Skip to content

Managing client document requests without the chasing

On this page

Managing client document requests well is mostly a question of what the first ask contained, not how firmly you follow up afterwards. A document arrives late for about six reasons, and only one of them is forgetting. A reminder is the correct treatment for that one. The other five are settled before you ever send it.

  • Most requests do not fail on a client’s willingness. They fail on an ask that could not be acted on the morning it landed.
  • A request is not one thing. It is seven items in seven different states, and a follow-up written as though it were one asks again for what already came.
  • Every delivery you acknowledge by name is a chase you never have to send.
  • Sending the same message twice is sending it once. A second attempt has to change the channel, the person or the consequence.

Why the document has not arrived

Chasing gets easier to think about once you stop treating silence as a single condition. A document that has not come is in one of about six states, and they are not equally common.

The first is that it does not exist yet. A broker’s consolidated statement, a partnership K-1, a corrected form that replaces one already issued: the client cannot forward what nobody has sent them. In a practice this is probably the largest single cause of an unanswered document request, and it is entirely immune to reminding. What the file needs is not the document. It is the day the document is expected.

The second is that the client does not know what you are asking for. You used the name the tax code uses. The paperwork on their kitchen table carries the name their bank uses, and the two do not obviously match. Somebody unsure whether the thing in front of them is the thing you meant will usually do nothing rather than send the wrong item, and from your side that looks exactly like being ignored.

The third is that they have it and sending it is a chore. It is paper. It is in a drawer at the office. It is behind a login they created in 2019, where the password reset goes to an address they gave up years ago. The distance between having a document and having sent it is where most of the middle of the year goes.

The fourth is that they sent some of it, which is common enough and misread often enough to have a section of its own below.

The fifth is avoidance, and it repays reading properly. A client who has not sent the bookkeeping has generally not done the bookkeeping. A client sitting on a brokerage statement occasionally knows what is on it. Silence around one specific item, from somebody who answers everything else promptly, is information rather than rudeness, and what moves it is a message that makes the next step smaller, not one that repeats the deadline.

Only the sixth is forgetting.

Laid out that way, the standard follow-up is doing something strange. “Just following up on the below” is a reminder, and a reminder treats one cause in six, and not the most frequent one. Which is why the second copy of it performs no better than the first, and why some firms end up sending four.

The ask that does not need chasing

Five of those six are decided by the first message, so the original request is by a distance the cheapest place to spend effort. A request that can be acted on the morning it lands does not need a chase. One that cannot will need three.

Name each document the way the client’s own paperwork names it, and say who issues it. “The 1099-INT your bank sends in January” describes an envelope on a table. “Interest income” describes a line on a return, which is your vocabulary and not theirs. Where they have sent the same thing before, pointing at that does more than any amount of precision: the same statement you sent last February, for this year. A memory is easier to act on than a definition.

One item per line, and few enough lines that somebody can hold them in their head. A block of nine is not a request, it is a project, and projects get postponed while errands get done. Where the list genuinely is long, the answer is not to shorten it but to stage it: ask for the three that release the most work, and hold the rest until those are in. Running the firm inbox makes the case for the client rather than the message being the unit you read by. The same logic runs one level further down, where the unit is the item rather than the request.

Say what it blocks, once, plainly, with no countdown attached. “We cannot start the return until this arrives” is a fact, costs eight words, and is the only sentence in the message doing any persuading.

Then the line almost nobody writes, and it removes the biggest cause on the list. Ask for a date where the document may not exist yet. “If it has not reached you yet, tell me roughly when you expect it” turns an unknown into an entry with a day against it. A client can answer that from a phone at a bus stop, which is a much lower bar than finding paperwork, and the answer is worth more to you this week than the document would be, because it tells you whether the file is on track without your having to ask again in a fortnight.

Two smaller decisions earn their place. Make sending cheap: a legible photograph of a paper statement, taken at the kitchen table, is worth more to the practice than a properly scanned copy that arrives in April. And keep questions out of a document request. A message asking for four documents that also asks whether they sold the rental gets a reply about the rental, because answering is quicker than fetching, and now both of you believe the thread has been dealt with.

Partial delivery is the normal case

Ask for seven things and you will get three, then two, then nothing. That is not a client behaving badly. It is what happens when somebody works through a list in the order the papers turn out to be findable, and it is how most document requests in a practice actually resolve.

Firms nevertheless track requests as though they were binary. The record says the request went out on the fourth and nothing has come back, or it says the client has replied. Neither is true. The real state is seven items, four of which arrived on the ninth, one of which arrived for the wrong year, and two of which have not been mentioned since.

Tracking at the level of the message cannot hold that, so the second message resends the original list. A client who spent forty minutes on Sunday evening digging out four documents then opens a note asking for all seven again. What they take from it is that their effort did not register, or worse, that what they sent never arrived. The next batch does not come faster. It comes slower, usually with a phone call attached asking whether you received the first lot.

So the unit of a document request is the item. Written down that way a request carries a small state: what was asked, what came, when it came, and what is still out. Everything useful about chasing follows from that state living somewhere other than inside the thread.

It also changes what the record is for. A firm that knows it is missing the K-1 and the vehicle log can answer the client’s own question, which is nearly always some version of what do you still need from me. Answering that in one line, correctly, on the day they ask, does more for the relationship than any amount of well-worded chasing. It is only possible if somebody has been keeping the item-level state as deliveries came in, rather than reconstructing it from a forty-message thread on the morning of the call.

The receipt that chases for you

The single change that reduces chasing most is not a better follow-up. It is acknowledging deliveries by name.

A reply that says what arrived and what is still out does four things at once. It confirms the send, and a client who has been told their statements landed does not send them again next week, nor later have to be argued with about whether they sent them at all. It corrects their copy of the list, which is always less accurate than yours and usually out of date. It is a follow-up that nobody experiences as one, because it opens with what they did rather than with what they did not. And it costs seconds, because it is written against a list you are already holding.

Thank you, that is the bank statements and the mortgage interest in. Still to come: the K-1 and the mileage for the van.

Two sentences, and the second is a chase.

Do that on every delivery and most of your chasing stops being a separate activity at all. It rides on messages you were sending anyway, and it arrives while the client is already thinking about their file, which is the best moment there is to ask them for anything and the one moment a scheduled reminder can never hit.

The record it leaves behind matters too, and that argument belongs to running the firm inbox, which sets out why a practice keeps its mail rather than clearing it. A receipt is the same evidence written from the other end, with the advantage that both of you agreed to it at the time.

When you do have to chase

Some requests go quiet anyway, and for those the rule is short. Sending the same message twice is sending it once. A second attempt that changes nothing has already been read and already been left, and all it adds is the impression that your messages can safely be left again.

So each attempt changes one of three things.

Change the channel. For documents the phone call is worth more than it is for almost anything else you do, because fetching a document is an errand and a call takes the errand off the client’s evening. In ninety seconds you find out which of the six states you are in, you can often identify the document while they are holding it, and where it has not been issued yet you come away with a date. Firms treat the call as escalation. For a document chase it is closer to a shortcut.

Change the person. A married couple filing jointly is two people, one of whom handles the paperwork, and it is not reliably the one who signed the engagement. A business client has an owner who answers nothing and a bookkeeper who answers everything. Writing to the second address is not going over anybody’s head, it is asking whoever has the document. Inside the firm the same move runs in reverse: the partner asking once is a different message from the preparer asking for the fourth time, and everybody involved knows it.

Change the consequence. This is the last rung and the only one with any force in it, which is exactly why it has to be true. A sentence saying what changes and when, of the form that if this is not with us by the twelfth the return goes on extension, works. It works once. Say it, let the date pass, and do nothing, and you have taught that client that your dates are decorative. Every chase you send them afterwards costs more, for as long as they are a client.

The dates in those sentences are not guesses. Running the firm inbox derives the first one from the filing deadline and the work still to do at your end, and inbox zero during busy season explains why the same sum returns a much earlier answer in March than in July. What belongs here is only that the rungs get used in order, and that none of them is used twice.

How fast people reply to anything at all, and why a long and careful message is answered slowest, is measured in unanswered emails. What is particular about documents is that you are not waiting on a decision. You are waiting on an errand, and errands respond to being made smaller far better than they respond to being repeated.

The clients you chase every year

Count the chases at the end of a year. Not the messages: the chases, meaning the second and third and fourth asks for the same item. They concentrate hard in most practices, and the same names come up in roughly the same order every year.

That concentration is the useful finding, because it moves the problem out of the mailbox. Ten clients generating most of your chasing is not a communication problem to be fixed with better wording. It is a fact about ten relationships, and the repair belongs at the start of an engagement rather than in the third week of March.

The repairs are unremarkable and they work. Ask that client earlier than everybody else, on the honest understanding that their reply time is longer. Open with a call rather than an email, since you already know how the email goes. Split their list in two and send the second half only once the first is in. Where the bottleneck is the client’s own records, the conversation is about bookkeeping rather than about documents. And where none of it moves, the honest version is a conversation about terms: what the firm will do, by when, and what happens when the papers are late, which is a question of scope and fees rather than of email.

What makes any of that possible is writing it down as a property of the client rather than of the season. A firm that rediscovers each February that one client needs six weeks and two phone calls is paying for the same knowledge annually. Recorded once, it becomes next year’s plan, and that client gets asked in December.

One caution. The chronic non-responder is not always the client. Often it is their bookkeeper, who is themselves waiting on a bank, or a third party who has no relationship with you at all. Pushing harder down that line achieves nothing. The move is to ask the client to chase the party they have standing with, which is a different message with a different author, and it belongs on your list as something you are owed rather than as something sent.

Where does Point fit?

None of this needs software. Practices chased documents on index cards for decades, and the careful ones did it well. What a tool changes is how much of it a person has to carry: the remembering, the matching and the first draft of the words. The judging stays where it was.

Point reads the mailbox before you get to it and ranks it on importance and urgency rather than on arrival, so a client with a date against them is not sitting underneath a software renewal. The weighing behind that ranking has its own guide. Each thread carries a one-line summary, which on a document thread is most of what you want to know: whether anything came, and whether it was the thing you asked for.

The chase rests on two capabilities rather than one. A request you send is held as something owed to you instead of disappearing into the sent folder, and it returns to you on the date you set rather than the day it occurs to you. And Point notices when the answer has already arrived, marking the follow-up as looking done and waiting for you to agree, so a chase that is no longer needed stands down instead of going out. That second half is what makes an automatic follow-up safe to point at clients at all.

For a named item you are waiting on, a standing request in plain words is enough: you hear once, on the day it lands, rather than checking. Attachments stop living inside threads and gather into a single list, each still joined to the message that brought it, which is the difference between knowing a statement arrived and being able to put your hand on it. Ask a question of a document and the answer comes back with the place in the file it was read from, which counts for a lot in a trade where most of what a client sends you is inside the paperwork rather than in the note above it.

Replies come back drafted the way you write, so the acknowledgment naming what arrived and what is outstanding is an edit rather than a composition, and Point runs the scheduling back-and-forth itself on the days a chase turns into asking for fifteen minutes. How far Point goes on its own is set separately for each kind of action and starts at review, which prepares the work and then holds it for you. The acknowledgment is the one worth raising first. The fourth chase, the one naming a consequence, is not, and wants a person reading it before it goes. Everything done is written down in order with times against it, and an action can be reversed from the line that recorded it, apart from the thing no mail software escapes: a message that has already reached the recipient’s server cannot be pulled back.

Point sits on top of whichever mailbox the practice already has, Google or Microsoft, so the address on your letterhead is unchanged and there is nothing to move.

What Point does not hold is the list. It does not know that one client owes you seven things, it will not tick an item in your tax software, and it has no view of which of the six states a silent client is in. Item-level state stays yours. What a tool carries is the waiting, the arrival and the words. Everything Point does is listed on the benefits page, and Point for accountants takes the same ground through a practice. Whether a tool ought to be reading client mail at all is a separate question with a full answer of its own in is it safe to use AI with client financial data.

Common questions

How often should I follow up on a document request?

An interval is the wrong instrument, because it answers a question about manners and what you need to know is which of the six states the request is in. Where the document exists and the client has it, the date comes out of arithmetic rather than politeness: the deadline less the work still to do at your end once it arrives, which running the firm inbox sets out, and it moves earlier as the queue on the desks grows. Where the document has not been issued yet there is nothing to chase, and the diary entry belongs on the day it is due out. The rule that holds across all of them is that a second attempt identical to the first is not a second attempt.

What should a document chase email say?

It should open with what you already have. A message beginning with the four documents that arrived is read completely differently from one beginning by asking for seven, and that is the difference between a client whose Sunday evening counted and one who suspects it did not. After that, name only what is outstanding, in the words their own paperwork uses, and give the one sentence about what it is holding up. What it must not do is repeat the original list unchanged, which is the commonest form a second chase takes and the reason so many of them need a third.

The client says they already sent it. What now?

Two answers, and the useful one is upstream. In the moment, say what you hold and ask for the rest by name: that settles it without anybody having to be wrong, and reasonably often they did send it, to an address the firm stopped reading two years ago. The upstream answer is that this conversation is nearly always the bill for an unacknowledged delivery. A client told on the day that their statements arrived does not later believe the K-1 went with them, because every receipt you wrote kept their record of what they sent in step with yours.

Would a client portal stop the chasing?

It removes one cause and leaves the other five standing. A portal is a better destination for a document than an email attachment: transport is cleaner, the confidentiality question is easier to answer, and files land somewhere structured instead of inside a thread. It does not make a document exist that has not been issued, tell a client which of their papers you meant, or make a scan feel less like a chore on a Sunday night. Firms that adopt one commonly find the volume of chasing unchanged and the chasing now conducted by email about the portal. Worth having for what it is good at, and not worth expecting to solve this.

Can the chasing be automated?

Partly, and the half that automates is the half nobody enjoys. Remembering is mechanical: the date was decided the moment you sent the request, and a scheduled return does it better than memory does. Wording the second and third attempts is not mechanical, because each one has to change something and what it changes depends on the client. The failure worth naming is a reminder with no idea what already arrived, telling somebody who sent six of nine documents last Thursday that you are still waiting on their paperwork. That is worse than sending nothing, and it is the ordinary result of automating at the level of the message. In Point the return is handled and a follow-up stands down once the reply is in; how much further it goes is set per kind of action and starts at review.

The short version

A document is late for about six reasons and only one of them is forgetting, which is why the fourth copy of a reminder does no better than the first. Most of the repair sits in the original ask: name the document as the client’s own paperwork names it, keep the list short enough to be an errand rather than a project, say what it blocks, and where the thing may not exist yet, ask for the date instead. Track the request as items rather than as a message, because partial delivery is the normal case. Acknowledge every delivery by name, since a receipt that also says what is still outstanding is a chase nobody experiences as one. When you do have to chase, change the channel, the person or the consequence each time, and never name a consequence you will not honour. Then count the chases at year end: they gather on a handful of clients, and those are fixed in December rather than in March. Each of the other recurring jobs in a practice has its own guide in this collection, and AI email for accountants is where the buying question gets sorted out.

Join the private beta

We're onboarding a few teams at a time. Leave your email, confirm it once, and we'll send an invitation the moment a place opens.