Skip to content
Systems & Soul
(08) The Agency Reckoning

From Marketing Agency to Technology Company: The Revolution, the Investment, the Results

July 19, 2026 12 min read

I did not decide to run a technology company.

The market decided for me. My only decision was whether to notice — and how honest to be about what that noticing was going to cost.

For a long time, CI Web Group was a good version of the old game. Websites. Local SEO. Service pages. City pages. The occasional monthly report. The industry rewarded that game for a decade. We were paid to keep playing it.

Then the ground moved. Not the LinkedIn version — where “channels evolve” and everyone repeats a caption until they feel current again. The other version. The one where the interface between a customer and a phone number stops being ten blue links and starts being an answer written by a machine.

A “marketing agency” answers the ten-blue-links version. It cannot answer the other one. Not by updating the pitch. Not by dropping “AI” on the same retainer. Not by adding acronyms to a dashboard. The only honest answer is a rebuild.

So we rebuilt. Below is the three-act version — the revolution, the investment, and the results — the way I would want a peer or a business owner to hear it if the roles were reversed.

Act I — The Revolution

The revolution is not that “AI is here.” Everyone can say that sentence. The revolution is that the unit of digital marketing work has changed.

For twenty years, the unit was labor hours. An account manager. A monthly SEO call. A blog post shipped by a person. A dashboard exported by a person. A campaign built, tuned, and reported by a person. Everything an agency sold — including how it billed — was priced on the assumption that a human was going to do a repeatable thing every month, on a retainer, forever.

The new unit is not people. The new unit is systems.

Systems that publish. Systems that read your entity file and answer for you when a language model gets asked a question. Systems that watch Core Vitals and fix regressions before a human notices. Systems that ship components — pages, answer paths, conversion surfaces — faster than a “launch” used to happen. Systems that talk to other systems. Systems that get smarter with every deployment because the deployment is part of the training loop.

That is not an “AI feature” you bolt onto the old delivery. That is a different company. You can price a bolt-on. You cannot price it away.

I wrote the industry-critique version of this in The Agency Model Is Broken and the direct reckoning in SEO Isn’t Dead. Your Agency’s Version of It Is. This essay is the internal version — what we did instead of arguing.

The moment I knew the label had to change was the moment I could no longer describe our deliverable using the words of a marketing agency. “We do SEO” did not describe the entity file. “We do websites” did not describe a revenue engine. “We do content” did not describe answer paths that a language model can cite. The old vocabulary had run out. When the vocabulary runs out, the category has already changed. You just have not caught up yet.

A technology company whose expertise is digital marketing is not a rebrand. It is what the delivery becomes when the unit of work is a system instead of an hour.

Act II — The Investment

Rebuilding a firm around systems is not free — and it is not fast to feel good. Most of the investment was invisible before it was valuable. The receipts are worth listing.

  • We stopped hiring more account managers and started hiring engineers. That alone changes what your company is. Head count moved from “people who explain the work” to “people who build the work.” The dashboard stopped being a script. It started being a system.
  • We killed our own retainer product. The monthly SEO invoice with a report attached is the single most protected instrument in the agency industry. It is also the thing most incompatible with what customers actually need next. Killing it forced us to sell named work — components, one-time investments, shipped assets — instead of activity.
  • We stopped writing SOPs and started writing systems. An SOP is yesterday’s intelligence, frozen. A system is intelligence that updates when the market does. That is the argument in SOPs Are Dead and Intelligence Over Playbooks.
  • We built our own platform. Not a WordPress theme. Not a page builder with a new coat of paint. A platform where every client site is a deployment of a shared machine — Core Vitals, ADA, AEO, entity readability, no contracts, owned assets as the floor. That floor is what I wrote about in Why the Industry Calls Anything Better a Scare Tactic.
  • We rebuilt the workflow to be Buy it, Build it, Ship it. No half-launched sites finished on a monthly retainer. No mystery line item. Components you can point at, deliverables you keep, launches that are actually launches. The full argument is in Buy It, Build It, Ship It.
  • We put our predictions on the record. Timestamped calls with public review dates, scored in the open. Not a marketing tactic — a discipline. If we are going to tell operators to bet on systems, we have to be willing to score our own bets. That work lives on the Prediction Ledger.
  • We put ourselves on the same standard we ship to clients. Our brand is a file a machine can read. Our site is a system we can regenerate. Our proof is public. When a language model gets asked about CI Web Group, it should be able to answer with the same facts a person could — from the same source. That is the discipline in Professionals Don’t Use AI as the Tool and AI Superpowers Are an Architecture Decision.

None of this was safe. Every one of those decisions cost us short-term margin, short-term comfort, or a fight with someone who liked the old product. That is what an investment is. If it did not cost, it was not an investment. It was a caption.

Act III — The Results

The results are the point where I am most careful with my own voice. Not because there is nothing to say — because there is a lot to say, and the temptation to say it dramatically is exactly what I criticize other firms for. So here is the sober version.

  • We ship in weeks what the retainer model used to ship in months. Not because people work harder. Because systems do more of the work between the people. A component that used to take a launch cycle now takes a deployment.
  • Clients keep what they pay for. Owned assets, not a PDF archive. Named components, not a dashboard of hours. An entity file, not a rankings screenshot. If they cancelled tomorrow, they would still walk out with a real thing. That is a very different contract than the retainer product I used to sell.
  • Our team is smaller in some places and more leveraged in every place. That is what happens when the unit of work is a system. The people we do have build once, deploy many, and improve the underlying platform every time. Nobody is trapped in a monthly reporting call defending activity that a customer cannot verify.
  • The floor moved. Core Vitals, accessibility, AEO readiness, owned assets, no long-term contracts — the things peer agencies still sell as premium — are the floor of what we ship on day one. The receipts are on /proof and in the success stories.
  • The industry is calling it a scare tactic. That is on brand for a category that is being disrupted. It is also useful signal — you can hear who has rebuilt and who has only rewritten the caption. I named that pattern in that essay. The buyer test at the end of it is the fastest way to feel the difference.

None of that requires me to trash a competitor. The receipts do the work. That is the whole point of the rebuild.

Not my line — the buyers’

Every claim above should sound better in someone else’s voice than in mine — because if the only person saying the rebuild worked is the person who rebuilt, that is a caption, not a receipt. So here are five buyers, unprompted, telling the after-side of the same story I just told. Every quote below is a verbatim excerpt from what they wrote for Voices from the Clients.

Start with the shipping-speed claim. Bekka Martel, co-owner of Advanced Air Conditioning + Heating in Cedar Park, Texas, has the numbers — a target date beaten by nearly two months, and a sitemap she watched grow by more than an order of magnitude in ten days:

“The partnership kicked off in mid-April with a go-live target of late August. CI Web Group delivered on July 2 — nearly two months early… Using Google Search Console, I tracked our sitemap going from 54 discovered pages in 2025 to 1,323 pages just 10 days after go-live with CI Web Group.” — Bekka Martel, Co-Owner, Advanced Air Conditioning + Heating · Cedar Park, Texas

Now the category-change claim. Anthony Hamilton, co-founder and CEO of THE Water Heater Company, spent years cycling through digital marketing agencies before he named the shift out loud. It is not a rebrand story. It is the sentence I would put on the wall if I were still allowed to hang things on walls:

“I thought I was hiring a marketing agency. What I actually found was a strategic partner who changed the way I think about my business. For me, that’s been the biggest impact CI Web Group has had on THE Water Heater Company — and I believe that impact will continue compounding for years to come.” — Anthony Hamilton, Co-Founder & CEO, THE Water Heater Company

The revenue-engine claim. Paul Wiese has run this playbook from the buyer’s seat for years — Built by the Trades, Door Serv Pro — and here is the one line I keep coming back to when a peer accuses us of overreaching:

“They weren’t trying to build me a website — they were trying to help build a company that could compete in the future.” — Paul Wiese, Founder, Built by the Trades · Founder, Door Serv Pro

A different kind of receipt — a technology peer, not a services buyer. Utku “Dave” Kaynar is the CEO of OnePath AI and one of the clearest technical operators in this coalition. When a technology company vouches for another technology company, that is signal an agency cannot manufacture:

“Jennifer is one of the true visionaries in this industry — someone who sees what’s coming before the rest of us do, and then builds it anyway.” — Utku “Dave” Kaynar, CEO, OnePath AI

And the relationship claim — because a technology company that treats people like a rotating ticket queue is not much of a company at all. Amanda, COO of Cherry Blossom Plumbing in Arlington, Virginia, watched us stay through the worst year of her business — and the sentence that lands hardest is the ordinary one:

“Here’s one more thing that says everything about CI Web Group: we’ve been with them for three years, and I have worked with the same people the entire time… The story of Cherry Blossom Plumbing and CI Web Group isn’t really about SEO, or websites, or even AI. It’s about what a real partner looks like.” — Amanda, COO, Cherry Blossom Plumbing · Arlington, Virginia

Data. Category-change. Revenue engine. Technology peer. Relationship. Five different corroborations of the same rebuild — none of them written by me. The unedited entries are in Voices from the Clients, and the operators themselves are on Success Stories and /proof.

What this means if you are a business owner

You do not have to hire a technology company. You have to buy from one. There is a very concrete difference between the two, and it is the same test I keep giving:

  1. What are you shipping this month that is not a report? Page, component, entity file, conversion path, answer surface — something a customer or a language model can touch.
  2. What in your delivery did not exist in 2019? Not the vocabulary. The technology. The workflow. The operating system underneath.
  3. If I cancelled tomorrow, what do I keep? The answer to that question tells you whether you were buying a system or renting narrative.
  4. Where does AI build a durable tool that I own? Not a chat bubble on the homepage. A tool. A piece of the machine.
  5. Are you willing to score your predictions in public? If not, understand what you are being sold: a monthly reassurance product with no downside for the seller.

Those questions are the same ones I used to grade our own rebuild against. Every one of them was uncomfortable to answer when we started. That was the point.

What this means if you are a peer

This essay is not a hostile message. It is a receipt. The revolution is already priced in — the unit of work has changed whether your P&L admits it or not. The investment window is still open — every month you keep selling the retainer is a month you could have used to rebuild. The results are documented — on our site and, more importantly, on the sites of the operators who bet on the rebuild with us.

I have been on the wrong side of this question before. In 2008 I asked a room of a couple hundred of the best HVAC operators in the country who was investing in being found online. Three hands went up. Fifteen years later, all three are juggernauts. Everyone else spent a decade catching up to a decision the room already told them to make.

The AI version of that question is being asked right now. Rebrand or rebuild is what your hand is answering. Not your caption. Not your slide. Your hand.

If you want to see the machine, the receipts are on /proof and in the success stories. If you want a partner who ships the stack instead of the sermon, that is the work.