Take a system apart, piece by piece, to find the part where change will matter most.
Big problems are hard to think about all at once. Breakdown makes them manageable. You split the whole, whether it's a process, a product, a service or a project, into its main parts, then split the parts that matter into smaller parts. Instead of one tangled lump, you get a map you can actually read.
The map isn't the goal. As you take each piece apart, you study what it does, what it depends on and where it goes wrong. That close look usually shows that most of the delay, cost or frustration lives in just a few parts, often in the handoffs between them. Those are the places worth your creative effort.
It works on things you can touch, like a machine or a building, and on things you can't, like a hiring process, a course or a monthly report.
Use it when a problem feels too big or tangled to tackle, when you need to find where a process loses time or money, or before an idea session, so the ideas land on the right part.
Split the whole into parts, and the busiest parts into smaller ones, until you can see where the time goes.
Where does this system start, where does it stop, and what's inside?
Which three to six big parts, taken together, make up the whole?
What does this part do, what does it need, and where does it go wrong?
Which parts wait on, feed or get in the way of other parts?
Which part deserves to be broken down further?
Where would a change make the biggest difference for the least effort?
Splitting a whole into parts sounds obvious. Done deliberately, it changes what you notice and where you put your effort.
We can only hold a few things in mind at once. Faced with a whole system, we fall back on a vague overall impression, like “it's slow”. A single part is small enough to examine properly, so you can see what's actually happening.
From a distance, a process just looks sluggish. Up close, you may find that most of the delay sits in one approval or one handoff. You can't fix a problem until you know where it lives.
Delay, cost and complaints are rarely spread evenly. A few parts usually cause most of the trouble. Breakdown helps you find those parts before you spend energy improving the ones that were fine.
It's hard to have a fresh idea about “onboarding”. It's much easier to have one about “the laptop request” or “the first-day schedule”. Specific parts invite specific, practical ideas, which is why idea tools like SCAMPER work better on parts.
Name the system and mark where it starts and stops. “From signed offer letter to first real task” is a boundary you can work with. “Onboarding” on its own isn't, because nobody agrees where it ends.
Divide the whole into three to six first-level parts, by stage, component or role. Check that they add up to the whole with no big gaps. Use sticky notes, because you'll move them around.
Take the parts one at a time: what goes in and out, how long it takes, what it costs and where it breaks. Note facts where you have them and guesses where you don't, and mark which is which.
Draw lines between parts that wait on, feed or affect each other. Many problems live in the handoffs between parts rather than inside any single one.
Pick the one or two parts that look slowest, costliest or most troubled, and break them down again. Stop when the pieces are small enough that someone could act on them next week.
Circle the part where change would help most for the least effort. Turn it into a sharp question, such as “How might we get every account working before day one?”, and take it to an idea tool.
A wall or whiteboard, sticky notes, a marker and whatever data you have on time or cost. With AI, start with the prompt below.
Subject: onboarding for a 12-person customer support team. Goal: new hires handle real tickets by day five. Constraint: IT and HR rules are set by head office.
From the signed offer letter to the first ticket a new hire closes alone. Recruiting and payroll stayed outside the line.
Five parts: equipment, system access, product training, shadowing and first tickets. On the wall, they ran strictly one after another.
Training took two days and got good reviews. Laptops arrived on day one. System access took eleven working days on average.
Shadowing couldn't start until the new hire could log in to the ticket system, so every access delay pushed everything else back.
Access split into seven separate requests, each filed only after the start date, each waiting on a different approver.
One bundled access request, filed the day the offer is signed, with every approver named in advance.
The next two hires had every account working on day one and closed their first tickets on day four. The fix was one form and one agreement with head office, not a new system.
AI is good at the mechanical side of breakdown: proposing a complete set of parts, spotting gaps and keeping a growing tree tidy. It can't see your system. Only you know where the real delays, workarounds and complaints are. Let AI draft the map, then correct it with what you actually see.
Get a first split in seconds, then fix it with what you know.
Break our new-hire onboarding into 4-6 parts by stage. Show it as an indented list.
Test whether your parts really add up to the whole.
Here are my parts. What's missing, what overlaps, and which handoffs did I leave out?
Split only the part that matters, one level deeper.
Break 'system access' into its smallest steps. Mark every wait, approval and handoff.
Have AI rank the parts, then check its reasoning.
Rank these parts by delay caused and effort to fix. Give one line of reasoning for each.
Describe the system you want to understand, pick a mode, then copy the prompt into Claude, ChatGPT, Gemini or Copilot. In coach mode the AI builds the breakdown with you one level at a time and asks what you know before it guesses.
The “Open in” links put the prompt in the web address. If your challenge includes anything sensitive, copy and paste instead.
After the session, click one to copy it, then paste it into the same chat.