Cory Hesseltine was walking his dog on a Sunday when he heard me on a podcast. That is the entire origin story. No RFP, no referral, no conference booth. A guy in Seattle with a company to run, some time to kill, and a pair of headphones.
What came out of that walk is the build I get asked about most, so I want to put the whole thing down in one place — including the parts that were slower and messier than the highlight reel suggests.
He interviewed a dozen agencies and pushed on all of them
Cory talked to roughly twelve agencies before he chose one. That number matters less than what he did during those conversations: he pushed back on purpose, to see what happened.
It is a good test and more owners should run it. Disagree with the person selling you something and watch which way they move. Most agencies fold immediately, because the sale matters more than the recommendation. A few dig in without listening, which is its own problem. The answer you want is the third one — someone who holds the position, explains the reasoning, and changes their mind when you produce a better argument.
He has said the reason we stood out was that we were doing something structurally different, not that we were cheaper or louder. I will take that, but I want to be precise about what the difference actually was, because "we're different" is what every agency says.
Thirteen stages before anybody designed a page
Before a single page got built, we mapped the whole business: every service area, every brand they carry, the Google Business Profile footprint, and the customer profiles they actually serve. Thirteen stages of onboarding.
Owners hate this part. It feels like paperwork standing between them and the thing they paid for. But a website is a model of a business, and you cannot model something you have not described. Skip the mapping and you get a pretty brochure that contradicts the company behind it — which is most trades websites, and a large part of why they gate everything the customer came for.
A website is a model of a business. You cannot model something you have not described.
The Learning Center had two audiences from day one
The content engine was built to do two jobs at once: rank in Google and be legible to AI systems. Those used to be the same job. They are not anymore.
Google rewards a page that answers a query. Answer engines — ChatGPT, Gemini, Claude — reward something closer to structure: clear entities, stated relationships, claims a machine can lift without guessing what you meant. Write only for the first and you become invisible in the second. Most contractors are about to discover this the expensive way.
The tools are where it stopped being a website
We built custom interactive pieces into the site: ductwork diagrams, side-by-side product comparisons, a generator sizing quiz, an electrical load calculator, and a persona experience that changes what the page shows based on who is reading it.
Every one of those started as a marketing idea. Then something happened that no one designed for.
Vince closed a sale from the website, standing in a driveway
A technician named Vince pulled the site up in the field and used it to help close a job on the spot. Not a leave-behind. Not a brochure he handed over. The live website, working as a sales tool in a customer's driveway.
That is the moment the category changed. A marketing asset is something you point strangers at and hope. An operating tool is something your own people reach for because it is the fastest way to do their job. The second one is worth vastly more, and almost nobody builds for it, because the brief always says "website."
From there it kept going. The site became a training resource for technicians. New hires learn from the same pages that rank for homeowners, which means the explanation the customer reads and the explanation the tech gives are finally the same explanation.
A marketing asset is something you point strangers at and hope. An operating tool is something your own people reach for because it is the fastest way to do their job.
Not a project — a living thing
Cory describes the site as a living organism rather than a one-time build, and that framing is the one I would steal if I were an owner reading this.
The traditional website is a capital expense. You spend a large amount once, you get a thing, the thing decays, and four years later you do it again while complaining about the last agency. A living site is closer to equipment: it gets maintained, extended, and used daily, and it returns more the longer you own it. The build sessions ran late and shipped fast, in cycles, the way software gets built — because that is what it is.
Then they gave the whole team AI, and it got messy
This is my favorite part, because it is the part most case studies leave out. Eco rolled AI tools out across the team and there was a stretch of genuine chaos before it settled.
That is not a failure of the rollout. That is the rollout. Every company that has actually put these tools in front of people has lived through the same window: unclear ownership, inconsistent output, some people sprinting and others quietly opting out. The companies that win are the ones that treat that mess as a phase to work through rather than proof it was a bad idea. The ones that lose pull the plug in week three and tell themselves AI is overhyped.
Which is the same shape as the three hands I saw go up in 2008. The advantage never goes to the company with the best tools. It goes to the company willing to be bad at something new for a quarter.
What I would take from this if I ran a trades company
Three things, and none of them require an agency.
Push back on whoever is selling to you, deliberately, and watch what they do with it. Insist on being mapped before you are designed — if nobody asks about your service areas, your brands, and your customers before showing you a homepage, they are decorating. And ask, about anything you are about to buy, whether your own team would reach for it. If the answer is no, you are buying a brochure, however good it looks.
Cory is in Hands Up, in the clients' own words, and he says something there I think about often: that the best part of the partnership is that both companies are building toward the same future. That is the whole thing. You cannot buy that arrangement. You can only choose people who are already pointed the same direction you are.
Hands up.