Take control of your business expenses with our expense management software. Simplify your financial tracking and save time. Try it now!
Every business has an expense process it is slightly embarrassed about. Claims arrive late, receipts arrive later, approval is nominal, coding is guesswork, and the whole thing runs on somebody in finance chasing people who would rather be doing something else.
The reason it stays that way is that it never becomes urgent. Expenses are individually small, nobody is obviously being harmed, and there is always something more pressing. So the process persists in a state nobody designed and nobody defends.
This article is a practical programme for fixing it, in the order that produces results fastest, without turning the process into bureaucracy.
Before changing anything, spend an hour looking at a sample of recent claims. Twenty is enough to see the pattern.
For each, note how long after the expense the claim was submitted, whether a receipt was attached, whether the purpose was described in a way somebody could assess, who approved it and how quickly, and whether the coding looks right.
Most businesses find the same things. Submission is weeks late. Receipts are missing or are card statements rather than tax invoices. Purpose descriptions are generic. Approval happened within minutes of submission, which tells you nothing was examined.
That sample is what makes the case for change, because it is specific rather than a general complaint. It also tells you which problem to fix first, which is usually not the one people assume.
The second useful piece of evidence is the effort involved, because expense administration is invisible work spread across several people.
Count the time whoever processes claims spends on it, including chasing. Count the time approvers spend, even if it is only a few minutes each. Count the time finance spends coding, correcting and querying. And count the time claimants spend assembling and submitting.
Across a year, in a business of any size, the total is generally larger than anybody expected, and a meaningful proportion of it is chasing rather than processing.
That number matters because it converts a process everybody finds mildly annoying into a cost with a figure attached, which is what justifies spending anything on improving it.
The first change worth making costs nothing and it is the foundation for everything else.
Most expense policies are long, and length is why nobody has read them. A policy people have not read is not a policy, it is a document, and the effective rules become whatever people assume.
The version that works fits on one page. What can be claimed and what cannot. Any limits that apply. What documentation is required. How quickly claims must be submitted. Who approves what. And what to do when something falls outside the policy.
That final point is the one most policies omit and the one people most need, because legitimate spend the policy did not anticipate will occur, and without an exception route it produces either an unfair refusal or a quiet approval that undermines the policy.
Write it in plain language. Put it where people encounter it when claiming rather than in a folder nobody opens.
Late claims cause more difficulty than any other single behaviour, and the fix is straightforward.
They land in the wrong period. They cannot be meaningfully approved because nobody remembers the context. They make budget monitoring unreliable because managers do not know what is still coming. And they produce a year end pile up when everybody finally empties their wallet.
Set a deadline, whether that is within the month, within thirty days or by the end of the period. Then attach a consequence, which does not need to be severe.
Requiring additional approval for late claims works. So does a simple report of who has outstanding claims, circulated to managers, which relies on mild social pressure and is remarkably effective.
The important thing is that the deadline is enforced rather than stated, because a deadline with no consequence trains people that deadlines here are advisory.
Finance codes expenses by inference, because finance does not know what a particular meal was for or which project a taxi related to.
The claimant knows. Moving coding to the point of claim improves accuracy immediately and moves the work to the person who has the information.
The requirement is that the choices offered are short and meaningful. A dropdown with two hundred accounts produces worse coding than one with twelve categories, because people select the first plausible option rather than searching a long list.
Where the business tracks projects, departments or sites, having those available at claim entry is what makes expense analysis by dimension possible later, and adding them retrospectively is considerably harder.
This change usually reduces finance's processing time more than any other, because correcting inferred coding is slower than reviewing coding somebody chose.
The single most useful field on an expense claim is the one describing why the expense was incurred, and it is almost always filled in uselessly.
Client meeting tells an approver nothing. It also fails to support the tax position if the claim is ever examined, because the substantiation requirement is about establishing business purpose.
The fix is partly a system one, meaning making the field mandatory with a minimum length, and partly a cultural one, meaning finance actually pushing back on inadequate descriptions rather than accepting them to avoid a conversation.
What good looks like is short and specific. Who was present, what was being discussed, which project it related to. One sentence.
Once approvers have that, approval becomes a real decision rather than a formality, which is the change that makes everything else worthwhile.
Approval is the step everybody relies on and the one most likely to be theatre.
The problem is structural. An approver looking at a total, weeks late, with no detail, has nothing to assess and approves because refusing would require an investigation they have no basis for.
Making it real requires four things. Line level detail rather than a total. The receipt attached rather than filed separately. A purpose description written by the claimant. And the claim arriving soon enough that the approver remembers.
It also requires the right approver, which is the direct manager who knows what the person was doing, rather than a finance approver who does not.
And approval thresholds should be set so routine claims pass quickly and larger or unusual ones get attention, rather than everything receiving identical nominal scrutiny.
Most receipt problems are timing problems. The receipt existed and then it did not.
Capturing at the point of spend, usually by photographing the receipt immediately, removes the entire category. The image is attached to a draft claim, and building the claim later is assembling items already captured rather than reconstructing from memory and a wallet.
This is the change claimants like most, which matters because a process people find easier gets complied with. Resistance to expense processes is usually about effort rather than about the rules.
It also improves what finance receives, since a photographed tax invoice is considerably more useful than a faded thermal receipt scanned three weeks later.
Whether that capture happens in a dedicated application or within your finance system matters less than that it happens at the moment of spend.
Cards change the process substantially and are worth deciding on rather than drifting into.
They remove the reimbursement cycle, which is a genuine benefit to employees funding business spend from personal money, and they produce a transaction feed finance can see before any claim arrives.
They also mean the money has left the business before anybody reviews it, so approval becomes a review rather than a control and recovering inappropriate spend is harder than declining a claim.
Where cards are used, the controls that matter are appropriate limits per cardholder, category restrictions where the provider supports them, a substantiation deadline, and a consequence for cardholders who do not substantiate.
That last one is where most card programmes are weak, and where a business ends up with spend it cannot support.
Expense categorisation should reflect tax treatment rather than only the nature of the spend, because that is what makes the compliance position derivable rather than reconstructed.
Fringe benefits tax exposure builds through the year invisibly, since the expense is processed as an ordinary cost and only becomes a question when somebody prepares the annual return. Entertainment, private use of company assets and some travel and vehicle arrangements are the common categories.
Where categories distinguish the treatments at the point of claim, the annual return is assembled from data. Where they do not, somebody reconstructs it from a year of transactions coded generically.
Input tax credits are the other half. A credit card statement is not a tax invoice, and businesses reimbursing against statements are frequently claiming credits they cannot support.
Getting the categories right once, with somebody who understands the treatments, saves the annual reconstruction indefinitely.
Individual claim review is a control. Reviewing the totals is where the findings are, and almost nobody does it.
Spend by category over time shows what is actually happening, and a category growing faster than the business is worth understanding.
Spend by claimant identifies outliers, most of which are explicable and some of which are not.
Spend by supplier occasionally reveals a procurement decision nobody made, where several people are individually buying something the business could contract for at better rates.
And the exception rate indicates whether the policy fits reality. A high proportion of claims needing exception approval usually means the policy is wrong rather than that people are non compliant.
An hour a quarter is enough, and it is the step most businesses skip entirely.
Once the process is sound, technology removes effort rather than papering over a bad design.
Receipt capture with automatic extraction of the amount, date and supplier removes most of the data entry from claiming.
Policy checking at the point of claim tells the claimant immediately that something exceeds a limit or lacks a receipt, rather than finance discovering it later and starting a conversation.
Card feed matching pairs transactions with substantiation automatically, so the unsubstantiated list is generated rather than assembled.
And approval routing based on amount, category and department removes the decision about who should approve, along with the chasing.
The sequencing matters. Automating a broken process produces a fast broken process, and businesses that buy tools before fixing the policy and the approval structure generally do not get what they expected.
Expense management sits in one of three places and the choice has ongoing consequences.
Spreadsheet and email works at small scale and provides no audit trail, no duplicate detection and no visibility until claims arrive.
A dedicated expense application handles capture, policy and approval well and creates an integration and a reconciliation with your finance system.
Or the ERP itself, where claims are entered, approved and posted in the same system as everything else, which removes the integration, the reconciliation and the duplicate employee record, and means expense data carries the same dimensions as everything else.
For businesses on NetSuite, expense capability exists within the platform, and using it rather than adding a separate application removes a category of ongoing work. Our piece on advanced NetSuite modules covers how to assess that.
The opposite failure is real and common, and worth guarding against as the programme proceeds.
Every additional approval step adds delay, and delay produces workarounds. People who cannot get a legitimate expense approved quickly either stop incurring legitimate expenses or find a route that avoids the process.
Requiring receipts for trivial amounts costs more in administration than the amounts involved.
And treating every claimant as a suspect damages the relationship with the honest majority in exchange for marginal protection against the few who are not.
The proportionate position is light controls applied consistently, with attention concentrated on higher value and higher risk categories rather than spread evenly.
If you are starting from a process that is not working, the sequence that produces results fastest is reasonably consistent.
Look at twenty recent claims and count what the process costs. Rewrite the policy to a page. Set a submission deadline and enforce it. Make the purpose field mandatory and meaningful. Move coding to the claimant. Give approvers something real to look at.
Then, once those are established, introduce capture at the point of spend, then policy checking, then approval routing.
Then review the aggregate quarterly, which is what keeps the process aligned with the business rather than drifting back.
The first six changes cost nothing and produce most of the improvement. The technology accelerates a process that is already sound.
Expense management is one of the few finance processes where a meaningful improvement is available without buying anything, and where the obstacle is attention rather than budget.
The useful first step is the sample of twenty claims, because it converts a general dissatisfaction into a specific list, and the specific list is what gets acted on.
Our piece on tight expense management procedures covers the control and compliance dimension in more detail, and payroll and bookkeeping covers the wider finance operation this sits within.
If you would like a view on where your expense process is losing time, get in touch.
