Floor Nexus

We assessed what could be kept and what could not, and the honest answer was that around sixty percent of the existing product needed to be stripped out and rebuilt. That is not a recommendation anyone enjoys delivering, but pretending otherwise would have meant stacking new features on foundations that were already the reason delivery had stalled.

The rebuild ran in phases behind live traffic over roughly five months. New implementations went in alongside old ones, usage shifted across gradually, and the old paths came out last. At no point did the roadmap freeze — the company kept shipping customer-visible work the entire time the structural replacement was happening underneath it.

Staffing was its own problem. We interviewed the founder's former team, brought two of them back on their merits alongside the two already retained, and let the rest go. That gave us four people who carried real institutional knowledge of the product and its customers.

On top of that we brought in four to six Fluid Web engineers to start. When the company secured investment, we scaled the Fluid Web side by another eight, taking the combined team to roughly sixteen. Today twelve engineers are active on the platform, and expansion is still underway.

Floor Nexus is not one application. Contractors run jobs from an installer platform and a mobile app, homeowners track their work through a customer portal, manufacturers and distributors operate through a vendor platform with its own storefront and point-of-sale, and the company runs all of it through an admin layer.

Each of those surfaces has different users, different permissions and different failure modes, but they resolve against one product model. Building them in parallel rather than sequentially is the only way they stay coherent — mobile that ships a year behind web is not a surface, it is a liability.

Billing was one of the clearest symptoms of the original build. Everything was modelled at the user level — there were no business-level plans at all — which is the wrong shape for a product sold to contracting companies where one business has a crew, an office team and an owner who all need access. Charging per individual and having no concept of a company account meant the commercial model the business actually wanted was not expressible in the software.

Fixing that meant reworking the entitlement model so plans attach to a business, with seats and permissions resolving underneath it. That is not a billing change so much as a tenancy change, because who pays and who has access have to agree with each other at every level.

On top of the model rework, we migrated the entire payment setup from an old Stripe account onto a new one, and from the legacy integration onto Stripe's newer API. Moving billing between accounts on a product that is actively collecting money is about as unforgiving as this work gets — active subscriptions, payment methods and in-flight transactions all have to land intact, and customers cannot notice.

The platform supports roughly forty-five manufacturers and around eight thousand products that installers can quote and order against. Keeping a catalog that size coherent — pricing tiers, availability, warranty rules, vendor-specific terms — is quietly one of the harder parts of the system.

Five manufacturers now run their CRM and point-of-sale directly on the vendor platform, with more onboarding in December. That turns Floor Nexus from a contractor tool into a marketplace with both sides operating on it.

Weekly shipping started the day we took over, and in practice it is often several releases a week. That was deliberate: on a rescue, restoring visible delivery early is what rebuilds confidence internally, and it forces the deployment path to become reliable before anyone depends on it for something large.

It also meant the founder could stay on customers, partnerships and fundraising instead of managing engineering, which is where the real leverage was.