head of Product
V4's functional roadmap built in Productboard with the Head of Product. Features identified, broken down into epics then user stories, linked to Jira for sprint tracking, written with my AI agents. Backlog of the decision-support AI for garages: part search, advice, assembly, similar parts.
- Productboard
- Jira
- AI agents
Redo every V3 module
Every V3 module rebuilt in V4's new format, with the product designers in Figma: login and password, search by registration plate or vehicle serial number (VIN), catalogue, basket, history, account, notifications, access to returns. Messages and error cases included, one journey shared by every catalogue (parts, tyres, consumables).
- Figma
- UX / UI
Keep V3 running alongside
During the build, V3 stays in production: its backlog, agile rituals and support. Coordinating QA, business, IT and vendors, with indicators tracked after every release. No major outage, business doubled over the period.
- Run & Build
- 3-week sprints
Test and switch over, all the way to production
Testing against the existing version, in a test environment then in production, with accounts enabled progressively (feature flags). Switch-over ordered catalogue by catalogue, a tyre pilot prepared on 25% of one distributor's garages. Tyres validated in production in September 2026; convergence continues, journey by journey.
- Feature flags
- Tyre pilot
My role on this project
Driven end to end, all the way to production.
- UnderstandNeeds gathered with the business, the field and partners.
- ScopeScope, business rules, specs and user stories.
- PrioritiseBacklog ranked by value, shared roadmap.
- Build3-week sprints, coordinating business, IT and vendors.
- TestFunctional acceptance testing by journey, in Qase.
- ReleaseCatalogue-by-catalogue switch-over (feature flags), tyre pilot, tyres live in September 2026.
- RunV3 kept running alongside: backlog, support, indicators tracked after every release.
Testing, properly tooled Acceptance test suite restructured in Qase, test cases generated and automated with AI, every fix retested and every incident turned into a regression test.
Migration & decommissioning
Build the target, then switch off the old, breaking nothing.
Same method for Autossimo: V3 stays in production while V4 is built alongside, then the switch happens catalogue by catalogue, rolled out progressively behind feature flags. In progress.
- 23,000
- professionals on the platform
- 0
- major service interruption
- ×2
- volume over the period
- 1
Keep the existing running
The old tool stays in production and keeps being maintained throughout the redesign: users work without disruption.
- In production
- Old tool maintained
- 2
Build alongside
The new version is built next to the old one. The existing and the new are driven at the same time, on a single trajectory.
- In construction
- Both versions side by side
- 3
Switch over step by step
Each feature, then each distributor, moves from the old version to the new one, in waves, at national scale. Once everyone has migrated, the link to the old system is cut.
- No big bang
- Bridge cut at the end
How I drove the roadmap
A functional roadmap set by what the technology can deliver.
- Own the functional roadmap, with the Head of ProductIn Productboard: features identified, broken down into epics then user stories, linked to Jira. The technical roadmap (APIs, new service layer) is owned by IT; I align it with the functional roadmap at every planning session.
- No development on mocksIf a feature's API isn't ready on the technical side, its user stories are deprioritised in favour of tickets whose API is delivered. Every sprint moves forward on what can go all the way to production.
- Order the switch-overThe move to V4 happens catalogue by catalogue, rolled out progressively: tyres first, with a pilot on 25% of one distributor's garages, validated in production in September 2026.
- Keep V3 alive without freezing V4Requests on V3 in production and the V4 build are prioritised on a single trajectory: what ships in V3 is decided in light of what will be rebuilt in V4.
- History: filter by statusOrders API: delivered
- Basket: change a quantityBasket API: delivered
- VIN search: unknown vehicleVehicle API: delivered
- Next sprint, waiting for the API
- Logistics rebound: suggest the fallback warehouseMulti-warehouse stock API: not ready
- Logistics rebound: fallback lead timeMulti-warehouse stock API: not ready
