The report is only as good as the balances underneath it
A finance director's weekly cash update looks, from the board's side of the table, like a single confident number. From the finance team's side, it is the end of a routine that started days earlier — statements requested, downloaded, opened, and read one account at a time, across however many entities the group holds.
Where that routine breaks down is rarely the analysis. Totalling balances and spotting a trend is fast once the numbers exist in one place. The slow, error-prone part is getting them into that one place at all — a dozen PDFs in a dozen formats, from a dozen banks, some in a different currency, gathered by whoever on the team had an hour free that week.
The pain scales with entity count, not with transaction volume, which is why it catches so many groups off guard. A single-entity business with a hundred transactions a week barely notices this problem — one statement, one format, one person who has read it a hundred times before. A twelve-entity group with the same total transaction volume, spread thin across twelve accounts and a handful of banks, spends far more time just assembling the picture than analysing it, even though the underlying numbers are no more complex. Growth by acquisition makes this worse in steps rather than gradually: each new entity adds another statement format, another login, another line item that has to be chased down before the weekly number can go out.
Where finance directors actually lose time
None of these five is a single dramatic failure — each one is a small, recurring drag that a finance director tolerates because no individual week is bad enough to justify fixing it. Added up across a year, they are usually the largest line item in what the treasury routine actually costs, well ahead of anything the analysis itself takes.
Retyping balances from PDF statements
The single largest time cost in most routines, and the one least visible to anyone above the team actually doing it.
Chasing a subsidiary's statement that arrived late
One missing account holds up the entire report, or gets carried forward silently and quietly discovered wrong later.
Reconciling internal transfers between entities
A payment from one group company to another is easy to double-count or drop entirely if it is not explicitly matched.
Explaining a currency movement that was actually a data problem
A figure that looks like an FX swing is sometimes just a misread statement — and untangling which takes longer than either would alone.
Rebuilding the whole process after someone leaves
Undocumented routines walk out the door with whoever built them, and the next person starts over.
A realistic weekly routine
Request or gather statements
Every entity, every account, as of the same cutoff — a live feed where one exists, a downloaded statement everywhere else.
Read every balance
Closing balance, statement date, currency and account, from whatever format each one arrives in.
Check continuity
Each account's opening balance against last week's closing, catching a missed update before it reaches the board.
Resolve flagged transfers
Confirm matching debits and credits between entities, so intercompany movement is never counted as new cash.
Total by entity, by currency, by group
Three views of the same underlying numbers, because the board, the CFO and a lender each read a different one first.
Compare against last week and the covenant thresholds
The single chart that usually gets the most attention in the actual meeting.
Notice that only step two — reading each balance — is genuinely mechanical. Steps one, four, five and six all involve a decision at some point: which statement counts as this week's, whether a flagged transfer is really internal, how a currency subtotal should be labelled. Removing the friction from step two is what frees the finance director's time for the decisions that actually need a person, rather than for retyping numbers a document already states clearly.
Who owns each step matters more than it first appears, too. In groups where “whoever has time” gathers the statements, the routine quietly degrades the first busy week — an account gets missed, nobody notices until the following week's reconciliation, and the report goes out with a gap nobody flagged. Naming an owner for steps one through four, even if the same person doesn't do all of them every week, is a small governance decision that keeps the routine from silently eroding.
The cost of doing it by hand, honestly
For a group of twelve entities averaging two to three accounts each, manual gathering and reading commonly runs four to eight hours a week — more in a week where a statement is late or a format changes without warning.
| Approach | Typical time per week | Where it goes wrong |
|---|---|---|
| Fully manual | 4-8 hours | Retyping errors, late statements silently skipped |
| Spreadsheet template, manual entry | 3-5 hours | Still relies on someone reading each PDF correctly |
| Document reader plus a template | Under 1 hour | Requires the account list and template set up once first |
At a loaded internal cost of £30-40 an hour, the gap between the first and third row is roughly £100-280 a week — £5,000-14,000 a year — for a task that produces exactly the same report either way. The larger and less predictable cost is what a late or wrong figure costs when it reaches the board or a lender before anyone catches the error.
That second cost is harder to put a number on but easier to recognise once it has happened once. A board that is told cash is comfortable and later learns a subsidiary's balance was mistyped does not just correct the figure — it starts asking for the underlying statements every time, which slows every future report down. A lender who catches one wrong number in a covenant pack asks harder questions on the next one, whether or not the substance of the position was ever actually at risk. The time saved by reading statements automatically is real and worth having on its own; the credibility protected by never having a retyping error reach the board in the first place is usually worth more.
Who does what, once the routine exists
A weekly treasury routine works best when it is a chain of clearly separated jobs, not one person doing everything under time pressure. Splitting it up also makes the routine survive someone being on leave or leaving the team entirely, which a one-person process never does.
| Step | Typically owned by | Judgment required |
|---|---|---|
| Gathering statements | Analyst or bookkeeper | Low — mostly chasing and downloading |
| Reading balances | Automated, or an analyst as fallback | Low — the numbers are stated on the document |
| Flagging unusual transfers | Analyst, escalated when unclear | Medium — needs context on the group's normal patterns |
| Reviewing the trend and covenant headroom | Finance director | High — this is the actual judgment the report exists for |
| Presenting to the board or lender | Finance director or CFO | High — framing and anticipating questions |
The pattern worth noticing is that the two lowest-judgment rows — gathering and reading — are also the two that consume the most clock time in a manual routine, while the two highest-judgment rows — reviewing and presenting — take the finance director's time but relatively little of it. Automating the low-judgment rows does not change who owns the high-judgment ones; it just stops the low-judgment work from eating the hours that should go to the judgment calls instead.
What changes
Hours back, every week
The mechanical reading work shrinks from hours to minutes, freeing time for the analysis that actually needs a person.
A traceable figure
Every balance on the report links back to the statement it came from, for when the CFO or the auditor asks where a number came from.
A routine that survives a handover
Documented steps and a fixed template, not tribal knowledge that leaves with whoever built it.
A trend, not just a snapshot
Weekly figures accumulate into a real history — see balance timeline for the running view this builds toward.
Scenario: the acquisition
A group completes a small acquisition mid-quarter. The new entity has three accounts at a bank the group has never used before, statements in a layout nobody on the team recognises, and a currency the group does not otherwise hold.
Under a fully manual routine, this is the week the report either goes out late or goes out with the new entity missing entirely, quietly, for someone to notice later. Adding three unfamiliar accounts to an already tight weekly window is exactly the kind of change a hand-built process handles badly.
A document reader that works from the statement rather than a specific bank's feed treats the new accounts the same way as every existing one — read the closing balance, the date, the currency, add three rows to the account list, and the new entity is in the position from its first available statement rather than absent until someone builds a special case for it.
Scenario: the covenant test
A facility agreement requires the group to maintain a minimum cash balance, tested quarterly, with the lender entitled to ask for evidence. Six weeks before the test, the finance director wants to know whether the group is comfortably clear or uncomfortably close.
Answering that from a single point-in-time snapshot is weak evidence — it shows one day, not a trend, and a lender reviewing it has no way to tell whether that day was typical or a temporary high point timed to look good. What actually answers the question is the balance history leading up to the test date, sourced to the statements behind it.
Because the weekly routine already produces a dated, sourced figure every week, that history already exists by the time anyone asks for it — nobody has to reconstruct six weeks of balances from scratch under time pressure the week before a lender call.
Scenario: the currency move nobody saw in the blended total
A group holds roughly a third of its cash in a second currency, spread across four entities. For most of the year that split barely matters — the group total is reported in the parent currency, converted at a standard rate, and the number moves in a way everyone finds unremarkable.
Then that currency moves sharply against the parent currency over three weeks. The blended, converted total the board has been watching shows a modest change — modest, because the loss on the foreign-currency third is partly offset in the arithmetic by the unchanged domestic two-thirds. Nobody in the room realises how much of the group's actual purchasing power in that currency has moved, because the blended figure was never built to show it.
A finance director whose weekly routine keeps currencies separate — rather than converting at the point of totalling — catches this the first week it happens, not the quarter it becomes a real problem. The four affected entities show a currency-specific figure moving sharply while the domestic entities do not, and that divergence is visible precisely because nothing was blended away before anyone looked at it.
What happens next — hedge, hold, or do nothing — is a treasury policy decision this report does not make. What the report has to do, and the reason currencies stay unconverted throughout this whole routine, is make sure the decision gets made deliberately, by someone who saw the real number, rather than never being made at all because a blended total hid that there was anything to decide.
What the board actually wants to see
One clear total, with the confidence to answer a follow-up question about any part of it.
A trend, not just this week's figure — whether the position is improving, holding steady, or deteriorating.
Currency exposure broken out, not blended into one converted number that hides which currency actually has the risk.
Headroom against any covenant or facility limit, stated plainly rather than requiring the board to do the subtraction themselves.
None of the four is achievable from a single account's statement, or from a snapshot with no history behind it. All four are what a consistent weekly routine, built from real balances, naturally produces as a side effect of just doing it the same way every week.
| A single snapshot | A weekly trend | |
|---|---|---|
| Shows | Where things stand today | Where things have been heading |
| A lender's question | 'Is this typical?' — unanswerable | Answerable directly from the history |
| Board confidence | One number to take on trust | A pattern the board can reason about itself |
What this does not replace
It is worth being precise about the boundary here, because a treasury report that overstates what it covers is more dangerous than one that understates it. This routine is deliberately narrow: it reads what banks say the group holds, on a weekly cadence, so that the finance director always has an accurate starting point. Four things it is not follow directly from that scope.
Not a treasury management system
No sweeps, no payment initiation, no limit enforcement — this is the reporting layer underneath a TMS, not a substitute for one.
Not the statutory close
A monthly close is a formal, audited process. This is the operational weekly view that sits alongside it.
Not a forecast
The report shows what the group holds now, from documents. What it will hold next month is a separate exercise, with its own assumptions.
Not a currency hedging decision
Exposure is shown by currency. What to do about that exposure is a treasury policy question this report informs, not answers.
None of these four limits are a reason to skip building the routine — they are the reason it stays trustworthy. A treasury report that quietly drifted into claiming to be a forecast, or a substitute for the close, would eventually be wrong about something important and lose the board's confidence for reasons that have nothing to do with whether the underlying balances were read correctly.
Why traceability matters more than it seems
Every figure in a treasury report should be able to answer one question instantly: which document did this come from? In a manual spreadsheet, that answer usually lives in someone's memory, or in a folder of PDFs loosely associated with a tab, and finding it again three months later takes longer than re-reading the statement would have.
This matters more than it sounds like it should, for three recurring situations. The first is an auditor, external or internal, asking to see the source for a balance that appeared in a board pack six weeks ago — a request that is routine, not adversarial, but still needs an answer within the day rather than a week of searching. The second is a refinancing or a lender review, where the whole cash position gets re-examined from scratch and every number needs a document behind it, not just the most recent one. The third is simpler and more common: the finance director themself, three weeks later, trying to remember why one entity's balance moved the way it did.
A routine where every balance is read directly from a stored statement, rather than retyped into a cell that then becomes the only record, answers all three automatically. The traceability is not an extra feature bolted onto the reporting — it is a natural side effect of reading the number from the document every time instead of copying it once and trusting the copy afterward.
Starting this week
Take next week's statements — every entity, every account you would normally gather by hand — and read them together as a trial alongside the existing routine, not instead of it yet.
Compare the time it took against a normal week, and check the totals agree. Most finance directors who run this comparison once do not need convincing further; the time difference is usually visible after a single cycle.
For the underlying method this routine is built on, see cash position across accounts and how to build a daily cash position.