Stop chasing clients for documents: what actually makes a request land

An accountancy practice is unusual among service businesses in that it cannot start most of its work until the client hands something over. A solicitor can draft. A designer can sketch. A bookkeeper looking at an unreconciled month with no bank statement can do precisely nothing, and no amount of capacity in the office changes that.

The real cost of chasing, then, is that work which could have been spread across four weeks arrives in the last four days, in a queue, behind a fixed deadline. The minutes spent sending the message are the visible part and the smaller one. Fixing the arrival date is the job worth doing, and it is a different job from working faster.

What we'll cover

• Blocked is not the same as busy
• What makes a request satisfiable
• Why clients actually don't send
• Knowing what has arrived
• Building it with CodeWords
• The ladder, and where it stops
• When it arrives in the wrong form
• Some of these records have to be digital anyway
• Making it survive
• Limits worth knowing about
• What to build first
• Frequently asked questions

Blocked is not the same as busy

Take a monthly bookkeeping client whose records are due on the 5th and arrive on the 26th. Nothing about the work got harder. Three weeks of possible working time were simply unavailable, and the job now competes with every other job that also arrived late.

An earlier arrival is worth more than a faster chase. Moving a client from the 26th to the 12th beats reducing the time you spend asking, even though the asking is the part that feels annoying.

The spread matters as much as the average. A practice where everything lands on day 20 can staff for day 20. One where records arrive anywhere between day 3 and day 28 cannot plan at all, and it is the variance that produces the evening work.

So the number worth tracking is the date each client's records actually land, month by month, and whether it is moving. Reminders sent is a measure of your effort, and effort is the thing you were trying to reduce.

What makes a request satisfiable

Most requests that get ignored are not ignored out of indifference. They are ignored because the client could not act on them in the moment they read them, and a request that cannot be acted on immediately gets deferred to a later moment that never arrives.

A satisfiable request names four things.

The specific document. Not "your records" or "the usual". The March statement for the business current account, the payroll hours for the period ending the 28th, the completed mileage log. A client with three accounts and two cards cannot act on "your bank statements" without first asking you which.

The period. Obvious to you, ambiguous to them, particularly where your year end and their mental model of the year do not coincide.

The format you can actually use. A PDF export rather than a photograph of a screen. If you will only have to ask again in a different form, say so the first time.

Where it goes. One destination, named. A request that ends without a route ends in an email to somebody's personal address three weeks later.

The test is whether somebody could satisfy it from their phone, standing in a car park, in under two minutes. The requests that get answered are the ones written as though they will be read that way.

Why clients actually don't send

Worth separating, because the responses are different and firms tend to apply one treatment to all three.

They do not have it. The statement is behind a login they have not used in two years, or the receipt genuinely does not exist. Chasing harder produces nothing.

They have it and it is not urgent to them. Your deadline is not their deadline, and nothing in the request explained what it blocks. This is the group a staged reminder actually moves.

They are avoiding it. The records are a mess, or they know what the numbers will say, or the last fee conversation was uncomfortable. More messages make this worse.

A practice that treats all three identically sends its most persistent reminders to the clients least able to answer them, which is both ineffective and the thing that damages relationships.

Knowing what has arrived

This is the part most firms are missing and the part that makes everything else possible: a list, per client per period, of what is expected and what has landed.

Without it, "have we had it?" is answered by somebody searching an inbox, slowly and sometimes wrongly. Reminders then go to clients who have already sent, which is the fastest way to lose the goodwill you need for the next request.

It does not need to be sophisticated. Four columns will do: what is expected, whether it arrived, when, and who confirmed it. It must be visible to the whole practice rather than living in one person's head, and it must update on arrival rather than when somebody remembers to tick it. A reminder is a function of that list, not of the calendar.

Building it with CodeWords

CodeWords connects to more than 3,000 integrations, and the connection is made once and reused.

1. Open CodeWords and start a new automation.
2. Describe what should happen in plain language to Cody, the automation builder: which clients owe what each period, the wording of each request, the timing between them, and the point a person takes over.
3. Authorize the connections to your practice management or ledger system, the channel your clients use, and wherever the expected-versus-received list lives.
4. Describe the exceptions: a client who has already sent, one who has replied asking a question, one you have agreed different dates with, one in the middle of a difficult period.
5. Run it alongside your current process for one cycle rather than switching, and compare what the list says against what the inbox says.

You describe the outcome; Cody builds it, connects it, and deploys it. The free plan covers light use, with Pro at $39 per month and Business at $100 per month as usage grows; details are on the pricing page.

The ladder, and where it stops

A sequence that sends the same message three times teaches clients that the first two can be ignored. Each rung should change in kind, not in volume.

First contact is informational and early. What you need, by when, why that date. No urgency, because there is none yet.

Second acknowledges time has passed and offers help. This is where you make it easy to say "I can't find it", because most clients will not volunteer that unprompted.

Third states the consequence plainly and without drama: what happens to their filing, their deadline or their fee if the records arrive after a given date. One accurate consequence does more than three increasingly tense reminders.

Then it stops being an automation. A client who has ignored three requests has told you something, and the next step is a person deciding what to do: a call, a changed deadline, occasionally a conversation about whether to continue acting. That is a commercial judgement. Practices that get this wrong let the ladder run forever; the ones that get it right treat its end as a tracked, dated trigger for a human decision.

When it arrives in the wrong form

Plan for this, because it is most of the traffic.

A client sends eleven photographs of one statement. A client sends the right month for the wrong account. A client sends a screenshot of a banking app showing a balance rather than transactions. A client replies "sent it to Dave last week" and Dave left in March.

Each has technically been answered and substantively has not. Treat arrival as opening a check rather than closing the item: mark it received pending review, acknowledge it, and let a person decide whether it is usable. Closing on receipt produces a list that looks clean and is wrong.

For anything sent repeatedly, fix the form once. Five minutes showing somebody how to export a statement properly removes a monthly irritation for years.

Some of these records have to be digital anyway

This is a constraint rather than a preference. Under Making Tax Digital for Income Tax, sole traders and landlords over the qualifying income threshold must keep digital records and report quarterly through compatible software, starting from April 2026 above £50,000, from April 2027 above £30,000 and from April 2028 above £20,000, judged on qualifying income two tax years earlier (HMRC guidance).

For an affected client a shoebox is not a slow way of doing it correctly; it is a way of not meeting the requirement. Retention is longer than people expect, too: at least five years after the 31 January submission deadline for the self-employed, six years from the end of the financial year for companies (self-employed, company). Anything collected through a messaging channel has to end up somewhere that survives a phone being replaced.

Making it survive

Check the list before every send, so nobody is chased for something they have already provided.

Keep agreed dates where the whole practice can see them, including the ones you have varied.

Stop the sequence on any reply and route it, rather than sending alongside a live conversation.

Log what was requested and when, because where a filing is late that record matters.

Report arrival dates, not message counts. Whether records are landing earlier than last quarter is the only question that tells you this worked.

Limits worth knowing about

A reminder cannot create a document that does not exist, and some of your chasing is for records the client never had.

Client-side chaos is not fixable from your end, only by changing how they keep records, which is a separate engagement and sometimes a separate fee.

Fixed deadlines do not move. Earlier arrivals help; they do not create capacity in the last week of January.

Some clients only respond to a phone call, and identifying them quickly beats trying to convert them.

The channel matters less than the specificity. A precise request by email beats a vague one anywhere.

What to build first

The expected-versus-received list. Not the reminders, not the sequences, not the integrations, just a per-client, per-period record of what you are waiting on and what has landed, visible to everybody, updated when things arrive.

Two habits make the difference. Write requests that can be satisfied in two minutes from a phone, because that is where they are read. And put a hard end on the ladder, because the chase that never ends is the one that costs you a client you could have kept by ringing them in week two.

Frequently asked questions

Why do clients ignore document requests?

Usually because the request could not be acted on at the moment it was read. Vague wording, a missing period, no format and no destination all push the task to later. A minority are avoiding the conversation, which is a different problem.

What should a document request contain?

The specific document, the period, the format you can use, and one place to send it. If it cannot be satisfied in two minutes from a phone, it will be deferred.

How many reminders should a sequence send?

Three, each different in kind: informational, helpful, then stating the consequence. After that a person decides what happens next.

How do I stop chasing clients who have already sent?

Keep a per-client, per-period list of what is expected and what has arrived, and make the reminder a function of that list rather than of the date. Reminding somebody who already complied is the fastest way to lose their cooperation.

Should a document be marked complete when it arrives?

No. Mark it received pending review. Much of what clients send is the wrong account, wrong period or an unusable format, and a list that closes on receipt looks clean while being wrong.

Does Making Tax Digital change how I collect records?

For affected clients, yes. Digital record-keeping and quarterly reporting through compatible software are required above the qualifying income thresholds, so paper-based collection is not a slower route to the same place.

How long do these records need to be kept?

Self-employed records for at least five years after the 31 January submission deadline for the relevant tax year; company records for six years from the end of the financial year they relate to.

What should I measure?

The date each client's records actually arrive, month by month, and whether the spread is narrowing. Message counts tell you about the automation, not about the practice.

Related reading

• Follow-up sequences and when to stop
• Receipts, records and what HMRC needs
• Which channel is the client file