Case study · RamClean
RamClean is a commercial cleaning and facilities company doing roughly $15 million a year, with clients including Amazon, Ford, Mercedes-Benz, and Chick-fil-A. It made the Inc. 5000 in 2025 on 310% growth. Its entire marketing department is two people. This is how that works.
A company growing that fast produces marketing work faster than it can hire people to do it.
Every new city is a page that doesn't exist. Every new service line is an ad group, a landing page, and a set of keywords nobody has researched. Every quarter the CRM drifts a little further from reality, and the spreadsheets that run operations multiply quietly in someone's Drive.
The usual answer is to add headcount linearly with growth, or to hire an agency and hope. RamClean did neither. The department stayed at two people and built systems instead — each one designed so the work happens on a schedule, with a human approving rather than doing.
Deal size compares pre-2021 to the 2022–2025 period, following the HubSpot deployment. Organic bid value and page count are trailing figures from internal reporting. The same page system also produced 30+ inbound form leads in a single 90-day window.
01
Ranking in every city a company serves normally costs a fortune in content. We built a pipeline that generates the copy, the images, and the schema for a city-level page, validated against a written content standard, and exports as ready-to-import XML. It went through three build iterations — hand-built, then a batch generator, then a reusable skill that works across service verticals.
200+ pages · ~$600K organic bid value · 30+ leads in 90 days02
Full HubSpot deployment — deal pipeline, workflows, dashboards, always-on reporting, and a bid strategy built around maximizing deal value rather than deal count. On top of it sits a qualification engine that reads a contact record, fills in the sales qualification fields from available data, and scans the pipeline for dormant leads worth another call.
Average deal size up 137%03
A rebuilt Performance Max structure, standing negative-keyword infrastructure across six lists with a match-type rulebook, a repeatable weekly search-term audit, and a semiannual review of every ad headline against performance criteria. Consolidating all-time analytics, Search Console, and CRM deal data surfaced an entire high-value service segment that nobody had been bidding on.
Weekly hygiene, without a weekly meeting04
Operational data lived in fragmented Google Sheets that disagreed with each other. We built an eleven-table Airtable system with automated payroll import pipelines, per-account margin calculation, a QA workflow run with the person who actually knows the data, and five interfaces built for five different roles — so nobody has to learn a database to use it.
Contract and site data consolidated and queryable05
A growing library of small tools that delete specific busy-work: a QuickBooks supplies extractor that turns a day of sorting into seconds, a bulk import app for scopes of work, monthly local-profile audits, content planning, press releases, job postings. Most of them aren't "AI" to the person using them — they're just tools. AI is how they get built quickly.
Hosted internally, used across departmentsThe systems matter less than the method. This part is the same every time, and it's the reason none of it fell over when the person who built it was busy with something else.
Not where people say they go. Sit with the work, watch the process, count the steps. Most of what looks like a strategy problem is three people doing the same manual task in slightly different ways.
Not a platform. The specific tool that takes the specific worst part of someone's week and makes it stop.
Nothing writes to a live ad account, CRM, or website without a person reviewing a plain-language summary and approving it. The system proposes; the human decides. This is not a compromise — it's what makes the thing trustworthy enough to actually use.
Every system ships with an SOP and an onboarding doc. If it only lives in one person's head, it isn't finished.
The test isn't whether it works. It's whether it works when somebody who didn't build it is the one running it on a Tuesday.
If it only works when I'm the one doing it, I've built myself a job, not a system.
That last step is the one most people skip, and it's the one that decides whether the work survives. A marketing system that requires its author to be in the room is a retainer with extra steps. The point is to leave behind something the team owns.
The department covers advertising, CRM, web development, social, email, data systems, documentation, and internal software. It hired and onboarded the company's first dedicated content role with a full system behind it — eight written SOPs, an analytics workflow, and a planning cadence — so the role worked from week one instead of month four.
Every decision and task is captured in a Notion workspace built for the purpose, updated daily from meeting transcripts and daily work. Anyone can ask what marketing is doing and get a real answer, including six months later. That repository is also what made this case study possible to write.
Client work behind the numbers:
Amazon · Ford · Mercedes-Benz · Buffalo Trace · Chick-fil-A · Sam Adams · Four Roses · Hyundai Transys · YMCA · Boys & Girls Club
The pattern transfers. A business with more locations than marketing people, work that only happens when somebody remembers, and systems that live in one person's head — that's the same problem in a different industry.
I build these one at a time, for a small number of clients, and I hand them over documented.