A year-end close is one ordinary month-end plus about a dozen jobs that happen once a year, and nearly all of the email comes out of the second half. The routine part is done from records you already hold. The annual part cannot be started until somebody outside the firm tells you something, and a few of those things stop being obtainable on the last day of the year.
- The size of a close’s mailbox has almost nothing to do with the size of the business. It tracks how many annual judgements the accounts contain, which is a different number entirely.
- If a third party recorded it, you can still get it in March. If only the business could have observed it, the year end was the last day it existed, and what replaces it is an estimate somebody has to stand behind.
- This is the one job in the year whose date the firm invents, so the message carrying that date has to name the thing downstream that has a real one.
- A query list is two lists in one envelope: what the client can answer from memory, and what they have to go and find. Sending them together is why it comes back two thirds done.
One month-end and a dozen annual jobs
Split a close in half and the two halves behave nothing alike.
One half is the twelfth month-end. Bank and card accounts reconciled, the feed cleared down, payables and receivables cut at the date, payroll agreed to the ledger, the suspense account emptied. All of it is done from records that already exist and mostly already sit in your own systems, and none of it needs anybody to write to you. It is work, and on a well-kept set of books it is a predictable amount of work.
The other half is the annual set, and it is a list of positions rather than a list of tasks. Depreciation, and whether the assets on the register are still in use or went to the tip in June. Accruals and prepayments. Stock or work in progress at the date. The balance on the owner’s loan or current account, and what the movements on it actually were. A customer who stopped answering in the spring, and whether that debt is ever going to be paid. The private-use proportion on a vehicle, a phone, a room in a house. Opening balances agreeing to last year’s final figures rather than to last year’s draft. And then whatever this business did once and will not do again: took a loan, bought a van, started renting a unit, wrote something off, changed how it gets paid.
Every one of those is a question to somebody who is not in your office, which is why the annual half produces nearly all of the mail. Reconciling a bank account generates no correspondence at all. Deciding whether the debt is bad generates a message, a wait, a chase and a reply, and it does that whether the business turns over a lot or very little.
Two things follow, and the first one is worth saying to whoever plans the firm’s January.
Do not size a close by transaction volume. Size it by the count of annual judgements, because that is what the mail and the waiting are made of. A one-vehicle trade with a room at home, a director’s loan, some stock in a lock-up and a habit of paying for personal things out of the business account will generate more year-end email than a company ten times larger with two bank accounts, no inventory and a bookkeeper. Volume counting the wrong thing is a general property of a practice mailbox, set out in running the firm inbox. The year-end version of it is sharper than most, because the two numbers can point in opposite directions.
The second is that the annual half has no habit attached to it. Monthly work has a rhythm and gets done because the rhythm carries it. Once-a-year work is remembered by whoever remembers it, and it lives in the mailbox rather than in the ledger for the whole time it is unresolved. Most firms have a close checklist, which solves the first half of that problem: the items are enumerated and nothing is forgotten. It does not solve the second, because the items on it are not tasks you can perform. They are questions addressed to named people, and a checklist has no addressee and no date.
None of this changes much with what the close is for. It may end in statutory accounts, in a management pack, in a set of financial statements, or in a trial balance handed to whoever prepares the return. Those differ downstream, in what gets produced and who signs it. The mail in front of them is the same mail.
The inputs that cannot be got later
Almost everything a practice waits for can be obtained late at some cost. A bank statement can be requested again. A loan balance can be confirmed. A supplier will send a copy invoice, a payroll bureau will rerun a report, a merchant service will export a settlement history. It is slow, occasionally it takes a phone call and a password reset, but the document is somewhere and it can be got.
A small number of year-end inputs are not like that, and the line between the two groups is clean enough to state in one sentence. If a third party recorded it, you can still have it in March. If only the business could have observed it, the year end was the last day it existed.
The second group is short and it is always the same. A count of what was on the shelves at the close. Work in progress on a job half finished. An odometer reading. Cash in the tin. Which tools were in the van and which had been written off months earlier without anyone saying so. Materials on site. Any of those can be reconstructed in February, and the reconstruction is a piece of work considerably more expensive than the observation would have been, resting on movements that themselves have to be evidenced, and the result is worse. It is an estimate wearing the clothes of a count.
Which makes the most valuable message of the whole close one that leaves in November, and it is not a document request. It is an instruction to do something on a particular day. It wants three things and no more: what to record, which day to record it on, and how to send it in. A photograph of a tally sheet taken on the shop floor is worth more than a spreadsheet that arrives in March, and saying so in the message removes the main reason people put it off.
It also wants the right addressee, which is rarely the person who signed the engagement. The count is done by whoever is standing in the stockroom. The odometer is read by whoever drives the van. Sending the instruction to the owner and hoping it is passed on adds a link to the chain and a week to the timing, and the general form of that mistake, along with what to do about a line on a list that your client has to obtain from somebody else, belongs to one PBC list, not five copies of it.
Two further notes on the November message, because it is doing more than one job.
It is the last point at which some things can still be acted on rather than merely recorded. What those are is a question about the client’s affairs and not about their mail, so it is not this page’s business. What is this page’s business is only that a message which arrives in February cannot do it, and the difference between the two dates is the whole of the value.
And where the input has already been missed, say so out loud rather than absorbing it. The accounts will carry an estimate, the client should know which figure is estimated and on what basis, and the fact belongs in a note against that client for next year rather than in somebody’s memory. A firm that takes on a new client in January inherits this problem in full, and the honest version of the first conversation includes the sentence that parts of this year will be estimated because the year has already ended.
The one date nobody outside the firm set
Every other deadline in a practice arrives from somewhere else. Filing dates, payment dates, the day a form has to reach the person named on it: none of them are yours, and their being external is what gives them force.
A close date is one the firm chooses. Even where statutory accounts have a filing date, it usually sits months beyond the date you are actually working to, so the date in the diary is still an internal one. This is the only substantial job in the practice’s year with that property, and it explains a failure that firms tend to read as a client problem.
An internal date has no consequence attached to it, and clients can tell. “We would like to have your year end wrapped up by the end of January” is a preference. It is answered at the rate preferences are answered, which is to say after everything carrying a real date has been dealt with, and the firm ends up mildly aggrieved about a message that never made a claim on anybody.
So the instrument is to find the first thing downstream that does have a date, and put that in the message instead of yours. There is usually one. A return that cannot be prepared until the books are closed. A payroll year end. A pack a lender wants, a covenant tested at a date, a facility up for renewal. A grant claim, a valuation, a sale, an investor doing their own diligence. Whichever it is, it is the client’s date rather than the firm’s, it has a cost the client already understands, and it converts a request about your workflow into a sentence about their year.
Where there genuinely is nothing downstream for months, that is worth saying rather than papering over. The cost of a close that drifts is not a penalty. It is a year of numbers nobody can make a decision from, a set of questions asked about a period so distant that the answers get worse, and a larger bill when the work is eventually done at speed. A client told that plainly may still choose to be late, which is their business, and it is at least a decision rather than a drift.
The internal date changes one more thing, and this is where it earns its place in the mailbox. It changes what a chase can honestly threaten. A chase escalates by changing the channel, the person or the consequence, and the consequence has to be one you will actually apply; that ladder is set out in managing client document requests. On a close, the true consequence is almost never a penalty and almost always a scheduling fact: if the records are not with us by the ninth, this moves behind the returns and gets done in March. That is easy to say, easy to mean, and it works because it is true. What it is not is a threat about a deadline you made up, which a client who has dealt with you for three years already knows is soft. Working out what your own side still needs after the papers land, and why that sum gets shorter every week of the season, is inbox zero during busy season.
What a query list is two of
Every close produces the same artefact: a list of transactions the ledger cannot explain, exported by the software as dates, amounts, payees and an empty column. Forty lines, or eighty, or two hundred. It is the most familiar object in the whole job and it is handled worse than anything else in the practice, because it goes out as it came off the screen.
Four things improve it, and the first one happens before anything is sent.
Cut the list inside the firm. A real share of any query list can be settled without the client: the same payee was coded the same way last year, the merchant is unambiguous, or the amount is small enough that no answer changes any number anybody will read. That triage is your work and not theirs, it takes a fraction of the time the correspondence would take, and it is the single largest reduction available. What should leave the office is only the part that nobody but the client can settle.
Then split what remains, because the list is two lists wearing one heading. Some lines are memory questions: what was the payment to the builders’ merchant on the third of August for. A client can answer that in three minutes, standing in a car park, from a phone. Other lines are paperwork questions: we need the invoice behind it. That is an errand with a drawer at the end of it, done on an evening that has to be found. Put both in one message and you get the ordinary result, which is that the easy half comes back on Tuesday and the rest never does, followed by a follow-up that asks for everything again. Send the memory half as a message that can be answered by replying to it, and let the paperwork half join the documents you are already waiting on, where the asking is a craft of its own and unaffected by the items being small.
Third, propose rather than ask, wherever you safely can. Eighty blanks is a form, and a form gets put off. Eighty statements, each ending with an invitation to correct it, is something a person reads through in one sitting, and for the routine lines your proposal will usually be right. “We have treated this as repairs to the unit unless you tell us otherwise” costs the client a glance where a blank column costs them a decision.
That has a limit and it matters. Do not propose where being wrong would change what is claimed, and do not propose where the answer is a judgement about the owner’s own affairs rather than a matter of coding. There, ask plainly, and accept that those lines take longer.
Fourth, order the list by what it changes rather than by date. A ledger export is in date order, which is the one ordering that carries no information whatsoever about which lines matter. Three or four of them usually decide the figures and the rest are tidiness. Say which ones those are, at the top, in the words their own bank statement uses rather than the words your ledger uses. A client who knows which four questions are the real ones answers those four first, which is exactly the outcome you want and the opposite of what a two-hundred-row spreadsheet produces.
One class of line should never be in the list at all. A payment to a family member. Personal spending running regularly through the business account. Cash with no counterpart. A round-sum transfer to an account you have not seen before. Those look like list items and they are conversations, with consequences on both sides and a record that will outlive the engagement. Take them out before the list goes, and handle them as what they are. The same applies to the query that turns out to be an advice question in disguise, which will otherwise be answered in the eightieth row of a spreadsheet by whoever is doing the typing.
Finally, send the list. Do not release it a line at a time as each one occurs to you. A client who receives eleven separate small questions over a fortnight experiences something quite different from a client who receives one message on a Tuesday, and the difference in how fast they answer is larger than it has any right to be. Why a batch beats a drip, and what it does to the count of things you have added to a job, is one PBC list, not five copies of it.
December goes on arriving in January
A close draws a line through a date. The mailbox does not respect the line, and for about six weeks it carries both sides of it in identical envelopes.
The traffic in the second and third weeks of January is a mixture. Invoices dated in December that were posted in January. A card statement whose period straddles the year end. A utility bill covering a quarter that ends in February. An expense claim from somebody who has finally found the receipts. A supplier’s year-end statement. A bank charge applied on the last day and visible on the third. A payroll run paid in January for work done in December. And underneath all of it, the ordinary business of the new year, which never stopped.
The thing your client cannot do is tell you which side of the line any of it falls on. Period allocation is your judgement and it rests on a basis they have no reason to understand. Ask them whether an invoice is for last year and they will answer with when they paid it, because that is the only date they have a feeling about, and it is the date most likely to point the wrong way.
So the question is never about the period. It is about the event. When was the work done. When did the goods arrive. What dates does the bill cover. People can answer events accurately and cannot answer periods at all, which is the same reason the useful January question about paperwork is about what happened rather than what exists.
Second, decide in advance what happens to anything that arrives after you have closed, and make it a rule rather than a case-by-case act of goodwill. There are only three answers: reopen the period, put it in the new year, or hold it as an adjustment because it is large enough to matter. Every firm has a threshold, most have never written it down, and the version the client gets is whatever mood the person answering was in. The mailbox part of it is narrow: the client should be told the rule at the moment the year closes, not after they have sat on an invoice for three weeks trying to decide whether it is worth bothering you with.
Third, a client with access to their own books will post into a period you have finished, and they will not mention it. Where the software allows the period to be locked, lock it, because that is a control and the alternative is not. Where it does not, the only signal available is the client saying something in passing, which is one of several reasons the year should end with a message rather than with silence.
A client who is also on extension has two tax years open at once, which is a related problem with an answer of its own, including the discipline of naming the year in the messages themselves. That is tax extensions and the six months of quiet after them. What is specific to a close is narrower: the ambiguity is not which year a document belongs to, it is which side of a cut a transaction falls, and that is a decision only your side can make.
Approval, and what silence is not
A close ends by sending somebody a set of numbers they did not produce, mostly cannot check, and are being asked to agree to.
The usual version of that message attaches the pack and asks whether there are any questions. It is a request with no action in it, and the ordinary reply is silence, or three words. Neither is approval. What actually happened is that a busy person opened a large attachment on a phone, understood that the figures were the accountant’s work, and had no idea which parts of it were theirs to confirm.
The repair is to separate the confirmation from the presentation. The pack is presentation. The confirmation is a short list of the things only the client knows and on which you have taken a position, put at the top of the message where they will be read.
The balance on the loan or current account with the owner, and whether the movements are what they believe them to be. The value put on stock or work in progress. Whether a named debt is going to be paid. The split between business and private use on the vehicle, the phone, the room. And every figure you estimated because the input had expired, named as an estimate, with the basis in half a sentence.
Three to six lines, each answerable with a yes or a correction, and then the pack underneath for whoever wants it. A client who replies that the loan account looks right and the stock figure is too high has told you something. A client who replies that it all looks fine to a forty-page attachment has told you nothing at all, and the difference lies entirely in what they were asked.
Say what happens if nothing comes back, once, with a date, and then honour it. On a close that sentence is unusually easy to write honestly, because what happens next is that the thing behind the approval does not start, and that is a fact rather than a lever.
Where the close ends in something that has to be signed and filed rather than merely agreed, the chasing and the filing of a signature is a job with its own shape, and it is engagement letters and the signature you are still waiting on. What belongs here is only that a signature is a different act from an approval, and asking for both in one message tends to produce neither.
The message that ends the year
Most closes stop rather than end. The last thing the client hears is a reply to their answer about a query, and then the file goes quiet, and both sides assume the other has finished.
There is a message worth writing at that point, and it is short. The year is closed as at this date. Here is what was adjusted, in a sentence each, and what it changed. Here is what was estimated, and why. Here is what remains open, if anything does. Here is what to do with anything you come across afterwards. And here is who is holding the work now.
It buys four things, three of which are invisible until they are needed.
It stops the year moving. A client who has been told the period is closed, and told plainly to send anything they find rather than deciding whether it matters, sends it. A client who has been told nothing either posts it themselves or sits on it until March.
It is the record. In eighteen months somebody will notice that a figure in the software does not agree with the accounts, and the question will arrive in a mailbox, and the answer is either in a message with a date on it or it is in a reconstruction.
It is the handoff. A close finishes and something downstream starts, frequently with a different person: another team in the firm, the client’s tax preparer, a lender’s analyst, the client’s own bookkeeper picking up the new year. That handoff is an email between two people each of whom quietly believes the other is holding the client, and it is the gap a fortnight disappears into. What it owes is small and entirely specific: the date the books were closed, what carries an estimate, what is still open, and who the client should now be writing to. On that last point, expect no compliance at all. People reply to the address that last wrote to them, which is the reasonable thing to do and is not a habit anybody is going to train out of them, so the arrangement has to work without it; naming the address that holds the record for a client is running the firm inbox. Where the other party is the client’s own bookkeeper, that relationship has a rhythm of its own in working with a client’s bookkeeper.
And it writes next year. The annual judgements this particular business actually has. The queries its books produce every single year, which are remarkably stable. The count, its date and the name of the person who does it. Recorded once as facts about the client rather than as things somebody rediscovers each January, they become November’s message instead of February’s surprise, and that carrying-forward of annotations rather than rows is the part of a rollforward that is worth the effort.
Where does Point fit?
Nothing here is a piece of software. Firms closed years on paper for a very long time, and the careful ones closed them well. A tool only moves where the effort falls: the remembering, the waiting, the matching and the wording of a first draft. The positions themselves stay exactly where they were.
Point reads the mailbox before you do and weighs each message on how much it matters and how soon, rather than on when it turned up, so a client’s answer about the loan account is not sitting beneath a delivery of receipts on a January morning when both arrived. What that weighing measures is a subject of its own. Every thread carries a one-line summary, which on a close is usually the whole question you had: whether something came, and what period it covers.
Three capabilities carry most of the load here.
A standing request in plain words covers the input you are genuinely waiting on. Ask to be told when a named client’s count sheet arrives, or the signed approval, and you hear once, on the day, instead of checking. Against the November instruction, which is set weeks before anything is due to happen, that is close to the whole discipline: leave the watch, then stop thinking about it.
Requests you send are held as things you are owed rather than vanishing into the sent folder, and they come back on the date you chose. Where the reply has already arrived, the follow-up flags itself as likely dealt with and holds until you say so, so nobody chases an approval the morning after it was given. And an ask sitting three paragraphs into somebody else’s message becomes a dated item without anyone lifting it out by hand, which is exactly the shape of the sentence about buying a van in September that turns up in the middle of a reply about something else.
Attachments come out of their threads and gather into one list, dated, each still joined to the message that brought it, and a question put to a document comes back with a pointer to the page it was read from. For a close that matters more than it does elsewhere, because January’s post is largely paperwork with no covering note, and the fastest route from a supplier statement to a period allocation is asking it what dates it covers and being shown where.
Replies come back drafted the way you write, so the query message, the approval request and the message that closes the year start as something you correct rather than something you compose, and Point runs the scheduling back and forth itself on the day a query list turns into twenty minutes on the phone. Each kind of action carries its own setting for how far Point takes it alone, and all of them ship at review: the work gets prepared and then waits for you. The dial explains that setting properly. Acknowledging a delivery is the first kind worth raising. The message asking a client to confirm a stock value is not, and wants somebody reading it before it goes, every year. Every action lands in a log with a time against it, and the log is where you undo one. The limit is the one every mail product shares: once a message has reached the recipient, it has gone.
Point signs in to the Gmail or Microsoft account the practice already runs, so nothing about the address on your letterhead or the archive behind it changes.
What Point does not have is the ledger. It has never seen your trial balance, it holds no close checklist, it cannot tell you which side of the cut a transaction falls on, it has no view on what is material and no idea which annual judgements this particular business contains. Those are the job, and they do not move. Everything Point does is listed on the benefits page; Point for accountants arranges the same list around a firm’s year. Whether a tool ought to be reading mail carrying a client’s records is a question that comes before any of it, and what a firm should satisfy itself about first is where it gets settled.
Common questions
When should year-end emails actually start?
Earlier than most practices send them, and the reason is not politeness. A small set of inputs stop existing at the year end, so the message asking for them has to arrive while the year is still running, which in practice means November. That message is not a document request and should not read like one: it is an instruction to record something on a particular day, addressed to whoever will physically do it. Everything else in the close can be asked for in January without loss, which is worth knowing, because it means the November message can be short enough to be acted on.
How do we get a client to answer sixty queries?
By not sending sixty. Settle inside the firm every line that last year’s coding, an unambiguous payee or a trivial amount already answers, which is usually a substantial share of the list. Split what is left into what they can answer from memory and what they have to go and find, and send those as two different things, because a list containing both gets postponed to an evening nobody has. Then propose an answer wherever being wrong would not change what is claimed, so the client is reading rather than filling in, and say at the top which three or four lines actually decide the figures. What comes back will be a different quality of answer as well as a faster one.
A client sends an invoice dated in December after we have closed. What now?
Answer it against a rule rather than a mood, because the same question will arrive from eleven clients. Reopen, carry it into the new year, or hold it as an adjustment if it is above your threshold: every firm has a position and most have never told a client what it is. Say it once, in the message that closes the year, along with an instruction to send anything they find rather than deciding for themselves whether it matters. And when you ask about the item, ask when the work was done or the goods arrived, never whether it is for last year, because the client will answer that one with the date they paid.
Is silence on a set of draft accounts an approval?
No, and neither is a reply saying it all looks fine, because both are the predictable response to being sent something unreadable and asked an open question. The fix is in the asking. Put three to six specific confirmations at the top of the message, each one a thing only the client knows and on which you have taken a position: the owner’s loan balance, the stock value, whether a named debt is collectable, the private-use split, and anything you estimated. Those can be answered on a phone in a minute, and an answer to them is worth something. A pack with no questions attached to it produces silence in every firm that sends one.
Should the close run in the accounting software rather than in email?
Partly, and it is worth using what the software does well. Most ledger packages can hold a query against a transaction and let the client answer it in place, which removes the transcription and keeps the answer next to the thing it explains. What no ledger holds is the rest of the close, which is most of it: the instruction to count something on a particular day, the conversation about a debt, the estimate you had to make and the reason for it, the approval, and the handoff to whoever picks the work up next. Those arrive and depart as mail whatever else you run, which is why a firm that moves its queries into the software usually finds the volume of year-end email roughly unchanged and the queries out of it. A useful split rather than a replacement, and it is the same shape as the answer about portals in managing client document requests.
The short version
A year-end close is one ordinary month-end plus a dozen once-a-year judgements, and the judgements produce nearly all the mail, so plan the close against how many of them the business has rather than against how big it is. A few inputs cannot be obtained late at any price: if only the business could have observed it, the year end was the last day, which puts the highest-value message of the whole job in November and addresses it to the person holding the tally sheet rather than to the person who signed the engagement. The date you are working to is one the firm invented, so carry the client’s own downstream date in the message instead, and let the honest consequence be a scheduling fact rather than an invented penalty. Cut your query list inside the office before it leaves, split what remains into memory questions and paperwork questions, propose answers where it is safe to, and say which few lines decide the numbers. Expect December to keep arriving through January, ask about events rather than about periods, and write down in advance what happens to something that turns up after the cut. Ask for specific confirmations rather than sending a pack and hoping. Then end the year with a message that says it is closed, what was estimated, what is open and who holds it now, because a close that merely stops is a close that reopens. Every other repeating job in a practice is taken on its own terms elsewhere in this collection. For choosing a tool at all, there is AI email for accountants.