Sunday evening, laptop on the arm of the sofa, scrolling right through eleven months of columns to work out whether 14 Baddow Road paid in June. That is usually the moment a window cleaning round spreadsheet stops being worth it. Not a crash. Not a lost file. Just the scrolling, every week, for the rest of your working life.
One thing first. Paper, memory and Google Sheets are all live working methods in 2026. A round book in the van does the job. So does a sheet, up to a point. None of this is an argument that your system is daft. It is what breaks when you move a round off a spreadsheet, and what to sort out first.
A window cleaning round spreadsheet does not break, it goes stale
The fear people have is corruption. One wrong cell, a formula dragged down wrong, and half a street thinks the clean is next week. That does happen. It is not the real problem.
You cannot update it at the door
The real problem is that you cannot update a spreadsheet at the door. Wet hands, phone in your pocket, and marking that clean off means opening the app, pinching, zooming, scrolling sideways to the August column and tapping one cell without catching the one below it. So you don't. You tell yourself you will do it tonight. Some nights you do.
Phone editing gets named by two separate trade sources as the number one reason sheets fail. What fails is freshness, not accuracy. Every figure in your sheet is correct. It is just four days out of date, and you are making Tuesday's decisions off it. Because the sheet is awkward on a phone, the schedule then drifts into the calendar app and the money into a notebook, and neither knows about the other.
Around 80 to 100 customers is where it tips
One UK trade source puts the pain threshold at around 80 to 100 customers, and forum behaviour backs that up. Under that, a spreadsheet is genuinely fine. Over it, payment tracking goes wrong first.
The order to do it in
The whole job in five steps. The rest of this post is what goes wrong inside them.
- Export customers, jobs, balances and notes out of your sheet.
- Keep an untouched copy of the original somewhere you will not tidy up by accident. Never work on it.
- Import into a test or trial account first, not your live one.
- Check four things and only four: record counts, next-job dates, prices, and debt totals last, because a balance that arrives wrong is the one error that costs cash rather than time.
- Run both for one full cycle. Four weeks with the sheet as a backup is how you find the customer who came out with no price.
Those four are exactly what breaks, and how you take payment decides what shape the owing list needs to be in.
One row per customer, one row per clean
Step one is where people come unstuck, and nobody mentions it until they are halfway through. The export is not a copy, it is a reshape.
Your sheet is one of two shapes
Most hand-built sheets are a grid. Name, address, price down the left. A column for each month across the top. A Y or an N in the cells. In this post, Y means the clean happened and N means it did not. If your Y means paid instead, the reshape is identical, you just fill a different field at the far end. The other shape is a running log: one row per clean, a real date column, an amount, often a paid column.
Almost every destination, meaning the software you are moving to, wants that log shape split in two. A customer record holding what rarely changes: name, address, phone, price, frequency. Then a job history, one row per clean: date, amount, paid or not. The month columns are the awkward part, because they are not really columns. They are events.
One customer, reshaped
| The piece | What your sheet has | What the system wants |
|---|---|---|
| The customer | Smith, 14 Baddow Rd, £14 | Customer: Smith, 14 Baddow Rd, £14, 4-weekly |
| The history | Jan Y, Feb Y, Mar N | Job: 15/01/2026, £14. Job: 12/02/2026, £14. Nothing for March. |
Two things there are not in your sheet. The first is 4-weekly: your grid probably holds the frequency nowhere, it is in the tab name or in your head. Either way it becomes one field you type once per customer, and it drives the next due date. The second is the day of the month.
Where the date comes from
A column headed March tells you the month, not the day. There is no honest way to get 15/01/2026 out of a cell that says Y under January, so you have two choices, and neither is clever.
Approximate: put the day you do that street, or the middle of the month, and accept that every historic date is roughly right rather than right. Fine as a record of what happened, not fine for anything an accountant sees.
Or drop the history.
Leave the old history behind
Bring the customer records, the last-done date and the outstanding balance, and let the new system build its own history from your next clean on. What you need on day one is who is due, what they pay and who owes you. Keep the old sheet as the record of everything before that; it is worth something if you ever sell the round.
The other option, if you want the history in the software, is to unpivot the grid: a second sheet with one row for every cell that says the clean happened, a Y becoming a row and an N becoming nothing. Customer, date, price, paid or not. For 300 customers and three years of columns that is thousands of rows, a rainy Tuesday rather than an evening.
Before you book that Tuesday: bulk import as a paid or assisted service exists in the UK market, turned round in about 24 hours in some cases, so the reshape need not be your job.
Dates written like a person, not like a computer
This one is for sheets with a date column. On a pure grid it bites later, when you type real dates into that second sheet.
Sort your date column before you do anything else. Anything that leaps to the top or the bottom in the wrong order is text pretending to be a date.
Sheets are full of it. "Last done 3rd wk March." "Due end of month." "12/4" with no year, which was obvious when you typed it and is not now. All of that imports as text, and what comes out is a blank next-job date or a wrong one. Wrong is worse, because you will not notice until you turn up at a street that was done a fortnight ago.
Fix it in the sheet, not after. One column, real dates, one format, UK order. A boring hour that removes the single most common breakage in the job.
Colour is not data, and neither is a tab name
Your round grouping almost certainly lives somewhere that cannot travel. A tab called "Mon week 1". A tab called "Tues, Kirkby". Or no tab at all, just yellow fill for one area and blue for another, which made perfect sense when you set it up.
Export that to CSV and the colours are gone. Nothing warns you. Names, addresses and prices arrive looking perfect, and the grouping you have carried in your head for six years quietly does not.
Add a column. Call it Round. Type the name in for every customer, the same name you say out loud. Half of window cleaning round management is really just this: knowing which names belong together on which day.
Move the notes column across whole
Key safe code. Dog in the back garden. Do not do the conservatory roof. Pays cash, leaves it in the porch. Gate padlocked, knock loud, she works Thursdays.
That column is the most valuable and least structured thing you own, and it has no obvious destination field anywhere, which is why every migration checklist names it separately, and why people lose it. Move it across whole and untouched, even if it lands somewhere untidy. Losing it costs you a year of learning the round again.
What people are actually frightened of
The blocker is almost never money. Round software in the UK generally runs about £15 a month, and the same round costs two evenings a month to keep alive on a sheet. Read enough threads and the same three worries come up.
- Will the addresses still be there if I lose the phone. The answer that settles it is cloud storage plus browser access, so a replacement handset or a borrowed laptop gets you back in. Test that in week one, on a different device.
- Will it be too complicated. Fair worry, and plenty say purpose-built software looks like more than they need. Judge it on whether you can add a customer and mark a clean done in under a minute, on a phone, in the rain.
- What if the software vanishes. The mitigation the trade converges on is being able to export your own data back out, in a file you can open. Do that during the free trial, while nothing depends on it.
Making Tax Digital bans retyping, not spreadsheets
This one bites sooner than most people assume. Making Tax Digital for Income Tax is phased in by qualifying income, meaning gross income before expenses rather than profit. Sole traders over £50,000 come in from April 2026, over £30,000 from April 2027, and over £20,000 from April 2028. A £1,000-a-week round is £1,000 times 52 weeks, which is £52,000 a year, so a large one-van round is in scope now. A round of 200-odd customers is nearer £28,000 and does not come in until April 2028. Work your own gross out rather than assuming. It means digital records, an update to HMRC at least every three months, and a final declaration by 31 January.
Spreadsheets are still permitted. What is not permitted is manually copying and pasting figures between systems, so a sheet has to be connected through bridging software rather than retyped each quarter.
Round software is not tax software either. What it gives you is a dated record of every clean and every payment to file from, instead of a column of Y's to interpret. The numbers a one-person round should track is a shorter list than most people expect.
If you want a hand with the spreadsheet part
RoundWiz imports from a spreadsheet, and the import page sets out the file it expects, including how your round names come across exactly as you wrote them. If your notes live in a column of their own, say so when you send the file over. It is £7.99 a month, VAT in, 30 days free, no card to start and no per-customer fee, so 412 customers costs the same as 40. You can export the lot back out whenever you like, so the worst case is an evening lost. The features list is the honest version of what it does and does not do, and if your sheet is an odd shape, ask; it is one person in Essex answering the email.