The following account is invented, assembled to show a method rather than to report a real event. A small maintenance planning team at a regional aircraft parts supplier kept receiving the same complaint: the wrong component was shipped, the customer sent it back, and a replacement went out days late. Over one quarter the returns pile grew, and each return was logged as “picking error.” This case study follows the team as they asked why five times and ended up somewhere unexpected.
The evidence they started with
The team had a return log with a reason column, a shipping spreadsheet and a month of order emails. Most reasons read “wrong item sent.” An analyst on the team grouped the returns by where the mistake first appeared, rather than by who handled the parcel. That small change of grouping turned out to matter more than any other step, because it moved attention from the warehouse floor to the paper trail upstream.
Why one: why was the wrong item shipped?
The picker took the item the pick sheet named. So the sheet itself carried a part number that did not match what the customer needed. The wrong-item story was really a wrong-instruction story, which cleared the warehouse floor of most of the blame.
Why two: why did the sheet carry the wrong number?
The sheet was generated from the order, and the order had been keyed by hand from the customer’s request email. In several cases two digits were swapped, and in others a similar-looking part with a different suffix had been chosen. Hand keying was the second layer down.
Why three: why was keying done by hand?
Customers wrote requests in free-form emails, and the order system had no import path for them. A coordinator read each email, searched the catalogue, and typed the result. During busy weeks one person handled dozens of requests in a sitting, and attention drifted.
Why four: why was nothing catching the slip afterwards?
There was a review step on paper, but it only checked that the quantity and the address matched the email. It did not compare the part number with the aircraft or component the customer had mentioned. Reviewers were checking what was easy to check, which was not what mattered most.
Why five: why did the review check only the easy things?
Because the review checklist had been written years earlier, when the catalogue was small and part families were few. Nobody had revisited it after the range grew. The checklist, not the people, had fallen behind the business.
That was the root cause, and the team could finally say it in one sentence: the order review step was designed for a smaller catalogue and was never updated.
What the team changed
- The checklist gained a line to compare each part number against the application the customer described.
- Requests were captured in a short structured form instead of free text, so the part number and the application travelled together.
- Look-alike part families were flagged in the catalogue with a note on what separates them.
- The returns log got a second column, “where did the error first enter?”, replacing the generic reason.
None of these changes asked anyone to be more careful. They changed the system so that carefulness mattered less.
A closer look at why checklists drift
It is worth pausing on the fifth answer, because it is the one most teams skip. A checklist feels like a settled object. Once it exists, people follow it without reading it, and the habit of obeying replaces the habit of asking what it is for. In this case the line about quantities and addresses had been added after a past incident, so it felt important, and the part-number comparison was assumed to be covered by the picker. Each step had a reason when it was written. The problem was that the reasons were never written down next to the steps, so nobody could tell which ones still applied after the catalogue grew. Adding a short “why this step exists” note beside each line is a modest way to keep that knowledge alive.
What the numbers did
The team did not publish figures, and neither will this account, because an invented number would add nothing. What they reported internally was directional: returns attributed to wrong items fell over the next quarter, and the share of returns tracing back to look-alike families rose briefly, which told them the flagging was working because the hidden cases were now visible. An analytical reading of that pattern is useful. When a fix surfaces a category that used to hide inside a vaguer label, the new category growing is evidence of better measurement, not worse performance.
Lessons for other parts and maintenance teams
Grouping failures by where they first entered, rather than where they were discovered, is the most portable idea here. Discovery points are usually the end of a chain: the warehouse, the customer, the inspection bay. Entry points are further back. The team that groups by entry point tends to find process problems; the team that groups by discovery point tends to find people to blame.
A second lesson concerns checklists. They age. Every review step has an author, a date and an assumption about the business at that time, so an annual read-through of each checklist is cheap protection. For a broader view of tracking where resources actually go, the supplier’s own site has a post on following monthly spending to its real destinations, a close cousin of tracing errors to their entry point.
Looking beyond your own sector for structure
When the team redesigned its request form, one member browsed other industries for inspiration on how to organise large lookups. A medical-tourism style directory at directory website for orthopaedic hospitals and doctors served as one example of a big catalogue split into labelled areas such as hospital, doctor, procedure, blog and forum pages. That prompted the idea of grouping parts by application first. The comparison went no further than that, since the businesses are unrelated, but borrowing a navigation idea from a different field cost nothing.
Questions to ask in your own shop
- Where did the last three errors first enter, as opposed to where they were caught?
- Which review step checks only what is easy to check?
- When was each checklist last read end to end by someone who did not write it?
- Is any manual re-typing still happening between two systems?
- Does the returns or defects log record the entry point?
If two or more answers make you uncomfortable, you have found the place to begin. Pick one, ask why five times, and write down the point where the answer turns from a person into a process. For more everyday travel-related reading from the same site, there is also a piece on starting a journey with a rental plan.

