Case study
A prefab-home marketplace
275 listings across the U.S. and Canada. Scraped catalog data going stale. Leads given away free and ignored. Manufacturers with no reason to stay.
- Strategy
- Research
- Product design
- Build
The situation
A marketplace for prefab, modular and tiny homes — roughly 275 listings, built on a website platform that couldn't do much more than display them. The listing data was scraped from manufacturers and went stale with no way to refresh it. Leads were handed to busy builders for free, and most were never followed up on.
The deeper problem wasn't the software. It was that the business had no reason to exist for either side of its own marketplace.
The reframe
“The site was a mirror of the manufacturers. It reproduced their catalogs and added nothing a buyer couldn't get direct. The fix isn't a better mirror — it's owning everything the manufacturer doesn't do.”
A manufacturer sells you a home. Nobody sells you the part that actually stops people: the land, the permits, the site work, the financing, the transport, the crane, the licensed installer, and the six months of sequencing that connects them.
That's the business. The catalog becomes the front door, not the product.
Why this one was mine to do
I've sold manufactured homes and I've set them. I know what a dealership's incentives look like from the inside, what a hauler will and won't take on, what a county permit office says when a home doesn't fit a category they recognize, and how often a buyer gets promised the moon by someone who won't be there on install day.
None of that was research I had to go do. It's why I could see inside a week that the business was a mirror — and why I could then build the software to fix it. Most consultants have one half of that. Most builders have the other.
What I delivered
- 01
A strategic reframe
Repositioned the marketplace as the operating system for buying a prefab home, structured as a nine-step buyer journey — discover, ask, find land, site prep, permits, ship, set and install, finance, delivered — with a vendor on every step.
- 02
Primary research, not a summary
A channel map of how prefab is actually sold (dealer-gated vs. factory-direct, and the seam that predicts which), plus a logistics and licensing study: four distinct transport pools by home type, the per-state installer-licensing patchwork, and why financing effectively blocks a DIY foundation. The finding that mattered: nobody prices delivered-to-your-parcel or books installers nationally. That's the wedge.
- 03
A working demo, not a deck
Five clickable pages in their own brand — living catalog with freshness and landed-cost-to-ZIP badges, a home detail with a cost breakdown, the nine-step project journey, and an AI concierge that answers zoning and permit questions in their voice. Built so the team could click it, not imagine it.
- 04
The build plan
An execute-in-order runbook: schema, data import, pages and SEO, the freshness engine that keeps listings honest, the AI layer, and a staging-first cutover that preserves their existing search rankings. Written so it could be handed to anyone — and then I started building it.
What it unlocked
Seven revenue avenues where there had been one
Vendors pay five ways across every trade in the journey; buyers pay for the guided build; and the accumulated market data becomes a product of its own. If one stream is slow, six others still pay.
A relationship instead of a sold lead
A buyer with an account and a live project is worth thousands over six to twelve months. A lead sold once is worth a referral fee. Same person, two different businesses.
A moat that compounds
The AI researches an unknown county's permit requirements once, then remembers it — so the next buyer in that county gets an instant answer. Every project makes the next one better, and it's not something a competitor can buy.
Got a business that's leaking value like this one?
Tell me what's actually wrong. If I can help, I'll come back with the reframe and a plan.