How do you chase a client for missing documents without annoying them?

Name each document with its date and amount, shorten the list at every reminder, cap the number of messages and stop all reminders as soon as the document arrives. After two unanswered reminders, a call from a team member works better than a third email.

How many automatic reminders before handing over to a team member?

Two automatic reminders are enough in most cases before handing over to a person. Treat the number as a starting point to adjust to your clients: what matters is that the limit exists and that the file escalates without anyone having to remember.

Can AI check that a document received is the right one?

It can read the document and compare it to the expected line: supplier, date, amount. It flags what does not match or cannot be read. Ambiguous cases go to a person, who decides.

Do we need to change accounting software to automate collection?

No. The loop plugs into the mailbox, a shared folder and your current software. Changing tools multiplies the risks without improving collection itself.

How much time can an accounting firm save on collection?

It depends on the number of clients and their habits. One vendor publishes more than two days a month from its partner firms’ data, but that is a vendor figure. Measure a real period in your own firm before deciding.

Accounting firms: collect client documents without chasing ten times

Why collecting documents costs more than the time spent chasing

Direct answer: to collect client documents without chasing ten times, replace improvised follow-ups with a fixed loop: a list of expected documents per client, a single drop-off channel, a short reminder cadence that stops on receipt, and hand-over to a person after two automatic reminders. AI keeps the list, drafts the reminders and checks what arrives. Your team keeps the sensitive cases.

In an accounting firm, chasing a client is never an isolated action. Someone has to find what is missing, open the right file, write a message, wait, and start again. Each reminder interrupts a team member who was doing something else, and getting back to the earlier task takes time. The real cost is not the length of the message but the fragmentation of the day. Then add delay: an incomplete file blocks data entry, entry blocks the close, and the close ends up squeezed into the last days of the period.

Firms know the problem. Qonto reports, from its data with partner firms, that cutting collection and follow-up work can free up more than two days a month. That is an order of magnitude published by a software vendor, not an independent time study: measure your own firm before relying on it. This article builds on our overview of accounting firm tasks worth automating by digging into the one that comes first: collection.

Why clients do not answer the first request

Before configuring anything, understand why a client lets it slide. It is rarely ill will. Four causes come back again and again.

The request is vague. “We are missing a few receipts” forces the client to guess which ones. A request that names each document, with date and amount, gets answered faster than a general one.
There are too many channels. One document arrives by email, another by WhatsApp, a third as a photo, a fourth by post. Nobody knows what has already been received, or where.
The list is too long. A client who receives twenty lines puts it off. A client who receives three deals with it the same day.
Nothing says what happens if they do not answer. Without a clear deadline or a concrete consequence, the request sinks to the bottom of the pile.

Qonto also notes that e-invoicing will not remove the “shoebox” effect on its own: as long as clients send documents in bulk, sorting and chasing remain the firm’s job. It is the collection loop that has to change, not only the data entry tool.

The collection loop in five parts

Firms that cut their reminders share the same mechanism. Here are its five parts, in the order we set them up.

A list of expected documents per client profile. A freelancer, a company with employees and a dormant company do not need the same documents. The list is built once per profile, then adjusted per client. It is what lets you know, at any moment, what is missing.
One drop-off channel per client. A dedicated email address or a shared folder you already use. Anything that arrives elsewhere is redirected to that channel, not handled separately.
A completeness check on receipt. The AI reads the document, compares it to the expected line and marks it as received, unreadable or off-target. A document that does not match what was requested does not close the line.
A short reminder cadence that stops by itself. As soon as a document is received, no more reminders go out for it. Nothing irritates a client more than being chased for something they just sent.
Hand-over to a person. After two automatic reminders with no answer, the file goes up to a team member, who phones. A two-minute call often settles what ten emails did not.

This structure matches what several vendors and providers in the sector describe: a list adapted to the file, a completeness check, reminders that stop on receipt.

A reminder cadence, day by day

The right cadence depends on your calendar and your clients. Here is a starting point, to adapt, for a year-end deadline.

When	Action	Who acts	Content
Day 0	Initial request by email	Automatic, list approved by the firm	The exact documents, with date and amount, the deadline, the drop-off link
Day 3	First reminder by email	Automatic	Only the documents still missing, shortened list
Day 7	Second reminder by text or message	Automatic, shorter tone	The number of documents left and the deadline
Day 10	Hand-over	A team member	A call, with the list and sending history on screen
At any time	Stop	Automatic	No more reminders for a document received and checked

Two principles matter more than the exact days. First: each reminder shortens the list, it does not repeat the first request. Second: a cap on messages per client, so that a badly set rule cannot harass anyone. Tensoria also recommends a small number of automatic reminders before handing over to a human, and measuring the average time between the first request and a complete file rather than the number of reminders sent. We follow that logic: the right indicator is time to completeness, not message volume.

What the AI does in this loop, and what it does not

In this loop, the AI has three clearly bounded roles.

It spots. It compares bank lines or the expected list with the documents received, and lists what is missing.
It drafts. It writes each reminder from the client’s actual list, with the right name and documents. Your team reviews the first sends.
It reads and files. It identifies the document received, extracts supplier, date and amount, and files it in the client’s folder.

What it does not do: decide that a missing document is acceptable, that a client deserves special treatment, or how to present a delicate situation. Those are firm decisions. The AI prepares, a person approves.

One firm-level practice deserves a mention: setting an amount threshold below which a missing document is no longer requested. Some firms mention thresholds of 20, 30 or 50 euros depending on the case, others up to 100. It is a trade-off specific to each firm, your responsibility and a matter for your advisers: the automation applies the rule you set, it does not choose it.

A worked example, to replace with your own numbers

Take a fictional firm with 120 active files. Assumption: a third of the files need three reminders per period on average, roughly 120 reminders. At 10 minutes per reminder, counting the search for what is missing, sending and follow-up, that is 20 hours. At an internal cost of CHF 80 an hour, the total is CHF 1,600 per period.

If the automatic loop handles the reminders and the team steps in for only one file in five, at 15 minutes per call, human time falls to about 6 hours. The gap is 14 hours, or CHF 1,120 per period, before any other gain. These figures are assumptions: neither the number of reminders, nor the time per reminder, nor the share of files that escalate is a measurement. Record a real period before deciding. If the yearly gap is smaller than the cost of setting it up, do not start the project.

When it is not the right choice

An automated collection loop does not suit several situations.

You have few clients and know them all. A two-minute call costs less than a project. Automation earns its place once chasing has become a job in itself.
Your clients already drop everything into a portal that chases on its own. If the tool exists and works, do not duplicate it.
Your document lists are written down nowhere. If each team member “just knows” what to ask for, write those lists first. An AI cannot guess a rule nobody has stated.
Your clients expect personal contact above all. A relationship built on the direct call deserves protecting: use automation for the list and the tracking only, and keep the message yours.

This article covers work organization only. It says nothing about the obligations of your profession, including which documents must be gathered or kept: those are a matter for your professional bodies and advisers.

How to start in four weeks

Week 1: measure. Record the number of reminders and the time spent, per team member, over a real period.
Week 2: write the lists. One document list per client profile, approved by the firm.
Week 3: test on ten clients. Every reminder is reviewed before sending, and the drop-off channel is single.
Week 4: compare, then extend. Look at the time between the first request and a complete file. Roll out only if the gain is clear.

The studio builds this loop on a fixed fee agreed before day one, plugged into your current tools. You can see the processes the studio automates and the plans and prices on the site. If AI is not the right lever for your firm, we say so up front.