An afternoon once, twenty minutes after that
Ask most finance teams why they do not build a daily cash position and the honest answer is rarely “we do not see the value”. It is “the first time we tried it took three hours and we never found the time again”.
That first attempt is genuinely slow, and there is no way around it: the account list has to be written down for the first time, the fastest source for each account has to be worked out by trial, and the routine itself does not exist yet. All of that is fixed cost, paid once.
What this guide is about is paying that cost deliberately, in a way that makes the second attempt faster than the first, and the tenth attempt a twenty-minute habit rather than a dreaded afternoon that keeps getting postponed.
Why daily, and not just monthly
A monthly position is enough for a business with months of cover and no immediate pinch points. It tells the story at a pace that matches how slowly the story actually changes.
Daily earns its keep in a narrower but common situation: when payroll, a large supplier payment and a tax instalment can land in the same week, and the question stops being “how are we doing” and becomes “will this specific payment clear”. A monthly position answers that question with data that is, on average, two weeks stale.
| Cadence | Answers | Right for |
|---|---|---|
| Monthly | How is the business trending? | Comfortable cover, few near-term pinch points |
| Weekly | Are we on track for the month? | Moderate cover, seasonal swings |
| Daily | Will this week's payments clear? | Thin cover, payroll or large payments due soon |
Most businesses do not need daily forever. A thin-cover period that needs it for six weeks does not need to keep it up for the following six months — the routine below is built to be picked up and dropped without losing the account list or the sources each time.
That last point matters more in practice than it sounds. The costly part of a daily routine was never the daily reading itself — it was the setup: writing the account list, working out the fastest source for each account, deciding the cutoff and the stale-balance rule. None of that decays when the business steps back down to weekly or monthly reporting for a calmer stretch. Keep the account list current even while it is not being read daily, and the next time cover tightens, the routine restarts from step five instead of step one — the difference between resuming a habit and rebuilding one from nothing.
Three decisions before you touch a single statement
Skip these and every subsequent attempt inherits whatever was decided by accident on day one. Fix them first and everything after is mechanical.
The cutoff time
The single moment of day every position is built from. Not '9-ish' — an actual time, because a position built at 9am one day and 5pm the next is not comparable to itself.
The account list
Every account, written down once, including cards, processors and anything held abroad. This list changes rarely and should live somewhere permanent, not be rebuilt from memory each morning.
The stale-balance tolerance
How many days old a carried-forward balance for a non-daily account is allowed to be before it gets flagged rather than silently reused.
The nine steps
Fix the cutoff time
Pick one time of day the position is always built from, and hold to it — a moving cutoff makes day-to-day comparison meaningless.
Write the account list once
Every account the business holds money in, including cards, processors and foreign accounts. This is the step most often skipped and the one that determines correctness.
Decide the source per account
A live feed, a downloaded report, or a statement — whichever is fastest and most reliable for that specific account.
Set the stale-balance rule
For accounts that do not report daily, decide how many days old a carried-forward balance is allowed to be before it is flagged.
Build the first one by hand
Slowly, checking every figure against the source. This is where mistakes in the account list and the sources get found.
Add the continuity check
Each day's opening should equal yesterday's closing for every account with a genuine daily balance.
Add the internal-transfer check
Flag matching debits and credits across your own accounts before totalling, so money moving between them is never counted twice.
Template the routine
Same file names, same order, same checklist — so day two takes a fraction of day one's time.
Reconcile weekly against the bank
A short weekly check that the daily figures still agree with what the bank actually shows, catching drift before it compounds.
Steps one to four are done once, or revisited only when something genuinely changes — a new account, a bank switch, a change in how thin cover has become. Steps five to nine are what happens every single day, and the whole point of doing one and two through four properly is that five through nine stay fast.
The template itself, column by column
A concrete template beats a vague description of one. The columns below are what most daily positions converge on independently, because each one earns its place by answering a question someone actually asks.
| Column | What it answers |
|---|---|
| Account name | Which account, in words a human recognises without decoding an internal code. |
| Currency | What unit the balance is stated in, before any subtotal ever combines accounts. |
| Balance | What the account held, exactly as its source stated it, as printed. |
| As-of date | When this figure was actually true — the one column that separates a fresh reading from a carried-forward one. |
| Source | Feed, statement or downloaded report — where the number came from, in case it needs checking. |
| Fresh or carried | A single flag distinguishing today's genuine reading from yesterday's figure reused because nothing new was available. |
| Notes | Anything that needs a human's attention — an unresolved transfer, an unusually large movement, a source that failed to load. |
Seven columns, one row per account, and the total is a single sum at the bottom of each currency's balance column — not the whole sheet summed into one blended figure, for the same reason the rest of this guide keeps currencies separate throughout. Everything above the total row is a reading; the total itself is the only line that is genuinely arithmetic.
Two columns are easy to skip on a first attempt and both are mistakes worth avoiding deliberately. Dropping “source” saves a few seconds of typing per row and costs whoever checks the figure later a much longer search to find where it came from. Dropping “fresh or carried” makes the sheet look tidier and makes it impossible to tell, three weeks from now, which numbers were actually verified that day and which were simply repeated.
Who owns which part of the routine
A one-person routine works fine at small scale and becomes a single point of failure the moment that person is on leave, in back-to-back meetings, or leaves the role entirely. Splitting the routine across two roles, even informally, is the cheapest insurance against that.
The builder
Gathers each account's balance by the cutoff, applies the stale-balance rule where needed, flags anything unusual in the notes column, and produces the completed table.
The reviewer
Runs the daily checks — continuity, transfer resolution, every account present — and is the one who actually reads the figure before it reaches anyone else.
In a small team these can be the same person on most days, and that is fine — the value of naming the two roles separately is not that they must always be different people, it is that on the day the builder is unavailable, the routine has an obvious second person who can step in without first reconstructing what the routine even involves. A routine that only lives in one person's head is not a routine; it is a habit that happens to be documented nowhere.
Weekends, holidays and accounts that do not report daily
A pure daily routine hits its first real complication almost immediately: not every account has something new to say every day.
No bank movement over a weekend
Carry Friday's figure forward. Nothing posted, so there is nothing new to read — reconstructing a fresh figure with no new information is theatre, not accuracy.
An account that only updates weekly or monthly
Carry its last known balance forward, marked with the date it is actually as of, until a fresher one exists.
A savings account with a fixed rate applied periodically
Update it on the day the statement or notice confirms the change, not daily by estimating accrued interest.
A processor with a rolling payout cycle
Read its own balance report directly rather than trying to infer it from bank deposits, which lag the actual balance by the payout delay.
The common thread: a carried-forward figure is fine, as long as it is visibly carried forward rather than presented identically to a figure that was actually read today. The distinction costs one column and saves a great deal of misplaced confidence later.
A worked week
Five accounts, one operating week, cutoff fixed at 9am. Two accounts report daily, one weekly, one is a card with a running balance, one is a processor on its own cycle.
| Day | Operating | Card | Processor | Savings (weekly) |
|---|---|---|---|---|
| Mon | 48,220 | -3,140 | 6,880 | 62,000 |
| Tue | 51,905 | -3,340 | 7,120 | 62,000 (carried) |
| Wed | 49,760 | -3,890 | 7,410 | 62,000 (carried) |
| Thu | 52,300 | -4,015 | 7,655 | 62,000 (carried) |
| Fri | 47,880 | -3,220 | 8,010 | 62,340 (fresh) |
The savings column is carried forward Tuesday through Thursday and marked as such, then refreshed Friday when its own weekly statement actually arrives. Nobody reading this table on Wednesday is misled into thinking the 62,000 was checked that morning — the label says otherwise, honestly.
Notice too that the card's negative balance is carried as printed, not flipped positive or excluded — it is real money the business owes, moving in its own direction from day to day just like the others.
Common mistakes
Starting from memory instead of a written list
Whatever account gets forgotten on day one stays forgotten, because nothing in a memory-based routine ever prompts anyone to add it.
A moving cutoff time
A position built at whatever time was convenient each day cannot be compared day to day, even though every individual reading is accurate.
Silently reusing stale balances as if they were fresh
The single most common way a position looks more current than it is — and the easiest to fix with one extra column.
Never confirming flagged internal transfers
Left unresolved, they either double-count money moving between your own accounts or drop it from the total entirely.
Rebuilding the routine from scratch every time thin cover returns
If the account list and sources are thrown away between busy periods, every restart pays the full afternoon cost again.
Converting multi-currency balances into one blended figure
Looks tidy, hides which currency actually has the shortfall, and bakes in a rate choice nobody remembers making by next week.
Most of these six share a root cause worth naming directly: each one is a shortcut that looks harmless the first time it is taken and compounds the second, third and tenth time it is repeated without anyone noticing the pattern. A memory-based account list is fine once. A moving cutoff is fine for one day. What turns a shortcut into a real problem is never the single instance — it is the routine quietly normalising it as how things are done here.
Best practices
Keep the account list in one place everyone building the position uses, not copied into a dozen personal spreadsheets that drift apart.
Record the as-of date on every single balance, even the ones read at the exact cutoff — it is the only way to spot a stale one later.
Review the account list monthly, whether or not anything obviously changed — accounts get added and closed more often than people remember to update a list.
Keep the daily file format identical day to day, so any tool or template built around it never breaks on a formatting surprise.
Reconcile the daily figures against a proper monthly close, so small daily reading errors get caught before they compound across a quarter.
Making it a routine, not a recurring task
The difference between a routine and a task that keeps sliding to tomorrow is almost entirely about how much of it requires a decision versus how much is mechanical.
A decision — which accounts count, what rate to use, whether a transfer is genuinely internal — takes judgement and cannot be rushed without risk. A mechanical step — opening a statement, reading a balance, typing it into a row — takes no judgement at all, and is exactly the kind of work that slows down disproportionately when done by hand, tired, at the end of a long day.
The practical move is separating the two kinds of work permanently: make the decisions once, during setup, and reduce every daily repetition to the mechanical part. Reading a PDF statement's closing balance, date and account identifier is exactly that kind of mechanical step, and it is the one most worth removing friction from — by hand, from a PDF, it takes minutes per account; read automatically, it takes seconds, with the source file kept alongside the figure for anyone who wants to check it later.
The weekly and monthly checks
A daily routine is fast precisely because each day does not re-verify everything from scratch. That makes a periodic proper check essential — without one, a small daily reading slip compounds silently for weeks.
| Check | Frequency | Catches |
|---|---|---|
| Every account appears, once | Daily | A forgotten or duplicated account |
| Opening equals yesterday's closing | Daily | A missed update or a typo |
| Every flagged transfer resolved | Daily | Double-counted or dropped internal movement |
| Daily figures agree with the bank's own statement | Weekly | Drift between quick daily reads and the source of truth |
| Full position reconciles to the formal monthly close | Monthly | Anything the weekly check was too quick to catch |
None of the five checks is arithmetic — adding balances correctly was never the failure mode. Completeness, staleness and double-counting are, and all five checks exist to catch one of those three.
Assign a rough time budget to each and the daily checks together add perhaps two minutes to the twenty-minute routine — a small tax worth paying every day for the much larger savings it produces at the weekly and monthly stages. The weekly check against the bank's own statement typically takes ten to fifteen minutes once the daily table already exists; without it, the same comparison done once a month takes considerably longer, because a month's worth of small drifts have to be traced back individually instead of one week's worth.
What happens when the weekly check gets skipped for a few weeks is instructive. Nothing goes visibly wrong at first — each day's figure still looks plausible on its own, because a small reading slip on any single day is rarely dramatic enough to notice. What accumulates instead is a slow divergence between the daily table and what the bank's own records actually show, one that stays invisible precisely because nobody is comparing the two. By the time the monthly close finds the gap, it is often several weeks of small errors tangled together, and untangling which day introduced which error takes far longer than the weekly check that would have caught each one within days of it happening.
What you actually need
Less than it looks like at the outset. Three things, and none of them is exotic.
A place to keep the account list
A spreadsheet tab is enough. It needs to be permanent and shared, not personal and easily lost.
A fast, reliable way to read each account's balance
A live feed where one exists; a downloaded report or a statement everywhere else. A document reader that extracts the closing balance and date removes the slowest part of this by hand.
A template for the daily output
One row per account, the same columns every day, so nothing built on top of it — a chart, a dashboard, a weekly summary — breaks on a formatting surprise.
What none of the three needs to be is expensive or complicated. Plenty of daily positions run entirely on a shared spreadsheet, a folder of downloaded statements, and a habit repeated at the same time each morning — no dedicated software, no integration project, no procurement process. The routine, done right, is what makes the daily figure trustworthy — not the sophistication of whatever produces it. Where a document reader earns its place is purely at the second of the three: replacing the slowest, most error-prone step with one that takes seconds instead of minutes per account, without changing anything else about how the routine works. For the underlying concept this routine builds toward each day, see cash position across accounts; for what the running shape of one account over many of these daily positions looks like, see balance timeline.
The three pieces above are also the three worth revisiting if the routine ever starts feeling slower than it used to — a stale account list, a source that stopped being the fastest option, or a template that quietly changed shape are the usual culprits, in roughly that order of likelihood.
Frequently asked questions
If a question about your own setup is not covered above, the same nine steps still apply — the specifics change with the number of accounts and currencies involved, but the underlying method of fixing a cutoff, listing accounts once and checking continuity daily does not.