How a Commercial Cleaner Grew Inside the Buildings It Already Cleaned

An Owner Who Believed He Had a Bidding Problem

A regional janitorial company: a little over thirty nightly contracts, mostly multi-tenant office buildings and a handful of medical office suites, five crews working after six in the evening, and an owner who still did every walkthrough himself. Retention was good. Accounts stayed for years. But growth had stalled, and his explanation was the obvious one. He was not getting invited to enough bids.

So he asked us for what most owners in this trade ask for: more requests for proposal, more walkthroughs, a website that would put him on more shortlists. We did look at that. Then we asked to see a year of invoices broken out by type, and a different picture appeared. Nightly contract revenue was steady. Everything else, the floor refinishing, carpet extraction, window work and post-construction cleans, was a thin line that had been shrinking for two years.

Three earlier pages here deal with winning the contract in the first place, and this story deliberately starts after it is signed. Getting onto the bid list and writing scope pages is in SEO for commercial cleaning companies. Vendor qualification and what a site should cost is in the commercial cleaning website cost guide. Retention, complaints and service records are covered on our AI for commercial cleaning page.

The Night Crew Sees the Next Job First

We asked one of the supervisors what her crew noticed in a typical week that was not part of the nightly scope. She did not need to think about it. A traffic lane worn grey in the carpet outside the elevator bank on four. Vinyl tile in a break room that no amount of mopping would bring back. Dust building on the return vents above a reception desk. Grout in a lobby restroom turning dark. Fingerprints at child height on the glass of a pediatric suite that the window contractor reached twice a year.

None of it was written down anywhere. There was a paper notebook in each janitor’s closet that occasionally reached the office. The owner, who did walkthroughs during the day, saw the buildings at their tidiest and their most occupied. The people who knew the true condition of every floor in his portfolio were working when he was asleep, and nothing connected what they saw to anyone who could sell it.

Who got the work instead

The facility manager did eventually notice. Usually a tenant mentioned the carpet, or a landlord inspection flagged the floors before a lease renewal. At that point the manager needed the problem fixed, searched for a carpet or floor care company, and hired whoever answered first. That specialist then had keys, a reason to walk the building and an introduction to the person who signs the cleaning contract. In two of the accounts behind this composite, the specialist who came in for the floors later submitted a bid for the nightly work too.

That was the uncomfortable finding. The company was not only failing to sell periodic work. By leaving it unsold, it was inviting competitors into its own buildings, one maintenance problem at a time.

The Contract Is a Sample of the Portfolio

The second thing the invoice review surfaced was who the company actually answered to. On paper it had thirty-odd clients. In practice, many of those buildings were run by a much smaller number of people: property managers at management firms who looked after six, ten or fifteen buildings each, and facility managers at companies with several locations. The company cleaned one of their buildings. Somebody else cleaned the rest.

Seen that way, every contract is a sample. The manager watches how you handle one building for a year, and that year is the most persuasive proposal you will ever write for the others. It is also an audition nobody had told the owner he was giving. He had never asked any of those managers what else they ran, and his website had nothing that spoke to someone thinking about adding a second building to a cleaner they already used.

The buyer changes jobs; the building stays

There was a quieter version of the same problem. Facility and property managers move between employers more often than buildings change hands. When the manager who hired you leaves, a stranger inherits your contract and often re-bids it in the first year simply to put their own stamp on the vendor list. When that same manager starts somewhere new, they inherit a cleaner they did not choose, and they will call the company they remember, if they can remember its name. Looking back over two years, the owner counted several contacts who had moved on. He had stayed in touch with none of them.

Reading the Account List Instead of the Bid List

Before anything was built, the owner and his office manager sat down with the contact list and answered, account by account, a few questions nobody had asked before. Who here signs, and who actually decides? Does that person manage other buildings, and do we clean any of them? When was the carpet last extracted and the hard floor last refinished, and by whom? Has the contact changed since we signed?

Most of the answers were “we don’t know,” which was useful in itself. The rest showed that the company’s best growth opportunity was not a new market or a new service. It was a list of buildings that it already had keys to, and a smaller list of people who already trusted it with one property.

The Build, Starting in the Janitor’s Closet

The order mattered. The first two pieces touched nothing on the public website at all.

One: an observation log a cleaner can use in under a minute

Every crew member got a short form on their phone: building, floor or room, one or two photos, a few words, in English or Spanish. It replaced the closet notebook. Submissions land in a list for that building’s shift supervisor, who looks at the spot on the next visit and either confirms it, adds a note on size and condition, or discards it. Nothing leaves the company unconfirmed.

Two: a scope note the facility manager can approve by reply

A confirmed observation becomes a one-page note. AI drafts it from the photos and the supervisor’s words: what was seen, where, roughly how large an area, what the work involves and how it can be scheduled around the tenants. The estimator prices it by hand and edits the draft before it goes out. The manager can approve it, ask for a different date, or say not this quarter, and a not-this-quarter answer is simply raised again when the quarter turns. It is ordinary business automation, with a person at both ends.

Three: a periodic calendar proposed at signing

New contracts now include a suggested schedule for the work that is not nightly: hard floors refinished on a stated cycle, carpets extracted by traffic level, vents and high surfaces on their own rotation. The manager does not have to accept it. But it puts the periodic work in the first conversation instead of leaving it for the day a tenant complains, and it gives the observation log a baseline to compare against.

Four: pages for the manager who already has a cleaner

Only then did the public site change, as part of an AI website build. The company added a page for each periodic and specialty service, written for a manager whose nightly cleaning is already handled by someone: how the work is scheduled after hours, how long a floor is out of use, what happens to furniture, what the crew needs from building engineering. It added a page called “Adding a building,” describing how a second site is brought on, what the first month looks like and how one point of contact covers both. Both are written plainly enough for an assistant to repeat accurately, which is the heart of our answer engine optimization work.

Five: a record of the people, not just the buildings

Each account now names its decision maker, the other properties that person manages, and the date the contact was last confirmed. When an email bounces or a manager announces a move, the owner gets a task, not a message sent in his name. He makes that call himself. The new manager at the old building receives a short history of the account and an offer of a walkthrough; the departing one gets a personal note from the owner.

Six: new-account search, last

The search work the owner first asked for came at the end. By then the specialty pages had a second job, reaching facility managers searching for floor or carpet work in buildings the company did not yet clean, which is a cheaper way into a building than a full nightly bid.

Rules the Supervisors Set

Crews never raise extra work with tenants or occupants; the only audience is the person who manages the building. No observation is sent without a supervisor having seen it in person. No more than a couple of notes a month go to any one building, so a report never turns into a stream of sales mail. AI drafts the wording and nothing else. It does not price, does not promise dates and does not decide whether something is worth raising.

Photos show surfaces only. No desks with papers on them, no screens, no people, and in medical suites nothing from exam rooms or any area where patient information might be visible; the medical clients’ own privacy rules set that boundary and the company built to it. Notes never comment on other vendors in the building. And the contact record holds business details the managers have shared directly. Nobody is scraped, and no outreach goes out automatically to someone who has changed employers.

How It Moved

Periodic work moved first, within weeks, because it needed no new buyer and no bid. The manager already trusted the company and already had the problem; the note just arrived before the tenant complaint did. Over the following months the periodic calendars reduced the rushed calls that used to follow an inspection or a lease showing.

Second buildings came more slowly, because management firms change vendors on their own budget and contract cycles, and the conversations were measured in quarters. New accounts from search moved on the usual curve described in how long SEO takes to work. All of this is the shape of the change, not a number anyone audited.

Count the Buildings Behind Each Account

Commercial cleaning is sold as a contract for one building, and owners understandably measure growth in contracts. But the people who buy it manage portfolios and change jobs, and the work they need beyond the nightly scope is visible every evening to the crew that is already inside. A cleaning company that writes down what its crews see, and knows who else each manager is responsible for, has a sales pipeline it has been walking past for years.

Claude Code is what let the observation form, the drafted scope notes and the new service pages go live in weeks while the supervisors kept the quality-control tool they already trusted. If your account list looks like this one, our AI for commercial cleaning page covers the wider system.