If fewer than 25 people submit expenses, you process under 50 claims a month, and one person approves everything, a manual process is still doing its job just fine. Somewhere past those numbers, though, it starts costing you more than the software would.
Figures verified against HMRC and gov.uk guidance on 28 July 2026.
Working out where that line sits is harder than it should be. A lot of businesses either buy software too early, when a decent spreadsheet and a bit of discipline would’ve done the job for another year, or they hang onto that spreadsheet long after it’s started quietly costing them in finance time, missed VAT, and claims nobody’s checking properly any more.
This page is here to help you find that line for your own business, rather than guess at it. You’ll get nine testable triggers, the exact figure that trips each one, and what to do about it if you haven’t hit any of them yet. If you want the underlying cost comparison, spreadsheets versus expense software models it properly. If you want the definition first, start with what expense management software is, and how it works end to end covers the claim lifecycle.
The three questions to ask yourself
Before the detailed test, these three questions answer if you need expense management software for most businesses.
- Does anybody in finance spend more than a full working day a month chasing, checking or retyping expense claims? A day a month is roughly the point at which the admin stops being a rounding error and starts being a job nobody was hired to do.
- Is there any month where you couldn’t produce, within an hour, the receipt behind a specific line in your VAT return? If the answer’s no, your evidence trail has a hole in it, and that hole gets bigger every month.
- Has anyone approved their own spend in the last twelve months? Including the founder, the finance director and the person covering for someone on leave. If yes, you don’t have an approval process—you have a habit.
One “yes” is a warning. Two is a case for change. Three, and the question is no longer whether you need something, but rather it’s what you buy and when.
Test your own process
Answer the nine questions below with the figures for your business. The test scores the pressure on your current process, names the specific triggers you’ve already hit, and tells you what each one is costing you. There’s no email gate and no form, simply a helpful tool to guide your decision making. If you come out in the lowest band, it’ll tell you so, and guide you towards what to do next.
Have you outgrown your current expense process?
Triggers hit: 2 of 9. Pressure score 14 out of 48.
Two or three triggers. Fix the process first, then revisit in six months.
You have crossed one or two thresholds, but not enough of them to make software the obvious answer. At this level a tightened policy, a fixed monthly claim deadline and a proper filing convention will buy you another year. Put a date in the diary to run this test again, and treat the triggers below as the things to watch.
Triggers you have hit
- You are VAT registered. Every claim with recoverable VAT needs evidence you can produce on demand, and a spreadsheet row is not evidence. Missing or unreadable receipts turn into VAT you cannot reclaim, and a photo pasted into a sheet then retyped breaks the digital link rule in VAT Notice 700/21.
- You are running expenses on a spreadsheet. A spreadsheet has no approval record, no version history you would want to rely on, and no link between a number and the receipt behind it. It works until someone leaves, a formula gets dragged, or HMRC asks a question about a claim from two years ago.
Next step: write down the three things that took the longest last month end. If two of them are expense related, you have your answer without needing a business case.
Thresholds are drawn from UK statutory requirements and from the operational points at which manual controls typically fail. Nothing is stored and nothing is sent anywhere.
Two or three triggers. Fix the process first, then revisit in six months.
You have outgrown it. Time to build the case.
Past the point where this is a cost question.
Your process is small enough that the controls a person can hold in their head are still adequate. Buying software now would add licence cost, an implementation, user training and a new system to administer, in exchange for solving a problem you do not have yet. Spend the money on something that moves revenue instead, and use the free fixes further down this page to make your current process defensible.
You have crossed one or two thresholds, but not enough of them to make software the obvious answer. At this level a tightened policy, a fixed monthly claim deadline and a proper filing convention will buy you another year. Put a date in the diary to run this test again, and treat the triggers below as the things to watch.
The volume, the number of hands involved and the compliance surface have all moved past what a manual process holds together reliably. The cost is real but it is mostly invisible, because it shows up as finance time, unclaimed VAT and errors nobody traces back to the process. This is the band where a business case usually pays for itself inside a year.
At this level the manual process is not a cost problem, it is a control problem. You have enough claimants, approvers, entities or cards that no individual can see the whole picture, which is the condition under which policy breaches and duplicate claims stop being caught at all. The question is no longer whether to change, it is how quickly you can without disrupting a month end.
Next step: do not buy anything. Write a one page expense policy, set a claim cut off date each month, and store receipt images in a dated folder structure. That covers you for the next stage of growth at zero cost.
Next step: write down the three things that took the longest last month end. If two of them are expense related, you have your answer without needing a business case.
Next step: count the finance hours spent on expenses over one full month, then price those hours. Compare that against a per user per month licence for the number of people who actually claim, not your total headcount. That single comparison is usually the whole business case.
Next step: before you look at any product, write down your approval thresholds by value, your entity structure and how your card transactions arrive. Those three things decide which systems can actually handle you, and they eliminate most of a shortlist in one meeting.
No triggers hit. Your process is inside every threshold on this page.
Claim volume above 50 a month.
Checking and coding claims stops fitting into the gaps between other work. At this volume the failure is not a wrong number, it is a claim that sits unprocessed for a month because nobody owned it, and an employee who is out of pocket for longer than they should be.
More than 20 people submitting claims.
Once more than about twenty people claim, the person checking cannot know each claimant’s normal pattern any more. That pattern recognition is the only real fraud control a manual process has, and it stops working silently rather than obviously.
Four or more approvers.
With four or more approvers you now need rules rather than habits: who covers holidays, what happens above a certain value, and who approves the approver. Email approval trails scatter across mailboxes and are not recoverable when the approver leaves.
More than one legal entity.
Costs have to land in the right company, and intercompany recharges have to be traceable. A shared spreadsheet with an entity column is the single most common source of misposted expenses, and the correction usually surfaces at year end when it is most expensive to fix.
You are VAT registered.
Every claim with recoverable VAT needs evidence you can produce on demand, and a spreadsheet row is not evidence. Missing or unreadable receipts turn into VAT you cannot reclaim, and a photo pasted into a sheet then retyped breaks the digital link rule in VAT Notice 700/21.
More than 20 mileage claims a month.
Mileage needs a running annual total per person, because the approved rate drops from 55p to 25p a mile once someone passes 10,000 business miles in the tax year. A spreadsheet almost never tracks that per person across a full year, so the overpayment becomes a taxable benefit nobody reported.
You have a company card programme.
Cards invert the process. The transaction exists before the receipt does, so the job is chasing evidence for spend that has already left the account. Without automated matching, the unreconciled pile grows every month and the VAT on those transactions is not recoverable.
You are running expenses on paper or a spreadsheet.
A spreadsheet has no approval record, no version history you would want to rely on, and no link between a number and the receipt behind it. It works until someone leaves, a formula gets dragged, or HMRC asks a question about a claim from two years ago.
More than 8 finance hours a month on expense admin.
A full day a month of qualified finance time on chasing and retyping is a real cost with a real hourly rate attached, and it is the number that makes most business cases. It is also the number nobody measures until they are asked to.
The nine trigger conditions, and the number that trips each one
Every page on this subject says you should buy expense software “as you grow”. None of them says at what. Here are the actual thresholds, what crossing one costs or risks if you do nothing, and the cheapest fix that doesn’t involve buying anything.
Nine triggers, with the threshold and the consequence
| Trigger | The testable threshold | What it costs or risks | Cheapest fix short of buying software |
| Claim volume | More than 50 claims a month across the business | Claims start queueing. Employees are out of pocket for longer, and the checking gets shallower as the pile grows | A hard monthly cut off date, and a single named owner for the queue |
| Claimants | More than 20 people submitting | The reviewer stops recognising abnormal patterns, which is the only fraud control a manual process has | A published per category limit, so a breach is arithmetic rather than judgement |
| Approvers | Four or more people approving, or any self approval at all | Approval evidence scatters across mailboxes and is lost when someone leaves. Self approval is the single most common audit finding | A written delegation list with named deputies, and a rule that nobody approves their own line |
| Entities | Two or more legal entities, or any spend in a foreign currency | Costs land in the wrong company. Foreign VAT gets claimed on a UK return, which it cannot be | A separate claim form per entity, and a rule that the original currency and rate are recorded on the claim |
| VAT registration | Taxable turnover above £90,000 in any rolling 12 months | Input tax you cannot evidence is input tax you cannot keep. Records must be retained for six years | Receipt images filed by claim reference, not by date, so any line can be traced back in under a minute |
| Mileage | More than 20 mileage claims a month, or anyone approaching 10,000 business miles a year | Paying 55p beyond 10,000 miles creates a taxable benefit. Nobody notices until a P11D is wrong | A per person running mileage total maintained monthly, reset on 6 April |
| Card programme | Any company cards where a receipt must be matched to a statement line | The unreconciled pile compounds. VAT on unevidenced card spend is not recoverable | A weekly, not monthly, receipt chase, and a written rule on what happens when a receipt never arrives |
| Audit or funder scrutiny | Crossing two of £15m turnover, £7.5m balance sheet, 50 employees, or taking grant funding with a reporting condition | Sample testing becomes someone else’s job. Missing approvals turn into audit findings or a clawback | Retain the approval email with each claim record, not in a separate mailbox folder |
| Finance time | More than 8 hours a month of finance time on expense admin | A full working day a month of qualified time. This is the number that carries most business cases | Measure it for one month before deciding anything. Most teams have never counted it |
Three or more of these together is the point at which the free fixes stop holding. One on its own almost never is.
Five signs you’ve outgrown spreadsheets
The thresholds above are the measurable version. These are the symptoms you’ll notice first, usually before anyone’s counted anything.
1. You can’t answer “why is this cost centre over budget” without opening three files
A spreadsheet holds the claim. The accounting system holds the posting. Someone’s mailbox holds the approval. Nothing joins them up, so every question about spend becomes a small investigation. The tell isn’t that the answer’s hard to find—it’s that you’ve started declining to look, because you know how long it’ll take.
2. Receipts arrive in four different ways, and none of them is the agreed one
Photographs by text message, forwarded confirmation emails, a paper wallet handed over at month-end, and the occasional claim with a note saying the receipt was lost. When capture isn’t enforced at the point of spend, it degrades to whatever’s easiest for the claimant, which is exactly the behaviour a manual process can’t correct.
3. Somebody’s approved their own expenses in the last year
Usually the founder, the finance director, or whoever was covering during annual leave. It’s rarely dishonest and almost always defensible individually. It’s also the first thing an auditor tests, and a manual process has no mechanism to prevent it, only a convention that people remember to follow.
4. Mileage is calculated from memory, and nobody tracks the annual total
The approved rate for cars and vans rose to 55p per mile for the first 10,000 business miles from 6 April 2026, the first change since 2011 to 2012, with 25p a mile above that and 5p per passenger. Two things follow. Most spreadsheets in use today still have 45p hard-coded somewhere. And almost none of them track a per-person running total across the tax year, which is what decides when the rate should drop. Our guide to the 2026 to 2027 mileage rates covers the detail, and mileage capture is the feature that takes the arithmetic off your plate.
5. Month-end has a queue in it with your name on it
The clearest sign is scheduling. If expense processing has become a block in the calendar rather than something absorbed into a normal week, the volume has passed what the method supports. That’s a capacity statement, not a criticism of the spreadsheet.
The benefits of expense management software, and which ones you’d actually get
When researching software, the lists of benefits are not a guarantee. In practice, each benefit only shows up under a specific condition, and if you don’t meet the condition, it doesn’t arrive. Here’s the honest mapping.
- Time saved in finance: This is real, and it’s usually the largest single line on any business case. But it scales with claim volume, not with good intentions. If you’re processing fifteen claims a month, you’ll save minutes here and there, not whole days, so it’s worth being realistic about the size of the win before you build a case around it.
- Fewer errors: Also real, but only where the errors are arithmetic, coding, or rate application (the kind of thing a system can catch reliably). Software doesn’t fix a badly written policy, and it’s worth remembering it’ll apply a wrong rule just as confidently as a right one, so the policy still has to be sound going in.
- Policy enforcement: This is where most of the value actually concentrates, but it only exists if you have a policy to enforce in the first place. If you don’t, that’s the thing to sort out first. Our expense policy template is a reasonable place to start, and policy compliance covers how it turns into rules the system can actually apply.
- Fraud reduction: Genuinely useful once you’re above roughly twenty claimants, because that’s the point where human pattern recognition stops being reliable. Below that number, your reviewer usually already knows what everybody’s normal spending looks like, so duplicate detection and flagging tools add comparatively little.
- Faster reimbursement: Real, and it matters a lot more to people than finance teams sometimes give it credit for. Being out of pocket for six weeks is a genuine grievance, especially for field staff who are regularly fronting large costs out of their own pocket.
- VAT recovery: Often the benefit that ends up paying for the licence outright, though almost nobody sits down and models it properly. It only applies if you’re VAT registered, and the actual saving comes from receipts that would otherwise have gone missing, rather than from any particular cleverness in the software itself.
- Real-time visibility: Genuinely useful once you’re above a few hundred claims a month, where things move fast enough that a delay actually matters. Below that, your accounting system is probably already telling you what you need, just with a slight lag that rarely causes a problem.
- Card reconciliation: Only relevant if you run company cards, but if you do, it’s often the single strongest reason to buy. Matching transactions to receipts is consistently the job that manual processes handle worst, so this is where automation tends to earn its keep fastest.
- Accounting integration: A real benefit, and one worth testing properly before you commit. Ask to see the actual export file rather than a screenshot of an integration logo, so you know what you’re really getting.
- Scalability: The vaguest of the benefits typically claimed, and the one we’d encourage you to discount most heavily. It only means something concrete if you can name the specific number you expect to cross, otherwise it’s just a word.
A real example: Jimmy’s Iced Coffee
Jimmy’s Iced Coffee hit several of the triggers above at once. A busy events team, spend across departments, and a shared card with no automatic matching added up to 30 hours of admin a month—approvals done one at a time, receipts chased by hand, and VAT quietly missed on mixed transactions.
After switching to Capture Expense with Pliant cards, that 30 hours came down to minutes. Receipts now come in by WhatsApp and match automatically, and splitting receipts to apply the right VAT rate means they’re reclaiming money they were losing before.
Read the full Jimmy’s Iced Coffee case study.
Capture Expense is the answer
If you’ve worked through the triggers above and hit one or none, you’re probably not under enough pressure to make software an obvious priority right now. That’s a fair place to be. But it’s worth being clear about what that actually means: it’s not that a manual process is a good long-term answer, it’s that you haven’t hit the point where its limits start costing you.
That point tends to arrive faster than people expect. Claim volume creeps up, someone new starts approving expenses, a company card gets issued. And even before any of that, a small, well-behaved expense process still picks up real benefits from software—faster reimbursement for the people fronting their own money, a receipt trail that’s already sorted when someone asks for it, one less spreadsheet that only one person really understands.
That’s really what Capture Expense is built for: handling the parts of expenses that get harder as you grow—VAT evidence, mileage tracking against the annual threshold, card reconciliation—while staying straightforward enough to use well before you’re relying on it for compliance reasons. Whether you’re sitting comfortably below every trigger on this page or ticking off three or four of them, it’s worth seeing what it actually does.
If you’d like to see how it works in practice, you’re warmly invited to take a look, and we’re happy to talk through your specific numbers if that’s more useful than a general demo.
Frequently asked questions
Do I need expense management software?
Not if fewer than 20 people claim, you process under 50 claims a month, one or two people approve everything and you have no company cards. Past those points, the manual process starts costing more in finance time and unrecovered VAT than a licence does. The strongest single trigger is a company card programme, because the transaction exists before the receipt and manual matching fails quickly.
At what size does a business need expense management software?
There’s no headcount that settles it, because a 30-person field engineering business with mileage and cards has far more expense pressure than a 300-person office business where twelve people ever claim. Count claimants rather than employees. Above roughly 20 to 25 regular claimants, or 50 claims a month, the manual controls start failing.
What are the benefits of expense management software?
Faster processing, fewer coding and rate errors, enforced policy, better VAT recovery, faster reimbursement, card reconciliation, an audit trail that survives people leaving, and visibility of committed spend. Each one only shows up under a specific condition, so it’s worth counting how many apply to you before assuming you’ll get all of them.
Can I just use a spreadsheet for expenses?
Yes, up to a point, and for a small team it’s a perfectly reasonable choice. It stops being reasonable when you’re VAT registered and keying figures manually into your return, because VAT Notice 700/21 doesn’t accept cut and paste as a digital link, and when the approval evidence lives in mailboxes rather than with the claim.
Is a photo of a receipt enough for HMRC?
Yes, for VAT purposes. VAT Notice 700/21 section 6 confirms that where the image contains all the required detail and is retained, the original invoice doesn’t need to be kept. If only selected data is typed into your system and the image isn’t retained, you must keep the original.
How long do I have to keep expense records in the UK?
VAT records generally for six years. PAYE records, which include taxable expenses and benefits, for three years after the end of the tax year they relate to. HMRC can estimate what you owe and charge a penalty of up to £3,000 where records aren’t accurate, complete and readable.
What is the mileage rate and does the software need to track it?
55p a mile for the first 10,000 business miles in cars and vans from 6 April 2026, the first change since 2011 to 2012, then 25p a mile above that, plus 5p per passenger. The tracking matters because the rate change is triggered by a per-person running total across the tax year. Note that the gov.uk rates and thresholds page for employers shows 45p, because that’s the National Insurance figure rather than the tax one.
What should I do if I’m not ready to buy anything yet?
Write a one page policy, set a monthly claim deadline, file receipt images by claim reference somewhere that outlasts an employee, ban self-approval in writing with named deputies, and check whether your accounting package already includes expense capture in the tier you pay for. Then re-run the test in six months.