The principle is the easy part
Every set of client money rules ever written rests on the same idea: money that belongs to a client is not the firm's money, must be kept apart from it, and must be accounted for in a way somebody else can check. Nobody in a professional practice needs that explained.
And yet firms get into difficulty, and when they do the story is almost never a dramatic one. It is a payment that cannot be tied to anything. A cost recharged with no document behind it. A ledger that has drifted a few hundred pounds over eighteen months, for reasons nobody can now reconstruct because the person who would have known has left.
In other words: the principle was understood throughout, and the paperwork was not maintained to the standard the principle needed. Those are very different failures with very different fixes, and firms tend to respond to the second by restating the first — more training on the rules, another reminder about care — which does nothing at all.
This article is about the second failure. Not what the rules say, which depends entirely on who regulates you, but what they ask of the filing cabinet — and where, in practice, that ask goes unmet.
A necessary caveat
This is not legal, regulatory or accounting advice, and it is deliberately not written as though it were.
The rules differ, a lot
Solicitors, surveyors, accountants, insolvency practitioners and property managers are each governed differently, and each differs again by country. An article cannot tell you what applies to you.
They change
Client money regimes are revised periodically, sometimes substantially. Anything written down in an article ages; the rules published by your own regulator do not.
We are not a party to any of it
FlowParse reads documents into rows. It holds no funds, moves nothing, and takes no part in any client account — handling client money is a regulated activity and this is not that.
What is left, and what this article is about, is the part that is common across all of them: a firm has to be able to show what happened to a sum of money, and to produce the document behind it, to somebody who was not there.
What tends to count as client money
The definitions vary and the boundaries are where the arguments happen, but the centre of the idea is consistent across professions.
| Situation | Whose money | What the record has to show |
|---|---|---|
| Funds received to be held for a client | Theirs | Every movement in and out, and the balance at any date |
| Money received on account of costs | Depends on the rules that apply | When it arrived, and when it stopped being theirs |
| A cost paid from the firm's own account | Yours, until recharged | What it was for, and the document behind it |
| Money received for a third party | Neither | Who it was for and when it reached them |
| Fees the client has been billed for | Yours once properly billed | The bill, and what it was made up of |
| An overpayment received in error | Theirs | That it was identified and what happened next |
The second row is where the professional rules diverge most sharply, and it is the one to check against your own regulator rather than against anyone's summary. The last row is the one firms handle worst — an overpayment that nobody noticed is a small operational error that becomes a serious one purely through the passage of time.
Notice the right-hand column, though. Whatever the category, the obligation lands in the same place: a record, retrievable, with the document behind it.
Holding money, and spending money on someone's behalf
These get discussed as one topic and they are two, with different risks attached.
Holding is heavily regulated, closely watched and generally taken seriously. Firms have systems for it, reconcile it on a schedule, and know it is the thing that gets them into trouble. The controls are usually real.
Spending on a client's behalf — a court fee, a search, an expert, a courier — attracts far less attention, because the money was the firm's own and the worst case looks commercial rather than regulatory. And so the discipline around it is looser, the documents are collected less rigorously, and a cost recharged to a client can turn out to have no evidence behind it at all.
That asymmetry is worth noticing, because a bill sent to a client includes both: fees and disbursements sit on the same document. A client querying the disbursements is querying the bill, and the firm's answer depends on records that were kept to a lower standard than the ones everybody worried about.
The mechanics of keeping that second set properly are the subject of a separate piece — how to track disbursements on a matter — but the reason it belongs in an article about client money is this: to the client, and often to a reviewer, it is all one bill.
What the rules ask of the paperwork
Strip away the differences between regimes and a short list remains. Every one of these appears, in some form, in every set of client money rules worth the name.
A contemporaneous record. Written at the time, not reconstructed later — because a reconstruction is an account of what someone believes happened.
Attribution. Every movement tied to a specific client and a specific matter, not to a general pool.
Evidence. A document behind each entry, and the ability to produce it rather than describe it.
Reconciliation on a schedule. The firm's records agreeing with the bank's, at stated intervals, with differences explained rather than carried.
Retention. Records kept for a defined period, in a form that can still be opened at the end of it.
Availability. Someone other than the author can find a given item. A record only one person can navigate is a record with a dependency.
Only the fourth is really about accounting. The other five are about documents — whether they exist, whether they are attached to the right thing, and whether anybody can lay hands on them. Which is why a firm with an immaculate ledger can still be in difficulty, and a firm with an ordinary ledger and excellent filing usually is not.
The sixth is the one that is never written on a compliance checklist and decides more outcomes than any of the others. It is the difference between a record and an archive.
Where firms actually come unstuck
Six recurring situations, none of which involves anyone doing anything wrong on purpose.
| What happened | Why nobody noticed | What surfaces it |
|---|---|---|
| A payment with no document | The document went to an inbox, not a file | Comparing payments against documents |
| A cost on the wrong matter | Two errors that cancel in the total | Per-matter review, not firm totals |
| A bulk statement posted as one line | It looked like one document | Reading statements as lines |
| An old balance nobody explains | Carried forward each month as 'known' | Ageing the difference, not just noting it |
| A receipt that faded | Thermal paper, filed unread | Capturing at arrival rather than at need |
| A record only one person can navigate | It worked, until they left | Someone else trying to find something |
Every row in the middle column describes something invisible from inside normal operations. That is the whole difficulty. None of these produce an error message, a failed reconciliation or an unhappy client at the time; they produce a question years later that the firm cannot answer.
The fourth row deserves particular attention because it is so easy to live with. A small unexplained difference gets carried forward, noted, carried forward again — and each month the note makes it slightly more normal. By the time anyone asks properly, the records that would have explained it are outside the window where they can be reconstructed.
The document that never arrived
If there is one failure mode worth designing against, it is this one: money left the account and the paperwork explaining it never entered the building.
It happens because payments and documents travel by different routes. The payment goes through the account automatically and lands in the statement whether anyone acts or not. The document goes to a person — an inbox, a portal, a coat pocket — and only exists in the firm's records if that person does something.
One of those two paths is reliable and the other is not, and no amount of care makes the unreliable one reliable, because it depends on somebody being at their desk and not busy at the moment it matters.
Which is why the single most useful control here is not a policy but a comparison: take the payments, take the documents, and look at what does not match. It is the only check that finds something the firm does not already know about, and it is usually the check nobody runs, because every other reconciliation compares two things that were built from the same source.
The practical version takes half an hour a month once both sides exist as rows. Statements come in the same way as cost documents do — the route is on bank statement to Excel — and the comparison is a sort and a filter.
How a review actually goes
Firms imagine an inspection as a judgement on their integrity. In practice it is much closer to a retrieval exercise, and it is worth understanding that because it changes what you would prepare.
Somebody picks items — often not many — and asks to see what is behind them. A payment from eighteen months ago: what was it for, and where is the document? A charge on a bill: what evidences it? A balance on a date: does the firm's figure agree with the bank's, and if not, why?
The questions are ordinary. What varies enormously is how long the answers take, and that is what forms the impression. A firm that produces each document in under a minute is doing something visibly under control. A firm that needs an afternoon per item is telling the reviewer something about its systems whether or not each individual answer turns out to be correct.
There is a second-order effect too, and it is the more important one. A reviewer who finds the first three items instantly tends to sample lightly. One who finds the first three difficult tends to look at more — and any records set that is examined for long enough will produce something to discuss.
So the practical target is not perfection. It is retrieval speed on an arbitrary item, chosen by somebody who does not know how your filing is arranged.
The unexplained payment, and the cost of leaving it
Every practice has a few. A round-figure payment two years ago with a supplier name nobody recognises. A transfer between accounts that made sense at the time. A refund that went out without a matching document.
None of these is necessarily wrong, and the vast majority turn out to be entirely ordinary once someone digs. The difficulty is what happens while they sit there: an unexplained item ages badly, because the people and the records that could have explained it both become less available over time.
The honest way to handle one is boring and effective. Write down, at the point you notice, what you know and what you cannot establish. Two sentences: this payment, this date, this is what we believe it was, this is the document we could not find and what we did to look.
That note converts a gap into something a firm has visibly dealt with. The alternative — leaving it undocumented in the hope that nobody asks — is the version that gets described unkindly later, because a gap somebody else discovers reads very differently from a gap the firm identified and recorded.
This is one of the few areas where the correct response costs almost nothing and the incorrect one costs a great deal, and firms still routinely choose the second because it requires no action today.
Residual balances, and the amounts nobody can return
Every firm that has held client money for more than a few years has a list it does not enjoy looking at: small balances belonging to people it can no longer reach. £14.20 for a client who moved abroad in 2019. £3.06 left over after a completion. £212 for an estate whose executor has himself died.
None of these arose from anything wrong. They are the residue of ordinary work: a fee estimated slightly high, a refund arriving after a matter closed, an interest credit on an account nobody expected to still be open.
What makes them awkward is that the money is unambiguously not the firm's, and the person it belongs to cannot be found. So it sits. And because it sits, it keeps appearing on every reconciliation, every year, as a set of lines nobody can clear — which is exactly the condition that trains people to stop reading the report.
Why they grow rather than resolve
A residual balance is never anyone's priority. It is small, it is old, and dealing with it means a search, a set of letters and a process that takes longer than the amount justifies. So it is deferred, and it is deferred again, and the list gets one line longer every quarter.
The compounding problem is that a long list of unresolvable lines is indistinguishable, at a glance, from a long list of unreconciled ones. A reviewer looking at fifty stale items cannot tell which are known residuals and which are genuine breaks that nobody has chased — and neither, after a while, can the firm.
The thing that actually helps
Not a clever fix — a separation. Residual balances that have been through a documented attempt to return them belong on their own schedule, with the date of the attempt and what was tried. Everything else stays on the live reconciliation where it can be chased.
That single split turns an intimidating list into two manageable ones: a short live list somebody works, and a longer dormant list that is reviewed annually and can be explained in one sentence to anybody who asks. The money has not moved, but the record now says what it is, which is the whole of the problem.
Most professional bodies have a route for dealing with genuinely unreturnable sums — often involving a threshold, a documented search and, above a certain amount, permission. What that route is where you practise is a question for your regulator, not for an article. But the schedule is what you will need whichever route applies, and it is worth building before it is asked for.
Interest, and the awkward question of whose it is
Money held for a client sits in an account, and accounts earn interest. Whose it is has a clear answer in principle — the client's — and a much messier one in practice, and the paperwork question is the one that catches firms out.
The mess comes from proportionality. Interest on £400,000 held for four months is a real sum that a client will notice and ask about. Interest on £600 held for nine days is a few pence, and the cost of calculating, recording and paying it exceeds the amount several times over.
Most firms therefore have a policy with a threshold, and most policies are perfectly reasonable. The failure is rarely in the policy — it is that the policy exists in somebody's head, was set at a time when interest rates were near zero, and has never been reviewed since rates moved.
A threshold written when a year of interest on a typical balance was pennies produces a very different outcome when the same balance earns a meaningful sum. The policy did not change; the world did. Firms that have not looked at their interest policy since it was written are usually applying a rule whose original justification no longer holds.
From a records point of view, three things need to exist: the policy itself, written down and dated; the calculation for anything above the threshold, retained; and evidence that clients were told what the policy is. The third is the one most often missing, and it is the one that turns a defensible policy into a difficult conversation.
Records that outlive the people who made them
Retention periods for client money records are measured in years — commonly six or more, depending on the regime. The people are not.
Over that period the practice management system is likely to be replaced at least once, the bookkeeper will change, the filing convention will be reorganised by somebody with good intentions, and the person who knew why a particular arrangement existed will have moved on. None of that is unusual; all of it is guaranteed.
What survives is only what was written into the data itself. Not conventions, not folder structures, not the shared understanding that “the old ones are in the other drive” — those evaporate with the people who held them.
The practical test is a single question, and it is worth actually running rather than imagining: hand a colleague a bill from three years ago, point at one disbursement line, and ask them to produce the document behind it. If they can do it without asking anyone, the record is genuinely available. If they cannot, the firm has an archive and believes it has a record.
There is a format point here too. An index that lives only inside a piece of software is readable for as long as that software is licensed and installed. The same index exported as a spreadsheet alongside the documents is readable in a decade with anything. Keeping both costs one export and removes a dependency nobody thinks about until a migration.
Small firms, where all of this is harder
Most writing on this subject assumes a compliance function, a finance team and someone whose job is oversight. A large share of practices handling client money have none of those — the same person does the work, sends the bills, and reconciles the account.
The obligations do not scale down. The available time does, and the effect is a specific pattern: the reconciliation gets done because it is obviously required, and the document filing slips because nothing fails when it does.
If only two things can be done properly in a small practice, these are the two with the best return.
One place where every cost document lands, reachable in seconds. Not a system — a destination. Most small-firm document losses are not a filing failure but the absence of anywhere obvious to file to at the moment the receipt is in hand.
One monthly comparison of payments against documents. Half an hour, and it is the only routine that finds what is missing rather than tidying what is present.
Everything else — matter allocation, ageing, the recharge decisions — improves the position but does not change the risk profile. Those two do.
What software can and cannot do here
It is worth being precise, because this is an area where vendors — including, in principle, us — have every incentive to be vague.
| Software can | Software cannot |
|---|---|
| Turn documents into records that can be searched | Make a firm compliant |
| Keep a link from a line back to its source page | Decide what your regulator requires of you |
| Make retrieval fast for someone who was not there | Produce a document that was never obtained |
| Show which payments have no document behind them | Explain a payment nobody can identify |
| Keep net, VAT and gross apart so either treatment works | Choose the treatment |
| Make the record independent of one person's memory | Replace the judgement of the person |
The right-hand column is the honest half and it is longer than most product pages admit. Compliance is an obligation on the firm, and no purchase transfers it. Anything sold as making you compliant is describing an outcome it cannot deliver.
The left-hand column is genuinely worth something, though, because it addresses precisely where the failures happen. Not the principle, not the intention — the retrieval, the completeness, and the independence from any one person's memory.
For what it is worth, FlowParse sits entirely in that left column and nowhere near the right one: it reads documents into rows and keeps the page reference. It holds nothing, moves nothing, and advises on nothing.
Four things firms assume that are not true
“The bank statement is the record.” It evidences that a payment happened. It says nothing about what the payment was for, which is the question actually asked. Two different records, and only one of them arrives on its own.
“If the reconciliation balances, we are fine.” A reconciliation confirms the totals agree. A cost attributed to the wrong client reconciles perfectly, because the money did move and the sum is right — and that is exactly the error a total cannot show.
“We keep everything.” Keeping and being able to produce are different claims. Retention is satisfied by storage; availability is satisfied by someone else finding it. Most firms can demonstrate the first and assume the second.
“It is a small amount.” Size is a poor predictor here. A small unexplained item and a large one raise the same question about the system that allowed it, and the small ones are more numerous — which is what makes a pattern.
What to do this quarter
Four things, in order of what they return for the time.
Run the retrieval test. Pick a bill from two or three years ago, choose one line, and ask somebody who did not work on it to produce the evidence. Time it. That number is your actual position, and it is usually a surprise.
Compare a month of payments against a month of documents. Not to fix anything yet — just to see the size of the gap. Everything else follows from knowing whether it is two items or twenty.
Write down the unexplained items you already know about. Two sentences each. The ones you are aware of and have not documented are the ones that read worst later, precisely because the firm knew.
Fix the destination, not the discipline. If cost documents are lost, the answer is a place to send them that takes two seconds, not another reminder. Reminders have been tried in every firm that currently has this problem.
None of that makes anyone compliant — see above, at length. It makes the records complete and retrievable, which is where the failures actually are, and it is the part a firm can do something about this quarter.