How to map your practice's processes before you automate with AI

Published Thursday 1 October 2026

Table of contents

• Why business process mapping comes before AI automation
• What business process mapping means for an accounting practice
• Which practice processes to map first
• Run your discovery process with a proven prompt
• A simple process-mapping template for your practice
• What to measure to build your automation business case
• Validate your maps and protect compliance before you automate
• Get the full workflow-mapping framework from Xero's Modern Practice Playbook
• FAQs on business process mapping

Key takeaways

• Documentation comes before automation: 87% of practices using AI every day rate their core processes as well-documented, against just 18% of practices with no plans to adopt AI.
• Map pain points, not everything, and start with high-frequency, rule-based work where the systems exist and the time saved is real.
• James Bergin's discovery prompt turns your rough notes into structured process maps you can validate with the people who do the work.
• Measuring each mapped process builds your automation business case and becomes an advisory service you can offer clients.

Why business process mapping comes before AI automation

Business process mapping is writing down how a task actually gets done, step by step, before you change anything. For a practice weighing up AI, it decides whether automation sticks or whether you pay for tools that never quite fit.

The evidence is clear. Xero's Modern Practice Playbook found that 87% of practices using AI every day rate their core processes as well-documented. Among practices with no plans to adopt AI, that figure falls to just 18%. Documentation is the foundation of every AI rollout that works.

James Bergin, EGM Technology Research and Advocacy at Xero, puts it plainly: ‘AI can't help improve processes your team can't yet describe.’ Rachel Harris, director of striveX, reiterates this point, fixing the problem in the process before reaching for any tool. Software should make a good process faster, never disguise a broken one.

Rob Collings of Flexal frames it the same way: ‘AI can only speed up what you've already understood’, and he won't automate a process he hasn't mapped. That caution is well placed. The Modern Practice Playbook reports that 82% of surveyed UK firms have tried AI, and 64% use the AI built into their accounting software. Most firms are past deciding whether to start; the real question is what to map first.

So can accounting be automated with AI? The rules-based, repeatable parts can, once you've mapped them clearly. The judgement and decision making stays with you and your clients.

What business process mapping means for an accounting practice

Three related terms get used interchangeably, and the difference matters when you decide where to spend your time.

• process mapping: drawing the current-state flow of a task so everyone sees how it really works
• process modelling: designing how a process should work, often as a formal diagram
• process mining: using system event logs to discover how a process actually ran across your tools

Process mining suits workflows that already run inside connected software, where the event logs exist. A hand-drawn map suits the judgement-heavy work that lives in people's heads. A generated diagram is not a verified process. A map is like a route drawn on paper: useful for planning, but it only reflects reality once someone who has driven the road checks it. Whichever method you choose, the output is only as good as the evidence and the review behind it.

Which practice processes to map first

You don't need to document the whole firm at once. Map pain points, not everything, and start where a workflow is slow, breaks often, or carries compliance risk.

A simple effort versus impact judgement works well. Your first target should be low effort and high impact, where the systems already exist, the rules are clear, and the time saved is real. In most practices that points to high-frequency, rule-based work:

• document and record chasing
• client onboarding and information collection
• reminder and deadline workflows
• enquiry routing and triage
• bank reconciliation and transaction categorisation

Compliance can reorder the ranking. A step that saves modest time but produces a cleaner audit trail may outrank a flashier time-saver. That matters more as Making Tax Digital (MTD) for Income Tax adds to record-keeping demands. Automated reminders that log every request sent to a client are a good example. They can matter more than raw minutes saved, because they leave a trail you can stand behind at review. It's also the first thing to check when a client asks how to tidy their own workflows.

Run your discovery process with a proven prompt

Mapping doesn't need a three-week workshop. Sit with the person doing the work, capture what actually happens, and let AI structure your rough notes.

The Modern Practice Playbook shares a discovery prompt from James Bergin to start exactly that. Paste it into the AI tool you already use:

"You are an expert at business process modelling. Help me model out the core processes of my accounting [or bookkeeping] practice. Start with the functions that are common to most accounting practices and ask me questions about the steps that we take in our practice to do things differently. Use Event-Driven Process Chain (EPC) notation in your diagrams."

Event-Driven Process Chain (EPC) notation links events, such as an email arriving, to functions, such as logging the query. Mapping the decisions between them lets the flow read in plain sequence. Once you have draft maps, interrogate them with fill-in-the-blank prompts to find where AI adds the most value:

• These ___ processes are the ones at the core of my business.
• My team spends ___% of their time on ___ activity.
• Our least efficient functions are ___ and ___.
• If I could save ___ minutes on the ___ function, that would improve things dramatically.
• Our highest-volume, lowest-cognitive-load tasks are ___ and ___.

Before you act on any of it, run a short AI readiness checklist for the process:

• The current process is documented and someone who does the work has reviewed it.
• The data feeding the process is reasonably clean and consistent.
• Clients have consented to their data being used where relevant.
• A named person owns the workflow and its exceptions.
• Human oversight sits over every AI-suggested action.

This is also how capability builds. The Modern Practice Playbook reports that 51% of UK firms surveyed rated self-led experimentation during work time as the most effective route to AI skill. Giving your team a prompt and an hour to test it beats waiting for a formal training programme.

This same prompt gives you an advisory asset. Run it alongside clients to help them map their own processes. Then advise them on where AI adoption is worth the effort, and where it isn't yet.

A simple process-mapping template for your practice

Good process documentation for accounting firms is lightweight and readable, not a polished diagram that lives in a shared drive and never gets opened. Document one process at a time using these seven fields.

1. trigger: what starts the task in reality, rather than in policy
2. inputs: what you need before you can begin
3. steps: each action in order, in plain English
4. decision points: where the path forks
5. exceptions: what happens when the normal path fails
6. outputs: where the finished work goes
7. systems: the tools each step touches

The exceptions field earns its keep. Practices rarely break on the happy path; they break when a client ignores the first request, a document arrives in the wrong format, or an approval stalls. A map your team can challenge is worth more than one that only looks tidy.

What to measure to build your automation business case

A map without numbers is a drawing. Attach four measures to each process so you can rank it honestly.

• task time: how long the step takes when someone does it by hand
• frequency: how often it happens in a day, week, or month
• error and rework: how often someone has to fix, chase, or re-enter something
• annoyance factor: how much the team dislikes it, because avoided tasks pile up

These numbers build the business case. The Modern Practice Playbook reports that AI already saves UK firms around 7.1 hours a week, rising to 10.6 hours for top performers, close to a full working day. The prize is what you do with that time: 61% of UK firms plan to redirect it into advisory and higher-value work.

The barriers to getting there are practical. The Modern Practice Playbook finds 31% of firms cite time and 30% cite technical know-how. Both shrink once one process is mapped and the first automation proves itself. Advisory work also carries the highest profit margin of any service in the same data. The hours you free up land where they pay best.

This is also the conversation to have with clients. Once you've mapped and measured your own workflows, you can help clients measure theirs, turning your practice's experience into a genuine advisory service.

Validate your maps and protect compliance before you automate

An AI-drafted map can read fluently and still be wrong. Validation is a human job, and it belongs with the people who actually do the work.

Walk the map with them and let them point at a box and say that's not how it happens. Compliance and data handling deserve the same care before anything is automated:

• Keep audit trails intact so every automated action can be traced and reviewed.
• Confirm client consent before their data passes through any AI tool.
• Set a clear framework for what can, and can't, be shared with third-party tools.
• Keep a person in the loop, because generative AI can produce confident but incorrect output.

Starting with the AI built into your accounting software keeps this simpler. JAX, Xero's AI finance partner, works inside the same audit trail your team already relies on, so its suggestions stay traceable and you keep the final say. Write that framework down once and reuse it, so a new hire or a client inherits the same guardrails rather than each person deciding alone. You can hold clients to the same standard, showing them how to protect audit trails and client data as they adopt AI.

Get the full workflow-mapping framework from Xero's Modern Practice Playbook

Mapping before you automate is the difference between practices getting real value from AI and those paying for tools that never stick. The complete framework sits in one place. To go further, Get James Bergin's full workflow-mapping framework and prompt template in the Modern Practice Playbook. Xero and JAX give you the platform-native, auditable place to start your first automation.

The same approach becomes a service you offer clients: map with them, measure the time at stake, and advise on where AI is worth it. The Xero Partner Programme is free to join if you want the tools and support to run this across your client base. It includes a dedicated representative and practice management features. Join the partner programme to get started.

FAQs on business process mapping

Here are some frequently asked questions accountants and bookkeepers might raise about business process mapping.

What if your team resists mapping their processes?

Resistance is normal – it often signals fear that mapping will lead to job cuts. Be clear: mapping protects jobs by showing where AI actually helps and where it doesn't. Involve the people doing the work from day one, not after. They own their process and their input shapes the map. Respect their expertise and they'll help you get it right.

Can you offer process mapping as an advisory service to clients?

Yes. Once you've mapped your own workflows, you have proof of concept. Offer to map a client's high-pain process, measure the time at stake, and advise on whether automation is worth it. You become the guide on what to automate and what to leave alone – that's valuable advisory work that builds on your accounting knowledge.

How do you keep process maps current as workflows change?

Assign one person to own each map and review it quarterly or when you make a significant change. A map that drifts from reality is worse than no map. Build the review into your calendar like you would a client review. When you onboard someone new, have them walk the map with the process owner to spot where it's outdated – new eyes often catch gaps.

What if the process is broken – should you fix it before mapping?

Map it as it currently runs, even if it's messy. That shows you exactly where it breaks and why. Then decide: fix the process first, or automate the broken version to expose the gaps faster? Usually mapping reveals the fix – the team often sees the problem once they describe it step by step.