Management accounts are only useful when a client understands what they show. The numbers can be correct and the pack can still arrive late, with no explanation of why margin fell or why cash is tighter than expected. Writing management accounts commentary is often the slowest part of month end, and it usually lands on a partner or senior manager who is already busy.
AI can help with the first draft. It cannot replace the accountant’s judgement, and it should never be left to invent reasons for the figures. Used properly, it gives your team a structured starting point that they check, correct and sign off.
This guide explains what good management commentary covers, where AI fits in the process, and how to set up a review workflow that keeps accuracy under control.
What management accounts commentary should cover
Commentary explains the story behind the numbers. For most clients, it should address five areas:
- Revenue movement: How revenue compares with the prior period and budget, and which clients, services or months drove the change.
- Gross margin: Whether margin moved, and whether the cause was pricing, mix, staff cost or subcontractor cost.
- Overheads: Any cost line that moved beyond an agreed threshold, such as rent, software or professional fees.
- Cash position: Bank balance, debtor days, creditor days and any large payments or receipts that changed the position.
- Actions: What the client should do next, such as chasing a debtor, reviewing a supplier contract or revisiting a forecast.
Good commentary is short. A typical pack needs a summary of two or three paragraphs and a short note on each variance that matters. Anything that does not need the client’s attention should be left out.
Where AI fits in the process
AI works best on the repetitive, rules-based part of the task: reading the figures, identifying what has moved and turning it into plain English. Where it can help:
- Comparing the current month with the prior month, the same month last year and budget.
- Listing the variances above an agreed threshold.
- Drafting a sentence or two for each variance, based on the data and any notes your team supplies.
- Producing a first-pass summary in the firm’s house style.
Where it should not be trusted without review:
- Explaining the cause of a variance that the data does not show.
- Stating figures that have not been checked against the ledger or management pack.
- Giving advice on tax, financing or compliance.
The accountant owns the commentary. The AI produces a draft, and the accountant decides what goes to the client.
A step-by-step workflow for drafting commentary
This is a practical process you can adapt to your practice. It works with any accounting software that exports a management pack to a spreadsheet or PDF.
1. Export the pack and the comparatives. Pull the current period’s management accounts, the prior period, and the budget or forecast if the client uses one. Make sure the figures are final before you start. Drafting from a pack that is still changing wastes time.
2. Set a variance threshold. Agree a rule for each client, such as any line that moves by more than 10% and £5,000. Only variances above the threshold go into the commentary. This keeps the output focused and stops the draft from listing every small movement.
3. Add context the numbers cannot show. Before generating the draft, note anything your team knows that is not in the data. Examples include a client who left, a large invoice that slipped into the next month, a new hire, or a one-off repair. Without this, the AI can only describe what happened, not why.
4. Generate the first draft. Give the tool the figures, the threshold, the context notes and the house style. Ask for a summary followed by one short paragraph per material variance, with the amounts stated.
5. Check every figure against the source. Compare each number in the draft with the management pack. Check the direction of every movement (up or down), the percentage and the pound amount. This step is not optional. A wrong sign in a client report damages trust quickly.
6. Edit for the client. Rewrite any sentence that sounds generic, is too technical or does not match what you know about the business. Remove anything that speculates. If you do not know the cause of a variance, say so and ask the client.
7. Sign off and record the review. Keep a note of who reviewed the draft, what was changed and when it was approved. This gives you a clear audit trail.
Controls to keep the commentary accurate
A simple set of controls will protect both the client and the practice:
- Human sign-off on every pack. No commentary goes out without a named reviewer.
- A checklist. Use the same checks each month: totals reconcile, signs are correct, comparatives match the prior pack, and no figure appears that is not in the source.
- Client-specific templates. Each client has its own threshold, sections and tone. A standard template stops the output from drifting between clients.
- Clear notes on drafting. Record which sections were drafted with AI and which were written by the team. This helps with quality review and supports your professional standards.
- Data handling rules. Decide what client information can be shared with the tool and under what terms. Check this with your data protection lead before you start.
What to avoid
Most problems with AI-drafted commentary come from a few common mistakes:
- Invented causes. A draft that says “higher costs due to supplier price increases” when no one has checked is a risk. Only state a cause that someone can confirm.
- Generic wording. Phrases such as “the business performed well this month” tell the client nothing. Replace them with the specific movement and its effect.
- Identical commentary across clients. If every client receives the same paragraph structure and phrasing, the reports look automated and lose value.
- Skipping the source check. Drafts can look polished and still contain errors. The check is the part that protects the firm.
- Overloading the client. A ten-page commentary with every variance explained is harder to use than a short summary with three clear actions.
Who this suits
This approach fits practices with 5 to 50 staff that prepare monthly management accounts for several clients. It is most useful where the same reporting structure repeats each month and where the partner or manager spends too long turning the pack into written commentary.
It is less useful for one-off reports or clients with very complex structures, where the judgement needed outweighs the drafting effort. You can see how we support firms like yours on our AI for accountancy practices page.
How to start
Do not roll this out across the whole client base at once. Start with one client for one month:
- Choose a client with a stable monthly pack and a willing contact.
- Agree the variance threshold and the sections the client expects.
- Run the workflow above and time each step, including the review.
- Compare the draft with what your team would have written. Note what needed the most correction.
- Decide whether to extend the process to more clients, and adjust the template based on what you learned.
A one-month pilot shows whether the process fits your practice before you commit more of your team’s time to it. If you are not sure where to begin, an AI readiness review will show which workflows are worth automating first. Our guide to AI agent development explains how agents differ from simple chatbots.
Frequently asked questions
Can AI write management accounts commentary on its own?
It can produce a draft. The accountant must check every figure, confirm the cause of each variance and approve the final text before it goes to the client.
Which figures should go into the commentary?
Those that move beyond an agreed threshold and matter to the client’s decisions. Keep the rest in the pack.
Does the client need to know AI was used?
Check your firm’s policy and professional obligations. Whatever you decide, the accountant remains responsible for the content.
Where should client data go?
Only into tools your firm has approved, under terms your data protection lead has reviewed.
Want help setting up this workflow for your practice? ElvraAI builds practical, sector-specific AI agents for UK accountancy and law firms, designed around the workflows your team already uses.
