Cleaning and maintenance: job sheets that arrive late

It's Friday afternoon, you're pulling the month's invoicing together, and six job sheets are missing. One is on the phone of a supervisor who was off today. One is a photo of a crumpled sheet of paper, buried in a WhatsApp group with two hundred messages. Two did arrive, but they say "all fine" and nothing else. If you run a cleaning company, a maintenance firm, or any service delivered on someone else's premises, this is about why job sheets turn up late and half-empty — and which part of that mess an AI agent can chase for you.

A job sheet isn't paperwork. It's your invoice and your alibi

When the work happens on the client's site, the job sheet is the only proof it happened at all. Three things hang off it: what you can invoice, what you can defend when there's a complaint, and what you can show when the contract comes up for renewal.

The fixed part rarely goes missing. The contract says so many hours a month across so many sites, and you invoice that almost on autopilot. What leaks away is the extras. A machine you had to bring in. A window cleaned because it was in a state. Two hours outside the normal slot because the client wanted it done before a visit. Consumables replaced. None of that gets billed unless someone wrote it down the same day. Run the numbers on your own business: if three extras of thirty or forty euros slip through every week, you're not losing a hundred euros — you're losing five thousand a year, and it's five thousand of clean margin, without winning a single new client.

Then there's the other side, the one that stings more even though it never shows up in the accounts: the building manager who calls to say the entrance wasn't cleaned on Thursday. With a job sheet that has a time, a name and a photo, that conversation lasts two minutes. Without one, it's your word against theirs, and it usually ends with a service given away for free.

Why they arrive late and thin (your team isn't slacking)

When job sheets don't come in, the instinct is to push harder. That's almost never the problem. It's usually four things, and all four are design faults:

• The timing. The sheet gets filled in at the end, when the trolley is packed away, you're tired, and the next site or the bus is waiting. Any friction at that moment wins.
• The format. A two-page PDF or a fifteen-column spreadsheet is fine on a desktop and impossible on a phone with one hand full.
• The channel. Some send a photo to the group, some ring the supervisor, some scribble it in a notebook and hand it in on Monday. With three channels, one always goes missing.
• The blank. "All fine" is what an open-ended question gets you. Nobody knows exactly what you're after, so they answer the minimum that ends the exchange.

Notice that none of the four gets fixed by more reminders at the Monday meeting.

Collecting a job sheet without rolling out a new app

The piece that actually changes the outcome is simple: the sheet shouldn't wait for someone to remember it — someone should ask for it at the right moment. That someone can be an agent on the WhatsApp your team already uses.

1. Send the first message. At the expected finish time, the agent messages whoever did the job. It isn't a form sitting there waiting; it's a question that lands while the person is still on site.
2. Three questions, not fifteen. Are you done? Anything you couldn't do, and why? Any extras or issues? That covers ninety percent of jobs. The odd field that only applies once a month gets asked only when it applies.
3. Let photos and voice notes count. People who work on their feet don't type paragraphs. A twenty-second voice note or a photo of the fuse board is faster and more reliable, and the agent can transcribe it into clean text.
4. Make it follow up. If the answer is "all good" but that site had an open issue from last week, the agent asks about it. That one follow-up question is the difference between a job sheet and a useful job sheet.
5. File it where it belongs. The closed sheet goes to that site's record with a date, a time and a name; extras go to a separate list, the one you invoice from; issues go to another, the one you review. If it all lands back in the same WhatsApp group, you've gained nothing.

None of this asks your team to learn new software, and that's half the battle won. Cleaning and maintenance run on shift work, split hours, high turnover and a wild variety of phones. Anything that needs a half-hour training session per person dies quietly.

What you get once sheets land the same day

The real prize here isn't "tidy paperwork".

You invoice extras within the week, while the client still remembers asking for them — arguing about an extra two months later is half an argument already lost. You answer a complaint with facts: "Thursday the 4th, 20:40 to 22:15, here's the photo." You spot sites going wrong, because three issues in a fortnight at the same address is a warning, and you get there before the building manager rings. And when renewal comes round, you're not turning up with goodwill and a memory: you're turning up with twelve months of logged services, which is a different conversation when the client has three other quotes on the desk.

A quick test for whether this applies to you: look at your last month-end and count the minutes you spent chasing job sheets, and the extras you actually billed. If the first number is high and the second is close to zero, you don't have an invoicing problem. You have a collection problem.

When not to build this

A couple of honest caveats, because this doesn't suit everyone.

If you're three people at one site, don't build anything. The WhatsApp group you already have works fine, and adding process will only cost you time. This starts paying off when there are several crews, several sites, and someone who can't be at all of them.

If your client makes you use their portal — common in facility management and public-sector contracts — the agent can collect the sheet and hand it to you ready to go, but someone still has to key it in over there. That's a genuine improvement, but don't sell yourself an integration you haven't got.

Be careful, too, with sheets that need a client signature on site, or with anything your health-and-safety obligations specify. Before you replace a signed document with a message, check it. A well-built digital job sheet usually stands up, but that's for your advisor to confirm, not an article.

The most important caveat is cultural: this is a collection tool, not a tracker. An agent that asks once, at the end of the job, is a colleague who saves you the notebook. One that asks where you are every half hour is a foreman — and with a foreman, people quickly learn to answer whatever gets them left alone. At that point you've automated distrust, which is precisely the opposite of the point.

Start with your three worst sites

Don't redesign the whole operation. Take the three sites that cause you the most grief, define a minimum job sheet in five fields, and run it for a fortnight. Measure three things only: what share of sheets arrives the same day, how many extras got logged, and how many times you had to chase someone. Two weeks is enough to know whether this works in your business.

That background grind — asking at the right time, following up when something's missing, filing each thing on the right list — is exactly what an operations agent does. It doesn't replace your supervisor; it takes the chasing off them. You approve what gets invoiced; it makes sure nothing is missing on Friday.