Month-End Close: Stop Chasing Invoices and Expense Claims by Hand

Why a late close costs money

A late close is rarely caused by the accounting itself. It is caused by waiting. A supplier bill that arrives on day four, a manager who has not approved a claim, or an accrual nobody has estimated can hold up the whole sequence. The APQC benchmarking study of 2,300 organisations, reported in 2018, put the median time from trial balance to consolidated statements at 6.4 calendar days, and the bottom quarter needed 10 or more. That measure starts only once the trial balance is run, so the chasing that happens before it does not appear in the number. It still delays the start.

The more recent picture is not much better. Ledge's 2025 month-end close benchmark, based on a survey of 100 finance professionals, found that half of finance teams take more than five business days to close. Ventana Research's 2023 study found that 59 percent of organisations close within six business days, slightly below the 2019 level. In other words, a close of six to eight business days is common, and a team that spends its first two days waiting for paperwork is not unusual.

The direct cost is staff time. The most cited figure for expense claims comes from the GBTA Foundation and HRS study of 2015. It put the cost of processing one travel and expense report at about $58 and 20 minutes of effort across employee, manager and finance team. The same study found that roughly one report in five contained an error, and each correction added about $52 and 18 minutes. That benchmark is old and covers travel expenses, so treat it as an order of magnitude, not a budget. Applied to 30 monthly claims, it still works out to about $1,740 of labour a month, which is my own arithmetic from the 2015 figure.

Chasing invoices has its own cost, and it is harder to see. Suppose 40 open items each need three follow-ups of about five minutes, including finding the right person and logging the reply. That is 10 hours of a controller's time in every close, before any accounting starts. These numbers are planning assumptions. Measure your own follow-up count for two months and replace them with your figures.

The late cost is decision timing. Leaders who wait for the close to see margins or cash are working from figures that are already one to two weeks old. A missing accrual that is found in day eight can change the month's result after the board pack is drafted.

The manual way and the automated way

The table below compares how most finance teams chase month-end items today with a setup that runs the reminders and tracking on a schedule. The accountant's work is the same in both columns.

Set the close calendar first

Automation needs dates to work from. Write down the close calendar before you buy or build anything. For each month, set the day the period locks, the cut-off for supplier invoices, the cut-off for expense claims, the day accruals are due from department heads, and the day the trial balance is run. Most teams aim for a close of about five business days, but the right target depends on your volume and how much is still manual. Ledge's 2025 figures suggest the top group close in three or fewer business days, so treat that as a long-term goal.

Give each department a named owner for each item type. The reminder system is only as good as the name on the list. If three people all think someone else sends the marketing invoices, the tracker will show them as missing every month.

Build the tracker

The tracker is a single table with one row per expected item. Columns should include the department, the supplier or claimant, the item type (invoice, accrual or claim), the expected amount, the owner, the due date, the status, and a link to the document once it arrives. A shared spreadsheet in Microsoft Excel or Google Sheets is enough for a small team. A larger group can use Microsoft Lists or the task module of its accounting system.

Where possible, feed the tracker from the source. Recurring supplier invoices can be pre-listed from the bills already in the accounting system. If you use QuickBooks Online or Xero, check what each one can export or connect to, and use their own connectors rather than custom code where a connector exists. Build the first version by hand for one close, so the team learns what the columns mean before anything runs on its own.

Schedule the reminders

Microsoft Power Automate can run a scheduled cloud flow at a set time, every day or on a chosen date, without anyone pressing a button. Microsoft's own documentation describes this recurrence trigger as the starting point, followed by an action such as sending an email. A typical flow reads the tracker, finds rows still marked missing, and sends each department head a list of their open items. It can run every morning from the start of the close, then again the day before each cut-off.

Keep the message short. A list of three open items with dates gets an answer. A list of thirty does not. Send the reminder to the person who owns the item, copy the controller only after the cut-off has passed, and make the reply path obvious: attach the invoice, or reply saying it will come by a stated date.

Use WhatsApp where department heads already work on it

Some managers will answer a WhatsApp message in minutes and ignore email for days. In that case, the WhatsApp Business Platform Cloud API can send the same reminder. There are three rules to know. First, Meta's On-Premises API was retired on 23 October 2025, so new setups should use the Cloud API. Second, a business can only send free-form messages inside a 24-hour window that opens when the other person writes first. Outside that window, every message must be a pre-approved template. Third, template messages are reviewed before use, and Meta can reject a template that does not follow its guidelines.

Reminders about an open invoice usually fit the utility category, which covers messages tied to a transaction. Write the template so the body reads like a short operational notice, with placeholders for the name, the item and the due date. Pricing for each message depends on category and country, so check the vendor's current pricing before you commit.

Capture invoices and receipts

Most of the time lost in a close goes to finding a document, not to reading it. Set up one dedicated address that managers and staff forward invoices and receipts to. A flow can file each attachment, read the supplier name, invoice number, date, amount and tax lines using optical character recognition, and then pass the result to the tracker. Large language models can help classify the document, for example to tell a supplier bill from a delivery note. Neither tool is perfect on handwritten receipts, poor scans or unusual layouts, so every extracted amount should be checked by a person before it reaches the ledger.

Give the controller a summary on day one

The last step is the report. A scheduled flow can count open items by department and status each morning, list anything past its cut-off, and send that summary to the controller. The controller reads one page instead of twelve inboxes. The accountant can then decide which open items need an estimated accrual, which can wait for the next month, and which need a call.

What breaks, and how to prevent it

Reminders get ignored. People stop reading a daily list that always contains the same three items. Prevent this by escalating a named item after the cut-off, rather than repeating the same message every day, and by reviewing the escalation list with the department heads once a quarter.

Duplicates appear. The same invoice can arrive by email and be uploaded by a manager. The tracker should match on supplier and invoice number, and anything that matches twice should be flagged for a person to merge.

Extraction errors slip through. A misread amount or a wrong department code can look right in the tracker. Keep the human check on every amount and every coding decision, and sample a few documents each month against the source.

Approval becomes a formality. If the manager's approval is only a reply that says ok, the control is weak. Expense policy checks still need a person, because a claim can be valid in format and still break the policy. Automation does not decide whether a meal or trip was allowed.

Personal data needs care. Receipts can contain card details, names and addresses. Limit who can open the shared inbox, keep the tracker in a restricted folder, and check the privacy terms of every tool before you connect it. The rules on personal data differ by country, so ask your own adviser which ones apply to you.

Automation can also hide a broken process. If a department never sends its invoices on time, a reminder only tells you that more clearly. Fix the process owner first.

When not to automate this

Do not build this if the close is only a few days of work and the team has three or four suppliers and one approver. A shared calendar and a checklist will be cheaper to run and easier to keep alive. Do not automate it while the expense policy is still changing, because the rules in the flow will be wrong by the next close. Do not connect a tool to bank or payment data until the accountant has agreed who can see what. And do not expect the system to produce accruals. Estimating an accrual needs judgement about contracts, services received and what is still owed, and that stays with the accountant.

Automation is also the wrong first step if no one owns the close calendar. The reminders will be sent, the tracker will fill up, and the close will still be late, because nobody is responsible for the cut-off dates.

What it typically takes

This is a matter of weeks, not months, for a single-entity business with a standard chart of accounts. Plan roughly one to two weeks to write the calendar, agree owners and build the tracker by hand. Then run one close in parallel, with the old chasing still in place, and fix what the reminders got wrong. Expect to tune the reminder timing and the message wording for two or three cycles before the team stops asking what the messages mean. A multi-entity group with several ERPs will take longer, because each source system needs its own export or connector. These are planning estimates from the work itself, not measured results, and the actual time depends on how clean your records already are.

Results also depend on your own figures. The free automation audit measures how many items you chase each month, how long each follow-up takes and where the close actually stalls, so you can see whether this change is worth making before you commit time to it.