Business Process Automation: Where to Start When Everything Feels Urgent

Start with the clock, not the tool

Most businesses reach for automation the wrong way round. Something goes wrong, or someone quits, or a customer complains about a delay, and the instinct is to go and look at Power Automate, Zapier, Make, or n8n and ask what it can do. That question is unanswerable in the abstract, because it starts with the tool instead of the problem.

Everything feels urgent when you have not measured anything. Whichever process caused the most pain this week feels like the priority, regardless of whether it actually costs the business the most time over a month. The fix is not a bigger to-do list or a better tool. It is a week spent finding out, with some precision, where the hours are actually going.

Find the time-stealing processes first

A time audit is unglamorous and it works. Pick a representative week, not one that includes a product launch, a big pitch, or someone's leave. For three to five consecutive working days, track your time in rough blocks: client work, internal admin, meetings, chasing information, correcting other people's mistakes, and re-entering the same data into a second system.

The processes that steal time rarely show up as a single dramatic event. They hide inside categories like "admin" and "communication": the weekly report copied by hand from one spreadsheet into another, the approval that sits in someone's inbox for two days because there is no reminder, the onboarding checklist that lives in someone's head rather than on paper.

Why "just get someone to do it" is harder than it sounds

The obvious next move is to hand the worst offenders to someone else: a new hire, a freelancer, an outsourced team, or an automation partner. This is usually where the plan stalls, and it is worth being honest about why.

Handing a process to another person requires you to explain it clearly enough that they can run it correctly without you standing over them. That is a genuinely hard bar to clear for a process that has never been written down. The first few attempts come back wrong, you correct them, the questions keep coming, and you end up spending more time managing the handover than you would have spent doing the task yourself for another month.

The same bar applies to automation, which is why the two problems are more similar than they look. A workflow tool cannot handle ambiguity any better than a new hire can. If a process cannot be written down as a repeatable sequence of steps and decision points, it cannot be handed to a person or a tool with any confidence, and the conversation required to get there is itself a time-stealing process that rarely gets counted.

What you can do yourself before you call anyone

Three things, none of which require a vendor, a consultant, or a new hire.

Write the process down as it actually runs. Steps, inputs, outputs, and every exception you can think of. Not the version in the staff handbook; the version that happens on a Tuesday afternoon when someone is on leave and a supplier is chasing an invoice.

Mark every decision point as a rule or a judgement call. "If the invoice is under $500, approve automatically" is a rule. "Decide whether this client is worth the discount" is judgement. Rules are what get automated. Judgement calls need a person, at least until the pattern behind the judgement becomes clear enough to write down as a rule too.

Time-stamp a handful of real instances. Note when the process starts, when each handoff happens, and when it finishes. This usually reveals that most of the delay sits in waiting, not doing: waiting for an approval, waiting for someone to check an inbox, waiting for a second person to be available. Waiting is often the cheapest thing to fix.

Check what you already own before buying anything

Businesses regularly evaluate a new automation platform before checking what is already sitting in their existing Microsoft 365 licence, unused. Power Automate is included in most Microsoft 365 business and enterprise plans and connects natively to Outlook, SharePoint, Excel, Teams, and Microsoft Forms, with no new procurement or vendor relationship required.

Before looking anywhere else, scan for three patterns: approvals currently done by email that could run through Microsoft Forms and a Power Automate approval flow; spreadsheets that get manually re-typed from one system into another on a schedule; and sign-offs that currently require someone to be physically present or reachable, which could route through Teams approvals instead.

Tools like Make or n8n earn their place for more complex, cross-platform, or higher-volume flows, once it is clear the native Microsoft stack genuinely cannot do the job. Starting there instead of with what you already pay for is the most common source of wasted automation spend. If the job needs an AI step in the middle, How to Hire Your First Digital Employee compares Copilot Studio, Make and n8n and walks through a safe first build.

Prioritise with an effort/impact matrix

Once you have a shortlist of candidate processes, resist the urge to tackle the biggest one first. Score each process on two dimensions, impact and effort, on a simple 1 to 5 scale, and plot the results.

Score independently before discussing as a group; it removes the bias of whoever is loudest in the room. Consistency in how you score matters more than precision in the scores themselves. Aim to land on three to five quick wins, not one large transformation project, as the starting list.

What to track, and how to record the wins

Before touching a single process, capture a baseline for each candidate: time per instance, how often it happens, and the current error or rework rate. A rough estimate is fine; the point is to have a number to compare against later, not to build a perfect measurement system.

After you automate or delegate, capture the same three numbers again. Keep this in one simple shared log rather than scattered across memory, Slack threads, or someone's personal notebook.

Think about where the data goes, not just the task

Every automated or delegated process now produces something it did not produce before: a timestamp, a volume count, an exception log, a record of who did what and when. Decide upfront where that data lands rather than letting each project quietly create its own spreadsheet.

If the process runs through Microsoft Forms and Power Automate, land the results in a SharePoint list or an Excel table connected to the same tenant, not a one-off export. The second and third automation project should be able to build on the data structure from the first, rather than repeating the same mapping exercise from nothing.

This week's checklist

• Track your own time for three to five working days in rough categories: client work, admin, meetings, chasing, correcting.
• Ask two colleagues to do the same for the same week.
• Write down, in plain steps, the three processes that came up most often as delays or repeated admin.
• Mark each step in those processes as a rule or a judgement call.
• Check Power Automate and your existing Microsoft 365 licence before evaluating any other platform.
• Score each candidate for impact and effort, and pick one or two quick wins to start with.
• Record the baseline time, frequency, and error rate before you change anything.
• Decide where the resulting data will live before you build the first flow.

None of this requires a platform decision, a hire, or a consultant's brief. It requires a week of honest measurement and the discipline to start with the smallest process that clearly wastes the most time, rather than the biggest one that feels most urgent. If the shortlist turns out to be bigger than the team has capacity for, or the rules behind a process are genuinely unclear, that is the point at which ScaleASEAN's fractional CTO and technology advisory work is worth a conversation.

Frequently Asked Questions

How do I find out which business processes are actually worth automating?

Run a time audit before evaluating any tool. Pick a representative week and track time in rough blocks for three to five consecutive working days: client work, internal admin, meetings, chasing information, correcting mistakes, and re-entering data into a second system. Asking two or three colleagues to do the same reveals time-stealing processes that are often invisible to an owner or manager.

What's the difference between a rule and a judgement call in process automation?

A rule is a decision that follows a fixed condition, such as 'if the invoice is under $500, approve automatically', and rules are what get automated. A judgement call, such as deciding whether a client is worth a discount, needs a person, at least until the pattern behind the judgement becomes clear enough to write down as a rule too.

Should I buy new automation software or use what I already have?

Check what is already included in your existing Microsoft 365 licence before buying anything. Power Automate is included in most Microsoft 365 business and enterprise plans and connects natively to Outlook, SharePoint, Excel, Teams, and Microsoft Forms with no new procurement. Tools like Make or n8n earn their place for more complex, cross-platform, or higher-volume flows once it is clear the native Microsoft stack cannot do the job.

How do I prioritise which processes to automate first?

Score each candidate process on impact and effort using a simple 1 to 5 scale. Quick wins (high impact, low effort) should be tackled first, usually three to five processes. Major projects (high impact, high effort) are worth doing but need proper planning. Fill-ins (low impact, low effort) are fine for spare capacity, and thankless tasks (low impact, high effort) should be dropped.

What should I measure before and after automating a process?

Capture a baseline for each candidate before touching it: time per instance, how often it happens, and the current error or rework rate. After automating or delegating, capture the same three numbers again and keep them in one shared log rather than scattered across memory or personal notes.

Why do business process automation projects usually stall?

Process improvement projects most often stall on resistance to change rather than on the technology itself. Research from HFS Research notes that automation initiatives meet resistance specifically when the benefits are not made clear early on, so a concrete before-and-after number from the first quick win is the clearest way to bring people on board.

Related Guides

• AI Agents vs Chatbots: What Your Business Actually Needs: What the difference actually means for a business, and which one fits each function.
• Building a Technology Roadmap Without a Full-Time Tech Leader: Prioritising technology decisions with a risk register, without a full-time tech leader.
• Your AI Readiness Checklist: 10 Things to Sort Before You Deploy: Ten things to sort out, including data quality and PDPA alignment, before deploying AI.