Systemising Pain: When to Fix a Process with Software, Automation or AI
Systemise Before You Automate: Why Order Matters
By Andrew Lee Ward 5 min read Updated 27 Jun 2026
Systemise before you automate, because automating an unclear process only makes the mess faster and harder to spot. First make the process explicit and repeatable: one named owner, documented steps, clear decision rules, and a definition of done. Then decide whether automating it is worth the build.
Automation is fast. Speed in the wrong direction is just a quicker way to be wrong. Systemise first.
Why should you systemise before you automate?
Systemise before you automate because automating an unclear process bakes in the mess, then makes it faster and harder to spot. Automation is a multiplier: it runs whatever logic you give it, at speed, without judgement. Point it at a process with no clear owner, undocumented steps and unwritten rules, and it reproduces every gap and guess thousands of times before anyone notices. Point it at a process you have already made explicit and repeatable, and it strips out the manual effort while keeping the control.
The order is the whole point. A messy manual process is at least slow enough for a human to catch the exceptions; a messy automated one fails silently and at scale. As Bill Gates put it, "automation applied to an inefficient operation will magnify the inefficiency."
What does it mean to systemise a process?
Systemising a process means making it explicit and repeatable so it no longer lives only in one person's head. It rests on four things, in order:
- One named owner. A single person accountable for the process running and improving — not a committee, not "the team". Without an owner, no one keeps the rest true.
- Documented steps. The task written down as a standard operating procedure: inputs, steps in order, plain language, short enough to actually follow. A standard operating procedure, in the American Society for Quality's definition, is the documented best-known way to perform a task — the baseline everyone works from until someone improves it.
- Clear decision rules. The choices made along the way, and the rule behind each — including how to handle the common exceptions, so the process doesn't break the first time reality deviates.
- A definition of done. An explicit statement of what finished looks like, so two people would agree the work is complete. This is the definition of done, and it is what stops "I thought you'd finished" arguments.
Get those four right and you have a business system: something a competent colleague could run without asking you, and a build team could later automate without inventing the logic themselves.
The practical sequence: owner, steps, rules, then automate
Work the sequence in order; skipping a step is where automation projects go wrong.
- Clarify the owner. Decide who is accountable before anything else. The owner does the rest of the work and lives with the result.
- Document the steps. Watch the task done once, then write each step in the order it happens. Have someone else run it from your notes and fix what they trip over — that is your real SOP, not the idealised version.
- Agree the rules and exceptions. Surface the decisions hidden inside the steps and write the rule for each. List the three or four exceptions that come up most and say what to do about them. This is the stage that exposes the disagreements people didn't know they had.
- Then decide whether to automate. Only now do you ask whether to build. Some systemised processes are worth automating; many are fine left as a checklist a person follows. That decision — whether the pain is frequent and structured enough to justify the build — is its own question, covered in when to automate a process.
Notice that three of the four steps happen before you spend a penny on software. That is deliberate. The cheap clarity comes first; the expensive build comes last, and only if it earns its place.
"Automating a process you haven't systemised doesn't remove the mess. It just hides it behind a button." — The Control Standard
Systemising reduces key-person risk
A documented process survives the loss of the person who used to run it. When a task lives only in one expert's head, that person is a single point of failure: a holiday, an illness or a resignation, and the work stops. This is the bus factor — the number of people you can lose before a process stalls — and an undocumented process has a bus factor of one.
Systemising raises it. Moving the steps and rules out of someone's head and into a standard operating procedure means the process no longer depends on their presence. The expert is freed to do more valuable work, new joiners get productive faster, and you can delegate with confidence because the standard is written down rather than assumed. Documentation is what turns "only Sarah knows how to do this" into "anyone can do this, and Sarah improves it."
Where this fits: the systemising-before-automating band
If you have run the free Pain Automation Score and landed in the 31–55 band — "Systemise Before Automating" — this is the work the score is pointing you at. The pain is real and worth solving, but the process needs clearer ownership, steps and rules before any automation will stick. Build on top of that clarity and the automation holds; build on top of an undocumented mess and you have bought a faster mess.
The standards bodies back the same sequence. ISO 9001, the international standard for quality management, is built on "documented information" — the requirement that key processes be written down, controlled and kept current precisely so quality doesn't depend on individual memory (ISO 9001:2015, clause 7.5). Systemising is not bureaucracy; it is the discipline that makes both quality and automation possible.
Where to go next
Systemising before you automate is the central discipline of turning repeated operational pain into something you can control. For the wider picture of when pain is worth solving with people, process or software, start with the systemising pain hub. When you have a process documented and want the practical playbook for building on it, read business process automation. And when you want to know whether to automate at all, when to automate a process gives you the test.
If you're not sure which of your operational headaches are worth systemising first, the free Pain Automation Score below scores one in a few minutes and tells you whether to clarify, systemise, or build.
Frequently asked
What does it mean to systemise a business?
What's the difference between systemising and automating?
What is a standard operating procedure (SOP)?
How do you document a process?
Can you automate without documenting first?
What is the bus factor, and how does systemising reduce it?
Isn't writing SOPs a waste of time for a small team?
How detailed should a process document be?
Keep reading
When Is a Process Worth Automating? A Practical Decision Framework
A process is worth automating when it is frequent, broadly stable each time, costly in time or errors, supported by data you already hold, and mostly rules-based rather than judgement-heavy. If it is rare, changes case by case, needs a person to decide each time, or the steps are not yet clear, automate later — systemise first.
GuideBusiness Process Automation: A Practical Starting Guide
Business process automation (BPA) is using software to run a repeatable business process — approvals, reminders, data entry, status reporting — with little or no manual effort. Start where the work is frequent, rule-based and structured, the data already exists, and an error is costly. Systemise the process first, then automate it.
GuideThe Ten Types of Innovation: A Lens for Where to Improve
The Ten Types of Innovation is a framework from Doblin (Larry Keeley and colleagues at Deloitte) that sorts innovation into ten types across three groups: configuration, offering and experience. Use it as a lens for where software or automation could add value beyond a faster process — in your structure, service, channel or customer engagement.
TopicSystemising Pain: When to Fix a Process with Software, Automation or AI
Systemising operational pain means turning a repeated, painful task into a clear, owned process before you reach for software. A process is worth fixing with software, automation or AI when the pain is frequent, structured and costly, and the data already exists. Most teams either automate too early or never systemise at all.
TermProcess Mapping
Process mapping is the practice of drawing a process as a step-by-step diagram — showing each task, decision, handoff and information flow from trigger to outcome. It makes invisible work visible, exposes where things stall, and creates a shared picture a team can agree on before they try to improve or automate anything.
GuideWhy Does My Team Keep Making the Same Mistakes?
Your team keeps making the same mistakes because the problem lives in the system, not the people. A mistake that recurs across different people and projects is a signal about how the work is shaped: unclear handoffs, fuzzy decision rights, missing standards and slow feedback. The fix is to redesign one of those, not to ask people to be more careful.