The best timesheet system I ever ran was a whiteboard, a biro and a foreman I trusted, and it still cost me most of a Monday every week chasing hours and arguing over half an hour here and there. The worst was a slick app that half the crew had deleted from their phones by the second Friday, which tells you where the real problem lives.

I've run crews for the best part of fifteen years, on fit-out jobs, on event builds, and for a stretch at a small production company where six freelancers needed paying from timesheets that arrived, if they arrived at all, scrawled on the back of a delivery note. Automating time tracking and timesheets eventually fixed the admin. But it only worked once I stopped treating the software as a way to catch people out and started treating it as a way to get everyone paid right without a fight.

The Short Answer

Automate the capture, not the judgement. Give people one tap on the phone they already carry, stamp the location at the start and end of the shift and nowhere else, keep the job list short, and have the foreman approve hours once a week before they flow into Xero, Sage or BrightPay. Expect to pay somewhere between nothing and about £7 per head per month, and pick a tool that prices in pounds and keeps working with no signal.

The rest of this piece is the detail: what UK law requires you to keep, where the regulator draws the line on tracking, which features matter on a muddy site at 7am, which tools I'd put in front of a British crew, and how to roll it out so it sticks.

Why Crews Ignore Timesheets

Nobody on a crew wakes up wanting to fill in a form. When a timesheet is late or wrong, it's usually because the system asked for effort at the worst possible moment: at the end of a ten hour day, with cold hands, from memory. The fix isn't nagging. It's removing the memory part.

The numbers back this up. A survey of 4,248 UK employees paid through PAYE, the system employers use to deduct tax at source, was run for the Global Payroll Association in 2024, and HR Magazine's report on the payroll errors researchfound a quarter had received an incorrect payslip from their current employer. THP Accountants' summary of the same survey adds that 78 per cent of those were underpaid, 46 per cent of the mistakes came down to human error in working out wages, and another 29 per cent were overtime or bonuses left off entirely. Those are exactly the errors a paper timesheet invites.

The other side of the coin is the crew fiddling the sheet. The most recent figures I could find come from a US survey of more than 830 hourly workers by the time clock company OnTheClock, published in August 2026, which found that one in four had clocked in or out for a colleague in the previous year and 43 per cent had handed in hours that didn't match what they'd worked. I haven't found a British equivalent, and I'd treat the exact percentages with care, but the mechanics are identical here: paper plus memory plus a mate with a pen equals drift.

Buddy punching, which is the trade term for clocking a colleague in while they're still in the van, isn't a moral failing so much as the predictable result of a system that runs on trust and nothing else. So the design brief for automation is simple. Make the honest thing the easy thing. If clocking in takes one tap and gets someone paid for every minute, most people will do it, and the ones who won't were never going to fill in the paper properly either.

What UK Law Actually Asks You To Keep

Let me clear up the biggest misunderstanding first, because vendors trade on it. The Working Time Regulations 1998 do not require you to log every hour of every worker's day. Regulation 9 of the Working Time Regulations says an employer must keep records adequate to show the 48 hour average weekly limit and the night work limits are being met, and keep them for two years. As the employment lawyers at DavidsonMorris note in their 2026 guide to record keeping, the regulations prescribe no format and no dataset, and they don't expressly demand daily time recording.

The rule with teeth sits somewhere else. Under the National Minimum Wage Regulations 2015 you must keep records sufficient to prove every worker was paid at least the minimum wage for the time they worked, and HMRC's own minimum wage manual states that since 1 April 2021 those records must be kept for at least six years and be producible as a single document for each pay reference period. HMRC is His Majesty's Revenue and Customs, the tax authority, and it polices minimum wage rather than any employment body. A timesheet app with a six year retention setting and a clean export per pay period satisfies this without you thinking about it; a filing cabinet of damp paper doesn't.

Two more 2026 dates matter. From 6 April 2026 a new duty came in, as Make UK's briefing on the Employment Rights Act explains, to keep records of holiday entitlement and holiday pay for six years. And on 7 April 2026 the government launched the Fair Work Agency, a single state enforcement body that brings minimum wage, agency worker rules and eventually holiday pay and sick pay under one roof. For the 2026 to 2027 year HMRC is still running minimum wage checks under contract to the new agency, and holiday pay enforcement won't start before April 2027, but the direction of travel is obvious.

The penalty for underpaying is 200 per cent of the shortfall, capped at £20,000 per worker, and the employer gets named publicly. The payroll firm Zellis, in its April 2026 note on HMRC's proactive enforcement team, lists the traps that catch decent employers: time spent arriving early to set up, queuing to clock in, tidying up after the official finish. All of that counts as working time.

Rounding Is Where Automation Quietly Breaks The Law

Most time clock apps let you round punches to the nearest five, ten or fifteen minutes, and I used to think this was harmless tidiness. It isn't. Relentify's guide to timesheet rounding works through the arithmetic: a worker who loses seven minutes a shift to rounding loses about thirty hours of paid time a year, and if that pushes anyone's effective hourly rate below the minimum wage you're in breach regardless of intent.

My rule now is no rounding down, ever. If a legacy payroll system forces you to round, round in the worker's favour or don't round at all, and pay travel between sites wherever the contract says it's working time. Automation doesn't remove this decision. It just applies whatever you chose to every shift, every week, silently, which is why you need to choose well before you switch it on.

The Monitoring Line You Must Not Cross

This is the part most articles on time tracking skip, and it's the part that will hurt you. The moment your app records location, photos or a face, you're processing personal data under UK GDPR, the British version of the General Data Protection Regulation, and the regulator is the Information Commissioner's Office, or ICO. On 3 October 2023 the ICO published its guidance on monitoring workers, which applies to anyone doing work for you regardless of contract, so subcontractors and agency staff included.

The guidance boils down to three tests. Be clear about your purpose, and don't monitor people just in case. Use the least intrusive method that achieves that purpose. And tell people what you're collecting and why, in plain language, before you start.

Where monitoring is likely to be high risk, and the ICO names biometric data and anything that could cause someone financial loss as examples, you must carry out a data protection impact assessment first. A DPIA, to give it its acronym, is a written exercise where you describe what you'll collect, why it's necessary and how you'll limit the harm. It's a couple of hours' work, and it's the document that saves you when someone complains.

GPS is where this gets practical. Michele Angelo Petraroli, who builds a clocking app himself, wrote in Personnel Today in August 2026 about the distinction I wish I'd understood ten years ago: an app can stamp a position at the two taps that open and close a shift, or it can record where someone is continuously, on breaks and after hours. The regulator treats those as opposite intentions. The two stamps prove attendance, which is the record you're meant to keep, while the breadcrumb trail proves nothing you need and everything a tribunal will ask you to justify.

Several of the tools below sell breadcrumbs as a feature. I'd switch them off on day one.

Faces And Fingerprints Are A Different Category Again

In February 2024 the ICO ordered Serco Leisure and seven community leisure trusts to stop using facial recognition and fingerprint scanning for clocking in, after finding they'd been processing the biometric data of more than 2,000 staff across 38 sites with no real alternative on offer. The regulator's reasoning matters more than the headline. ID cards or fobs would have done the job, staff had been told the scanner was the only way to get paid, and given the power imbalance nobody could meaningfully say no.

Biometric data is special category data, so you need a lawful basis, a separate condition for processing it, a DPIA and a real alternative such as a PIN. Deputy, TimeKeeper and Jibble all offer face verification on a kiosk, and it does stop buddy punching. If you use it, make PIN entry available to anyone who asks, without questions, and write down why the face option is proportionate on your site. If you can't write that paragraph, don't turn the feature on.

On 8 July 2026 the government opened a consultation on workplace monitoring technologies, reported by Personnel Today, with options ranging from a statutory code of practice to a legal duty to consult staff before introducing monitoring tools, and it closes on 30 September 2026. Whatever comes out of it, the sensible move is to consult your crew now anyway. People accept a system they helped design.

The Five Features That Matter On Site

After trying more of these apps than I care to admit, here's what actually decides whether a crew keeps using one. The whole point of automating time tracking and timesheets is to make the record appear without anyone remembering to make it, and each of these features serves that.

Offline mode comes first. Basements, steel frames and rural sites kill signal, and an app that refuses to clock someone in without data will be abandoned within a fortnight. Onetrace, a UK compliance software firm, built its 2026 comparison of construction timesheet apps around real site conditions, poor connectivity included. The app should accept the tap and sync later, every time.

A geofence is the second. A geofence is a virtual boundary drawn around a site, and it lets the app confirm a clock in happened at the job without tracking anyone in between. Set the radius wide enough to cover the car park, because the first week of any rollout is people standing at the fence line trying to make the button work.

Third is a kiosk option: one shared tablet on the site cabin wall that people clock in on with a PIN. Not everyone wants a work app on a personal phone, and a kiosk removes the argument. Fourth is a job or cost code picker with no more than a handful of options, because cost codes are the labels that tell your accounts which job the hours belong to, and if the list runs to forty entries nobody picks the right one.

Fifth is a proper approval step and a payroll export in the format your payroll software wants, whether that's Xero, Sage, BrightPay or Moneysoft. Three features that look impressive in the demo and don't matter: live maps showing everyone's dot moving around, screenshot capture, and AI productivity scores. The first is a legal liability, the second is irrelevant on a site, and the third is nonsense for physical work.

Tools I'd Put In Front Of A UK Crew

Prices below are as published in September 2026 and exclude VAT. Where a platform bills in US dollars I've given an approximate pound figure at roughly 74p to the dollar, because the conversion moves.

Deputy

Deputy's UK price list bills in sterling: Lite at £3.25, Core at £4.25 and Pro at £6.50 per user per month on an annual commitment, paid upfront or in monthly instalments. Lite covers rotas, timesheets, clocking and payroll integrations, and Core adds timesheet auto approval and biometric kiosk clocking. Expertsure's UK review points out the catch for tiny teams: rolling monthly plans carry a £20 minimum, so a four person crew pays the minimum rather than the per head rate, and add-ons such as the £2 HR module and £1.40 Messaging+ can push a Lite user past the Pro price.

Best for: crews of six or more who also need rotas and want approved hours pushed straight into Xero.

TimeKeeper

TimeKeeper is the one I recommend most often to small British firms, partly because its pricing FAQ is in plain pounds: Core at £4 per employee per month, or £3.33 on annual payment, and Pro at £6, or £5 annually, which adds tracking time against jobs. There's a free trial of two weeks with no card required. It does kiosk clocking with a four digit PIN, GPS geofences and an optional face check, and it exports to Xero, Sage, BrightPay and Moneysoft, which is what most small UK employers actually run.

Best for: trades and installation crews of three to fifty who want sterling billing and a tool that does one job well.

Connecteam

Connecteam is the darling of the American contractor lists, and it earns some of that. The Small Business Plan is free for up to ten users with every feature switched on, which for a small crew is remarkable. Above that, Capterra's listing shows Basic at $29 a month on annual billing, about £21, for the first thirty users on the Operations hub, with Advanced at $49, about £36, adding geofencing for up to ten sites and automatic clock out. Monthly billing is $35 and $59, roughly £26 and £44, and the platform bills in dollars, so those pound figures are approximate.

The catch is the hub structure. Chat, training and HR documents live in separately priced hubs, so a business wanting all three can find the bill has tripled. It also ships a location breadcrumb feature, which you should leave off for the reasons above.

Best for: crews of ten or fewer who want scheduling, forms and a time clock for nothing, then a predictable flat fee as they grow.

Jibble

Jibble's free plan is free for unlimited users, and it includes the things a crew needs: kiosk clocking with PIN or face, a location stamp on each punch, offline mode and two geofences. The paid tiers went up in early 2026. Dupple's July 2026 review put Premium at $2.99 and Ultimate at $5.99 per user per month on annual billing, which is roughly £2.20 and £4.45, again billed in dollars, and Ultimate is where the payroll exports to Xero and QuickBooks live.

Best for: any size of crew that needs attendance and nothing fancier, at a price that's hard to argue with.

Clockify

I'm including Clockify because people will suggest it, and I'd steer a crew away. In April 2026 its parent company CAKE.com capped the free plan at five users and moved kiosk mode, approvals, time off and CSV exports behind paid tiers. Paid plans start at $3.99 per user per month, about £3, with approvals on Standard at $5.49, about £4, and GPS only on Pro at $7.99, about £6, all billed in dollars. It's a fine timer for an agency or a studio billing by the hour, but it is not a site tool.

Best for: office and studio teams tracking billable hours, not crews clocking on in the rain.

What About An HR Suite

If you already pay for HR software, check what's bundled before buying anything. Expertsure's May 2026 cost guide for UK time and attendance systems lists Sage HR Core at £5 per employee per month plus £2.50 for the timesheets module, RotaCloud at a flat £10 to £15 a month for one to five employees plus £4.50 for the clocking add-on, and BrightHR at roughly £16.67 per head at five employee scale, dropping steeply with volume. None of these are built around site work, but for a crew that's also a rota they can be enough.

A word on the American names that dominate search results: Workyard, ClockShark, ExakTime and busybusy. Buddy Punch's roundup describes Workyard's features as including Workers' Comp in payroll and certified payroll reports, which are US constructs, and these tools are built around American payroll and price in dollars, so I wouldn't start there for a British crew.

The Rollout That Sticks

Software is the easy half. Here's the sequence that's worked for me across several crews, and the order matters.

Start with one crew and one foreman who's on your side, for two full pay periods. Don't run paper alongside for longer than the first week, because the moment people know the old sheet still exists they'll default to it. Set the geofence generously, allow clocking in from ten minutes before the shift, and decide the missed punch rule up front: a missed clock out is fixed by the foreman by 5pm the same day, with a note, and the worker sees the correction in the app.

Then make the first payroll from the new data visibly right. Pay every minute, including the early arrival to unload the van. When the first payslips land with the overtime that used to vanish, adoption stops being your problem, because the crew has just discovered the app works for them. That single pay run does more than any training session.

Approvals should be weekly, on a fixed day, and take the foreman less than fifteen minutes on a phone. If it takes longer, your cost code list is too long or your rounding rules are generating exceptions. And write the privacy notice before day one, a page at most, saying what's collected, where the stamp happens, who sees it and how long it's kept. Hand it out, keep a signed copy, and treat that page as your DPIA's public face and your answer to any complaint.

Where The Usual Advice Is Wrong

The standard advice says buy the tool with the most features. I'd say buy the one with the fewest that still covers offline clocking, a geofence, a kiosk, short cost codes and a payroll export. Every extra feature is a thing you'll one day have to justify to the crew, the ICO or a tribunal.

The standard advice says GPS everywhere. The ICO's own guidance says least intrusive, and the Serco case shows what happens when a firm can't explain why it needed more than a fob. Two stamps a day is enough.

The standard advice says automation removes the need for a human check. It doesn't. Automation removes the memory and the arithmetic, which is where the errors were, but the foreman's weekly sign off is what stops one wrong rounding rule becoming a six year liability across the whole crew.

A Reality Check

None of this makes the crew love timesheets. The best outcome is indifference: they tap in, tap out, never think about it, and you stop losing your Mondays. Expect a fortnight of grumbling, a few people who forget their phone, and at least one site where the geofence needs redrawing.

Expect, too, that automating time tracking and timesheets will surface overtime you weren't paying. Budget for that rather than resenting it. It's the moment the crew starts believing the system is on their side.

What To Do This Week

Pull last month's timesheets and count the corrections. If it's more than a handful, you have your business case. Write the one page privacy notice and the short DPIA.

Then sign up for a free trial of TimeKeeper or the free tier of Jibble, set up one site with one geofence and a five line cost code list, and ask your most sceptical foreman to run it for a fortnight. If they'll keep using it, the rest of the crew will too.