Post-Occupancy Evaluation
Find out how the building is actually used, and tell someone
Once installed, Claude loads it on its own when your conversation matches. You can also call it directly with /post-occupancy-evaluation.
Install just this one
npx archtmpl@latest --skill post-occupancy-evaluation --globalFirst time? The whole install, step by step
- Open Claude Code — the terminal version or the desktop app, either one.
- In a terminal, paste the line above and press Enter. In the app, paste it into the chat and ask Claude to run it.
- Restart Claude Code. That's the whole install.
Set up plugins for me: run `claude plugin marketplace add https://archaiflow.com/plugins/marketplace.json` and then `claude plugin install post-occupancy-evaluation@archaiflow`Paste into the Code tab (not Chat or Cowork) and approve when Claude asks. The third-party marketplace it mentions is this site. Windows may ask to install Git once.
What this skill does
Post-Occupancy Evaluation
Almost nobody goes back. When they do, they walk round with the client, hear that everyone is very happy, and leave. The visit that is worth making asks different questions, looks at what people have changed rather than what they say, and ends with findings sent to someone who can use them.
Workflow
Step 1. Establish what was intended, before going
Ask, in one message:
- What did the project set out to achieve? The brief, the narrative, any stated target. Point at
design-narrativeandbrief-review.- How long has it been occupied, and through which seasons? A building visited in one season has been half seen.
- Who is there: occupants, staff, managers, the people who maintain it?
- Has anything already been raised: complaints, defects, changes made?
- Was it commissioned properly, and was there any seasonal commissioning?
- Is anyone measuring anything: energy, temperature, occupancy?
- Who is paying for this visit, and what do they want from it? A client, the practice's own learning, a certification requirement, or a dispute.
Question 7 matters. A visit made for the practice's learning and one made because a client is unhappy are different visits and should not be confused, and the second one is not this skill.
Step 2. Look before asking
The order is the method. What people have done is more reliable than what they say.
Walk the building and record, without a question asked:
- What has been changed. Furniture moved, a door propped, a room used for something else, a screen put up, a sign added, a lock fitted, a heater brought in, a blind taped down, a controller covered with a note.
- What is worn. The route people actually take shows on the floor. It is often not the one drawn.
- What is unused. A space that is empty at the time it should be busy.
- What is overfull. Storage that overflowed, a room with more people than it was for.
- What is broken and has not been fixed, which says something about how the building is maintained.
Every handwritten sign in a building is a design failure somebody papered over. Photograph all of them; that alone is worth the visit.
Step 3. Ask the questions that do not produce politeness
Asking whether people like the building produces yes. Ask instead:
- What do you do differently here from your last building?
- What do you have to explain to a new person on their first day?
- What have you moved, and why?
- What do you do when it is too hot, too cold, too bright, too loud?
- What has to be booked, queued for, or waited on?
- Where do you go when you want to be away from people?
- What breaks?
- What do visitors get wrong?
Ask the people who maintain and clean it separately, and give that conversation
as much time. They know more about the building than anyone, they are almost
never asked, and cleanability is where finishes fail. Point at
design-interior.
Step 4. Keep three columns apart
| Finding | Observed / Reported / Measured | By whom | When |
|---|
- Observed is what was seen. Strongest.
- Reported is what somebody said. Real, and it is their experience, not a fact about the building.
- Measured is an instrument reading, with whose instrument and when.
Never merge them. Most POE reports are entirely reported and read as though
they were measured, which is how a finding gets dismissed by the one person
who needed to act on it.
Step 5. Compare against what was intended, and be honest
Against Step 1, for each stated intention:
| Intention | What is happening | Gap | Whose |
|---|
Whose sorts each into: the design, the construction, the commissioning, the
operation, or the brief being wrong in the first place. All five are real
findings and the last one is the most useful and the least reported.
Include what worked. A POE that lists only failures teaches nothing about what to repeat, and the practice will quietly drop the thing that was working.
Step 6. Send it to someone, or it was not worth doing
Each finding has a destination, and this is the step that is usually missing:
| Finding | Goes to | Why them | Sent? |
|---|
- The practice, for the next project:
lessons-learned. - The client's facilities team, where it is operational.
- The commissioning agent, where it is a system never set up properly.
- The contractor or the warranty, where it is a defect.
- The design team's own standards, where it is something to stop doing.
- The occupants, where the answer is that something can be adjusted and nobody told them.
Report the count of findings with no destination. Those are the ones that disappear.
Step 7. Save
Ask where. Then, separately, put the practice-facing findings into
lessons-learned, because that is the file read at the start of the next
project and this one will not be.
Rules
- Look before asking.
- Photograph every handwritten sign.
- Interview the maintenance and cleaning staff separately, and at length.
- Keep observed, reported and measured in three columns and never merge them.
- Sort every gap into design, construction, commissioning, operation, or brief.
- Record what worked, not only what failed.
- Every finding names a destination, and the count with none is reported.
- Never supply a benchmark or a satisfaction scale.
Anti-patterns
- Walking round with the client and asking whether they are happy.
- A report entirely of reported opinion, presented as findings.
- Visiting in one season.
- Skipping the cleaners.
- Listing failures only.
- Blaming the occupants for using the building differently from the drawing.
- Confusing a commissioning failure with a design failure.
- Writing a report nobody receives.
Resources
None. This skill is one file. Output is written directly at the path you choose.
What it does not check
What this does. Plans and records a return visit: what to look at, what to ask and what not to, how to separate what was observed from what was said and what was measured, and where each finding has to go afterwards.
What this does not do.
- It carries no benchmarks and no scales. What good looks like comes from what the project set out to achieve. A satisfaction score against an industry average describes a different building.
- It measures nothing. Energy use, temperature, air quality, acoustics and light are measured by whoever has the instruments, and this records what they found.
- It is not a defects inspection. Point at
punch-listand the warranty process; a defect found on a POE visit gets recorded and routed, not investigated here. - It is not commissioning. Point at
senior-commissioning-consultant, and note that a building that was never properly commissioned will produce POE findings that are really commissioning failures. - It does not record what the practice learned.
lessons-learneddoes, and it is the first place these findings go. - It does not replace the Architect of Record.
What you need before starting. The building, and permission to visit. What the project set out to achieve, from the brief. How long it has been occupied. Who is there. Whether anyone has complained about anything.
Files it puts on your disk
.claude/skills/post-occupancy-evaluation/1 file · 8.3 KBSKILL.md8.3 KB