Onboarding a new tax client is the only job in a practice that begins with nothing behind it. Every other item in the mailbox rests on a prior year: last year’s return says which forms to expect, the file says which entities exist, and eleven years of threads say who actually answers. A new client arrives as a message from somebody the firm has never dealt with, and the first six weeks are spent building, out of email, the record that every other job in the practice takes for granted.
- Half of what you need in the first month is not a document. It is permissions, identity and decisions, and none of those behaves like a bank statement.
- The item that decides when you can start is held by the firm the client is leaving, and nobody in that conversation works for you.
- You do not yet know who the client is. The entities, the addresses and the person who can actually instruct you are discovered over the first few weeks, usually from a signature block, and a wrong guess here is a confidentiality event rather than a filing error.
- Nothing marks the end. The return goes out, six things on the original list never arrived, and nobody will ever ask for them again.
The only job with nothing behind it
Take any other recurring job in a practice and notice how much of it is already answered before anybody opens the mail.
You know the client owns two rentals because you depreciated them last year. You know the brokerage statement will be a consolidated one and will arrive in the second half of February, because that is what happened the last four times. You know the wife handles the paperwork and the husband signs whatever she puts in front of him. You know they answer in about a week, that the mileage log is always the last thing, and that the K-1 from the partnership will be late. None of that is written down anywhere in particular. It is the residue of having done the job before, and it is doing an enormous amount of work.
Strip all of it out and you have onboarding.
A new client is not simply a client you know less about. It is a client about whom the firm holds no expectations at all, which means every ordinary inbox judgement stops functioning. You cannot tell whether their silence is normal for them. You cannot tell whether a document is missing or was never going to exist. You cannot tell whether the person replying is the person who decides. You cannot even tell whether their description of their own affairs is accurate, and it usually is not, in the honest direction rather than the dishonest one: people forget the small partnership, the state they moved out of, the account their mother set up.
So the first weeks of a new client are unlike anything else in the mailbox in one specific respect. Everywhere else, mail is an input to a job. Here mail is also the raw material of the client file itself. What you find out in these six weeks becomes the baseline every later year is read against, and if it is wrong, it is wrong quietly for a long time.
That is the reason to treat onboarding as its own job with its own list, rather than as the first stretch of the first return. It has different contents, a different clock, and a different way of failing.
What the first message is already telling you
Onboarding starts before you have agreed to anything, and the first message is doing more work than firms usually let it.
It arrives in the general mailbox, or into a partner’s inbox because somebody passed the name along, and it sits among the vendor pitches and the newsletters. It is usually short. But almost every enquiry contains, without meaning to, the two facts that decide whether this is a client you want: why they are leaving, and what state they are in.
The tells are ordinary and they are worth reading for.
Somebody who leads with a notice has an open problem, not a return. Somebody who mentions that they have not filed for a couple of years is describing several engagements, only one of which has this year’s date on it. Somebody whose first paragraph is about their previous accountant, at length and with feeling, is telling you what the relationship was like from the other side, and it is worth a moment’s thought about whose account you are hearing. Somebody who asks the fee before describing the work is not being rude; they are telling you what will matter in year two. And an enquiry that arrives on the last day of March, from someone with a complicated year, is a specific proposition covered further down.
None of that decides anything on its own. What it does is convert the enquiry pile into a screening pile, which changes how it should be handled. These messages are not waiting on you in the way a client’s message is; they are waiting on a decision, and the decision benefits from one short reply asking three or four questions rather than an immediate quote. What did last year’s return include. Who prepared it. Are there entities, and are they current. Is anything open with a tax authority right now.
Four answers, and most bad engagements are visible in them.
Declining deserves a sentence of its own, because firms are worse at it than they need to be. A decline that is prompt and unexplained is kinder than one that arrives in three weeks with reasons, and reasons invite negotiation about a decision you have already made. Where you can point them somewhere, do. Where you are declining because of capacity rather than fit, say so, because that is a different message and they may be a good client in September.
One more thing lives in this stretch and it causes trouble later. A great many new clients arrive through somebody: an attorney, a financial adviser, an existing client who put their name forward. That person is frequently on the original thread, and if you simply hit reply for the next six weeks they stay on it through the fee discussion, the prior-year returns and the first draft. Nobody intends that. It happens because the thread was already open. The habit worth building is that the enquiry thread ends when the engagement begins: thank the referrer in their own message, and start a new conversation with the client alone.
Two lists, and only one of them has a deadline
Once you have said yes, the mailbox starts carrying two different jobs about the same person, arriving interleaved and looking identical.
The first is the firm’s own list: the things that must be true before you can properly act for this client at all. Depending on where you practise and what your firm’s policy says, that is some combination of a conflict check, whatever identity verification you are required to perform, an engagement letter that has come back signed, the authorisation that lets you deal with the tax authority on their behalf, a decision about scope and fee, and the client’s acknowledgement of how you will handle their information.
The second is the work: last year’s return, this year’s documents, the books, the numbers, the filing.
These two lists behave in opposite ways, and confusing them is the most expensive mistake available in the first month.
The work list has a date on it that somebody else set, and it is loud. If a document is missing, a preparer opens the file, cannot proceed, and the job visibly stops. The firm’s own list has no external date and makes no noise whatsoever. Nothing in the tax software asks whether the conflict check happened. No stage of the return requires the authorisation to be on file. Engagement letters, and the signature nobody chases makes this argument about the letter, and everything in it applies with more force to the whole acceptance list, because the letter is only one line of it and the others have nobody chasing them either.
The practical consequence is the same one that piece reaches: a list nothing surfaces cannot be handled by attention, so it has to be a gate. Which items are absolute, and which a partner may waive by name, is for your firm and your insurer to settle between them. What does not work is a preparer deciding case by case, because a preparer receives no signal at any point that anything is outstanding.
There is one useful ordering rule inside the acceptance list, and it saves weeks. Two of its items are not just obligations; they are keys. The signed letter is what lets your people start. The tax authority authorisation is what lets you go and get information yourself instead of asking the client for it. Both take days or weeks to come back and neither depends on anything else, so both go out on day one, before the document pack, before the questionnaire, before anything you have to think about. Everything else in onboarding can begin while those two are in flight.
Most of the pack is not documents
The onboarding pack is the longest list you will ever send a single client, and it goes to the one person who has never sent you anything and has no habit of doing so.
That combination is worth sitting with. A returning client gets an organiser and knows the drill: same time of year, same shape, and they have done it eleven times. A new client gets a longer list, with unfamiliar words in it, from a firm they have known for a fortnight, at a moment when they are also cancelling something with their old accountant. Firms put real care into the letter that welcomes a new client and then attach forty lines to it.
The first thing to see is that most of those lines are not documents at all. They fall into four kinds, and the kinds travel at wildly different speeds.
Some are sixty seconds on a phone. A signature. A photograph of an identity document. Confirming a date of birth, a dependant’s name, a bank account for the refund. These can be done on a sofa, immediately, and the only thing stopping them is being bundled with something slower.
Some are an evening with their own records. Last year’s charitable receipts, the mileage, the closing balances, the list of assets bought during the year. These are real work for the client and they get scheduled, which means postponed.
Some are held by somebody else entirely. Prior-year returns, the depreciation schedule, the accounting file, the payroll filings. The client cannot produce these no matter how willing they are; they can only ask, and they are asking a firm they have just dismissed. That has a section of its own below because it is nearly always the long pole.
And some are decisions rather than deliveries. Which entity is which. Whether the side business is a business. Whether last year’s treatment of something should carry forward. These do not arrive because you asked; they arrive after a conversation, and putting them on a document list is a category error that guarantees they are the lines nobody answers.
Sending all four as one attachment makes the whole pack move at the pace of the slowest, and the engagement letter guide sets out why that is a bad trade for the quick items. In onboarding it is worse, because there are four speeds rather than two, and because the fast items are the ones that unblock your side.
So the pack goes out in order of who can act and how soon. Signature and authorisation first, alone, with nothing else in the message. Then the ask to the outgoing accountant, which starts a clock you do not control. Then the client’s own records, staged rather than dumped, with the two or three that release the most work named first. The decisions are not a list at all; they are twenty minutes on a call, and a call in the first fortnight with a new client is worth more than it will ever be worth again.
The craft of the individual ask does not change because the client is new. Naming a document the way their own paperwork names it, keeping a request short enough to be an errand rather than a project, saying what it blocks in one sentence, and asking for an expected date where the thing may not exist yet: all of that is managing client document requests, and it is the same here. What is different is only that you have no idea yet which of it they will find hard, so the value of asking for dates rather than assuming is at its highest in a first year.
Where the list stays open for weeks and both sides need to be looking at the same version of it, the machinery for that is one PBC list, twenty replies, and no version of the truth. An onboarding pack is that object, with the added property that the client has never seen one of yours before.
The half of the file somebody else is holding
The single most common reason a new client’s first return is late has nothing to do with the client. It is that the previous accountant has not sent the file.
This item deserves to be treated as its own small project, because it is structurally unlike everything else you are waiting for. The holder has no relationship with you, no obligation to you, and occasionally an unpaid invoice and a bruised feeling. The client is the only party with standing, and the client is embarrassed about the whole thing. The material takes weeks. And it is the material that decides whether the return you prepare is right.
Ask for it on the day the engagement letter goes out, not when you open the file.
What to ask for is worth naming precisely, because a request for “last year’s return” gets you a PDF of the return, which is the least useful thing in the folder. The return tells you what was filed. It does not tell you what the numbers were built on. What you actually need is the working material behind it: the depreciation and amortisation schedules with cost, date placed in service, method and accumulated depreciation for every asset; carryforwards of every kind, which means net operating losses, capital losses, charitable contributions, passive activity losses and unused credits, at both federal and state level; basis in any partnership or S corporation interest, which is the item most often missing and the hardest to reconstruct; the elections that were made and are still binding; estimated payments made and any overpayment applied forward; and for a business, the closing trial balance, the chart of accounts, the fixed asset register and the accounting file itself.
Two practical notes about the request. It should go from the client, not from you, because the client is the one the outgoing firm has a duty to. And it is much more likely to be answered if it is specific: a firm receiving a list of eight named items produces them, while a firm receiving “please send everything” produces a PDF and a delay.
Now the part that changes your planning, which is what to do when it does not come.
Some of it you can get yourself, and this is why the authorisation went out on day one. With the client’s authorisation on file you can obtain transcripts from the tax authority directly, and they cover more than firms expect: what was reported about the client by third parties, what was actually filed, and what was paid and when. That answers a surprising share of the return. Wage and income information fills in across the spring rather than being complete in February, so it is a better fallback in June than in March, which is another argument for getting the authorisation in early.
What transcripts will not give you is exactly the material that matters most: depreciation detail, basis, and the reasoning behind an election. Those either come from the outgoing firm, or they get reconstructed from whatever the client has, or they get estimated and documented as estimated.
That last option is a real one and it should be a decision rather than a drift. If the file has not arrived by a date you choose in advance, you reconstruct, you write down what you reconstructed and how, and you tell the client in writing that this is what happened and what it may cost to correct later. Firms that skip the writing-down step end up, two years on, with a depreciation schedule nobody can explain. The note costs five minutes at the time and is the only thing that answers the question later.
Set that date early and put it in the client’s diary as well as your own. A client who knows on day three that the deadline for their old accountant is the end of February will make a phone call in February. A client who first hears about it in April will make the same phone call, three weeks too late, and will remember it as your delay.
Who the client turns out to be
Everywhere else in a practice, the client is a settled fact: a named entity or household with known addresses and a known signer. During onboarding, none of that is settled, and the mailbox is where it gets decided, largely by accident.
Consider how the facts actually arrive. The enquiry comes from a personal address. The engagement letter is signed by a name that is slightly different from the one on the enquiry, because the legal name has a middle initial and the email does not. The bookkeeper appears in week two from a domain you have not seen. The spouse writes once, from an address nobody recorded, about the dependant care question. And in week four a message mentions the LLC, which nobody had told you about, because to the client it is obviously part of the same thing.
The firm’s record of who this client is gets assembled from those fragments, and the default assembly is a signature block copied into a contact card.
Three facts need to be established deliberately rather than absorbed, and none of them is reliably in the mail.
The legal identity, in full, with entity type. Not the trading name, not the name on the email. This one is boring and it is the one that produces the filing that has to be amended.
Every entity in the picture, including the dormant one. Ask directly and ask twice, in different words, because clients do not volunteer entities they think of as inactive, and an inactive entity is still a filing obligation. The question that works is not “do you have any other entities”; it is closer to “is there anything with its own bank account or its own tax number”, which is a question about the world rather than about tax vocabulary.
Who can actually instruct you, and who can only receive. In a household, the person who does the paperwork and the person who signs are frequently not the same, and the second one has not read a word of your correspondence. In a business, the controller answers everything and may or may not be able to commit the entity. That distinction has consequences for the engagement letter, which the letter guide covers, and it has a separate consequence here: authorising somebody to receive their employer’s tax information is not the same as authorising them to make decisions, and email flattens the two.
Getting this wrong is not an administrative slip. Adding the office manager to a thread about the owner’s personal return, or replying to both spouses on a matter one of them raised privately, is a confidentiality problem, and it happens most often in the first month, when nobody in the firm yet has a settled picture of who belongs to what. The general question of what a practice owes each of those parties is set out in client confidentiality in an accounting practice. What belongs here is only the timing: the risk is concentrated at onboarding, and it is at its worst in exactly the weeks when the firm is trying hardest to be responsive.
The same weeks are the one window where a client will learn how you work. Email management for accountants makes the case that clients reply to whoever wrote to them last, that this is sensible of them, and that no arrangement should depend on training it out of them. That holds. But a new client has no habits with your firm yet, and one or two conventions offered in week one will stick where the same conventions offered in year three will not.
Use the window sparingly, because the size of the ask is what decides whether it holds. Two rules survive. Where documents should go, and who to write to with a question. A page of instructions about mailbox etiquette is read by nobody and quietly teaches the client that your messages can be skimmed.
One more thing to catch while it is cheap. Somewhere in the first fortnight you will find that the work is not quite what was described: a second rental, a partnership interest, a state you had not counted on. At that point the letter is days old and the fee has not been billed, so the conversation is a straightforward one about scope. Six weeks later, with the work half done, it is a much worse conversation. The engagement letter guide covers what to do when scope moves after signature; the onboarding-specific point is that the discrepancy nearly always surfaces in the first three weeks, and that it is worth going looking for rather than waiting to trip over.
The client who arrives in March
New clients do not arrive evenly through the year. They cluster where people become dissatisfied with their accountant, which is immediately after a bad filing season, and where people discover they have a problem, which is when a notice lands or a deadline approaches. So a real share of onboarding happens in the weeks when the firm has least room for it.
The March enquiry is a decision, not an emergency, and it is worth doing the arithmetic in front of you before replying.
Onboarding has a floor made of calendar time that no amount of effort compresses. The letter has to go out and come back. The authorisation has to be signed and processed. The request to the outgoing accountant has to be sent by the client and answered by a firm that is also in season. None of those responds to how quickly your people work. Add them up honestly and the earliest realistic date on which a preparer can open the file is often three or four weeks out, which in March means the file opens after the deadline.
Which is the answer, and it should be said on day one rather than discovered on the eighth of April.
An extension is the ordinary and correct outcome for a client onboarded in the last few weeks of a season, and a client told that in the first conversation hears a plan. The same client told it three weeks later hears an excuse, and told it two days before the deadline hears an emergency. What an extension does and does not buy, and what has to be said in the message that offers one, is extensions, and the second season nobody plans for. What belongs here is only that the sentence is much cheaper in the first reply than in any later one, and that a new client will judge the whole relationship on how early they were told.
Two things follow from that.
The estimate is the hard part, not the extension. A client whose file you have never seen still owes what they owe by the original date, and you are being asked to estimate a liability from material you do not yet hold. That reality shapes the March pack: the items you need for an estimate come first, ahead of the items you need for the return, and they are a much shorter list. Last year’s return, the year’s income sources, and what has already been paid will usually do it.
And capacity is a separate question from fit. Inbox zero during busy season makes the case for deciding in advance what waits, and a new client in March is the clearest example there is of work that can be deliberately placed after the deadline. A firm that has decided, in December, that new clients arriving after a certain date are onboarded for an extension rather than for April is not turning work away. It is making the same decision earlier, in writing, when it can be explained calmly instead of apologised for.
Nobody says when onboarding is over
Onboarding has an obvious beginning and no ending at all, and the missing ending costs more than anything else in this piece.
Here is the shape of it. The pack went out with twenty-two items. Fourteen came back in the first fortnight. Five more arrived over the following month, two of them wrong and later corrected. Three never came: the basis figure the old accountant never sent, a set of receipts the client could not find, and the depreciation detail for a small asset that was worked around. The return was prepared, reviewed and filed, everybody was relieved, and the file closed.
Nobody will ever ask for those three things again. Not because anyone decided to stop, but because next January the client is not a new client any more. The organiser goes out, the year runs normally, and the gap is now simply part of the history. It surfaces three years later when the asset is sold and the basis has to be defended.
So onboarding needs a closing act, and it is a short one.
When the first return goes out, take the original list and reconcile it. What was asked, what arrived, what did not. For each item that did not arrive, one line saying what was done instead: obtained from transcripts, reconstructed from the client’s records, estimated on this basis, worked around and immaterial. That note goes in the permanent file rather than in the year’s folder, because its readers are next year’s preparer, a successor firm, and whoever eventually asks how a number got there.
It costs perhaps twenty minutes and it is the difference between a known gap and a silent one.
Two more things belong in the same twenty minutes, and they are both about not paying for the same knowledge twice.
Write down what you learned about how this client works. How long they take. Which of the two addresses actually answers. That the bookkeeper is the useful one and the owner never opens anything. That the brokerage statement was late because the account is held somewhere slow. Managing client document requests makes the case for recording that as a property of the client rather than rediscovering it every season; the point specific to a first year is that year one is when all of it is learned and the only year in which writing it down costs nothing, because you are already thinking about it.
And decide explicitly that they are no longer new. That sounds like ceremony and it is not. Until somebody says it, a new client stays in a special mode where partners answer personally within the hour, everything is checked twice, and the ordinary process does not quite apply. That mode is expensive and it is also a promise. The service level a client experiences in their first six weeks is the one they will hold you to in their fourth March, and the gentlest way to manage that is to have been ordinary a little sooner. Where a firm publishes response times, that is a topic of its own; here it is enough to notice that onboarding sets an expectation whether or not anybody meant to.
Where Point fits in a first six weeks
None of the above needs software, and firms have onboarded clients well with a paper checklist and a good habit for a very long time. What a tool changes is where the effort sits: the noticing, the holding of what you are owed, and the first version of the words. The acceptance decision, the list and the judgement about what a client actually needs stay exactly where they are.
Start with the honest limitation, because it is specific to this piece. Point reads your mail and gets more useful as it sees more of it, and a brand-new client is the one relationship it knows nothing about either. In the first fortnight the automatic part is at its thinnest. What helps immediately is telling it who this is: the people you deal with get sorted out, duplicate contacts merged and the ones who matter marked, so a couple writing from two addresses reads as one relationship from the start rather than as two strangers. Marking a new client on the day they sign is a ten-second job that pays for itself through the burst.
Through that burst the ordinary machinery does most of the lifting. Messages are weighed on how much they matter and how soon rather than on arrival, so the reply carrying the signed letter is not sitting below a newsletter on a March morning; how triage decides what needs you explains the weighing. Every thread carries a summary in a line, which matters more during onboarding than at any other time, because a first month produces one long, branching conversation with a stranger and the subject line stopped describing it in week one.
Three capabilities fit this job in particular.
Things you are owed do not live in the sent folder. A request you send is held as an open loop with the date you chose against it and comes back to you on that day, and where the answer has already arrived it marks itself as probably settled and waits for your agreement. Across a pack of twenty-two items in four speeds, that is the bookkeeping this job is mostly made of.
A standing request covers the wait you cannot influence. Ask in plain words to be told when the prior-year file arrives, and you hear once, on the morning it lands, instead of checking a mailbox for three weeks about a firm that has no reason to write to you.
And attachments stop disappearing into threads. They gather into one list, each still tied to the conversation it came in on, which is the difference between knowing that a new client sent twenty files from four addresses and being able to find the one you need. An attachment can also be asked a question directly, and the answer comes back pointing at the part of the file it was taken from. Against a sixty-page prior-year return that is precisely the shape of the work: what method was used on the van, what the carryforward was, whether the election is in there.
Search reads for meaning rather than exact words, so what the client said about the rentals in their very first email is findable in the week you discover there are three of them. Replies come back drafted the way you write, so the receipt naming what arrived and what is still out is an edit rather than a composition. And Point runs the scheduling back-and-forth itself, which is what the twenty-minute call in the first fortnight actually needs, since that call is the item most often lost to a thread about diaries.
How far Point goes is a per-action-type setting, and out of the box each one sits at review: the work is prepared, then held until you look at it. The dial covers that properly. Onboarding is a reasonable place to raise the acknowledgment of a delivery and a poor place to raise anything going to the outgoing accountant or anything touching scope and fee, both of which want somebody reading them. Every action Point takes lands in a log, timed and in sequence, and most of them can be taken back from the entry itself. The exception is the one no mail product can offer: a message sitting on somebody else’s server is out of reach.
Connecting is a sign-in to the Gmail or Microsoft 365 mailbox the practice already runs, so the address on your website, and everything already in it, stays where it is.
What Point does not have is the acceptance list. It does not know that this person has become a client, it cannot run a conflict check, it holds no view on whether the identity verification was done, and it will not stop a preparer opening a file. It does not know what should have been in the pack, and it cannot tell you that the entity nobody mentioned exists. Those are the parts of onboarding that are the actual work. Everything Point does is set out on the benefits page, and Point for accountants puts it in the setting of a firm. Before any of that, there is the question of whether a tool belongs near a client’s tax affairs in the first place, which has its own full treatment in is it safe to use AI with client financial data.
Common questions
What should we ask a new tax client for first?
Two things, on the first day, in a message containing nothing else: the engagement letter for signature, and the authorisation that lets you deal with the tax authority on their behalf. Both are quick for the client, both take real calendar time to come back, and both are keys rather than inputs. The letter is what lets your people start; the authorisation is what lets you obtain information yourself instead of asking. Everything else in onboarding can proceed while those two are in flight, and every week they are delayed is a week added to the end. The document pack, the questionnaire and the conversation about scope all come afterwards.
The previous accountant will not send the file. What can we do?
Keep the request with the client, because they are the party the outgoing firm owes a duty to, and make it specific rather than general: a named list of eight items is produced far more often than a request for everything. In parallel, use the authorisation you obtained on day one to get what you can directly. Transcripts will tell you what was reported about the client, what was filed and what was paid, which covers a useful share of a straightforward return, though the third-party information fills in across the spring rather than being complete in February. What they will not give you is depreciation detail, basis or the reasoning behind an election. For those, set a date in advance at which you stop waiting and reconstruct, tell the client in writing what you reconstructed and how, and keep that note in the permanent file.
Should we take on a new client in March?
Often yes, provided you and they agree on the same day what taking them on means. Onboarding has a floor made of calendar time: the letter out and back, the authorisation processed, and a request answered by a firm that is also in season. Add it up and the file frequently cannot be opened before the deadline, which makes an extension the ordinary outcome rather than a failure. Said in the first reply that is a plan; said in April it is an excuse. Deciding the cut-off in December, as part of what the firm has already agreed will wait, is what makes the conversation calm. Extensions covers what one does and does not buy.
How long should onboarding a new client take?
Measure it from signature to the day a preparer can open the file, and the honest answer for most firms is three to six weeks, of which very little is your own work. The long items are all waits: a signature, an authorisation, and a file held by somebody with no reason to hurry. The way to shorten it is not to work faster but to start those three on day one and to stop bundling them with anything slow. A firm that sends the letter, the authorisation and a forty-line document pack in one message has converted three fast items into one slow one.
Is it safe to ask a new client to email us their identity documents and tax numbers?
That is a decision about your firm’s own obligations and your data-protection duties rather than an email question, and it is worth settling before a client asks you about it rather than after. What the mailbox contributes is only this: onboarding is the point in the relationship where the most sensitive material moves in the shortest time, from someone who will do whatever is easiest, and whatever channel you offer in week one is the channel they will use for the next decade. Client confidentiality in an accounting practice works through what a practice owes here, and if you are weighing what any tool should be allowed to see, the questions to put to a vendor is the list to send first.
Would client onboarding software fix this?
It fixes more of onboarding than a portal fixes of chasing, which is worth saying plainly. A structured intake collects the identity details, the entity information and the signatures in one pass, keeps them somewhere better than an attachment, and gives you a completion state a mailbox cannot produce. What it does not touch is the three hardest parts. It cannot make the outgoing accountant reply. It cannot ask the second question that reveals the entity nobody mentioned. And it does not handle the conversation, which is where scope, fee and the decisions actually get settled, and which will be in your mail whatever else you buy. Treat it as a better container for the collectable half and keep the waiting, the chasing and the judgement where you can see them.
The short version
Onboarding is the only job in a practice with no prior year behind it, so the mailbox is not just carrying the work, it is building the client record that every later year will be read against. Read the enquiry as a screening item, because why they are leaving and what state they are in are usually both in it, and end that thread before the referrer sees the fee. Keep the acceptance list separate from the work list: it has no external deadline, nothing surfaces it, and it therefore has to be a gate rather than a habit. Send the signature and the authorisation on day one, alone, because they are keys and they take calendar time. Get the request to the outgoing accountant moving the same week, ask for the working material rather than the return, and choose in advance the date you will stop waiting and reconstruct. Establish the legal identity, every entity and who can actually instruct you, deliberately rather than from a signature block. Take the March client on an extension agreed in the first reply. Then close the thing: reconcile the original list, write one line against every item that never arrived, record what you learned about how this client works, and let them become an ordinary client before your first-month service level becomes a promise. The rest of this collection takes the jobs that come round every year; if the question underneath is which tool a firm should be running, that one is settled in AI email for accountants.