Contact us
A U.S. automotive commerce platform brings major retail partners, dealers, lenders and insurers together into one branded car-buying experience. Traffic sat flat for weeks, then a partner's campaign sent it surging, and every new partner meant setting up branding, access and data by hand. We put the platform on ECS with autoscaling, gave each partner tenant standardized provisioning, enforced data separation between tenants and joined marketing to one customer record.
Weeks of quiet traffic, then a partner's promotion.
Large partners wanted their own brand and their data kept apart.
As the platform reported.
The platform turns a retail partner's customers into car buyers, leasers and trade-ins, working with dealers, lenders and insurers behind the scenes. Its traffic is uneven: quiet for weeks, then multiples of normal within hours when a partner runs a national promotion. The old infrastructure was sized for an ordinary day, and scaling meant operators adding capacity by hand, mid-event.
Large partners put their brand on the experience, and their legal teams treat customer-data exposure as a deal-breaker. Tenant data was separated only logically, which doesn't satisfy that review. And a signed partnership still left engineers with branding, environment setup, access controls and marketing workflows to configure by hand. Meanwhile leads from expensive promotions could go weeks without a follow-up.
We treated each partner as a standard tenant, provisioned with its own data boundary, branding and access, and made capacity and follow-up automatic, so a promotion brings in customers without putting the platform at risk.
3 decisions, in the order we made them. Each one says what we did and why, with what it did drawn beside it.
ECS orchestrates the containerized services, with autoscaling tuned to promotional traffic, so each service scales on its own rather than the whole platform growing together, and the pre-provisioning that used to eat into operations time before every partner promotion is gone. The disaster-recovery design carries a one-hour recovery-time objective, built into the architecture and exercised regularly, which gives a partner contract a recovery commitment it can hold the platform to.
Partners judge the platform during their promotions, and capacity added after the spike has passed doesn't help anyone.
An Iterable integration automates multi-step lead nurturing and partner-branded campaigns across every brand at once. A customer-management integration provides one shared record across the marketing and sales workflow, so a shopper isn't three different people in three systems, which cuts duplicate outreach and missed handoffs.
As the platform reported.
More partner brands onboarded through the same provisioning, and the shared customer record extended across the dealer and lender workflows.
This project came before our current offers. Here's how we'd approach the same work now.
Through each promotion, AK Run's AI agents would take the first look at every alert and draft the fix, and a named engineer would confirm the cause and ship it with a test, with one team running the cloud, the applications and the security day and night. On AK AI Factory, an AI agent drafting follow-ups would get its own identity, only the tools its job needs and one partner's tenant in reach at a time, through Cedar policies checked outside the model on every call.
A free workshop: a few hours with the architects who'd do the work, on your own systems. No cost.
Book a free workshop