"Tiling." That is the whole description on item 214. Classification: tiling. Trade: tiler. Description: tiling. The photo is of a bathroom, taken from the doorway, and somewhere in it is the thing that is wrong.
The tiler rang on Thursday from the car park. What tiling? Where? He had the photo open on his phone and could see a bathroom. He had brought grout, silicone and a box of spare tiles, which is the tiling equivalent of bringing everything, and he still had to ring.
The reader is not you. It is a trade in a van, eleven days from now.
Here is the opinion. Nearly every bad defect description was written for the wrong reader. It was written for the person writing it, who was standing in the room, could see the fault, knew what they meant, and typed the one word that would remind them. "Tiling." "Paint." "Door." It reminds them perfectly. It tells the trade nothing.
The reader who matters is the one who was not there. The tiler, in the van, eleven days later, deciding what to bring in from the back. If the description makes him ring you, it has failed, and the item has just acquired a phone call, a wait and probably a second visit. The other reader, less often but at much greater cost, is whoever is deciding a dispute about the item a year on, who also was not there and who will read "tiling" and draw their own conclusions about how carefully the register was kept.
Both of them want the same three things.
What, where on the thing, how much
"Grout missing, shower niche, left return, 300 mm run."
Nine words. What is wrong: grout missing. Where on the thing: the shower niche, the left return. How much: 300 mm. The tiler reads it in the van and knows he needs grout in the colour on the job, a float, twenty minutes, and which way to turn when he walks into the ensuite. Nobody rings anybody.
Notice what is not in the nine words. Not the room, because the record already knows it is 3.07's ensuite. Not the trade, because that is a field. Not the priority, same. The description does not have to carry anything the record carries elsewhere, and a description that repeats the room and the trade and the level is wasting the reader's attention on things they already have.
The extent is the part people leave out and the part that matters most later. "Grout missing" could be a 50 mm touch-up or a strip-out of the whole niche. The 300 mm decides which, it decides what the tiler brings, and when the item is disputed it is the first thing argued about. A measurement or a count: 300 mm, 9 of 40, approximately 4 m². If you do not have a tape, pace it, and say you paced it.
Symptom first, diagnosis second, if at all
"Membrane failure, ensuite ceiling." Confident, specific, and quite possibly wrong.
What you saw was a water mark on the ceiling. Whether it is the membrane, the shower screen, the waste, or condensation from a bathroom with no exhaust is the trade's job to find out, and if you have written "membrane failure" on the record and it turns out to be the exhaust, the record has said something false and the waterproofer has been issued a plumber's problem. Write the symptom: "water mark, ensuite ceiling, approx. 400 mm from the wall above the shower." If you have a view, add it after, flagged as a view: "suspect membrane; check niche above." The symptom is a fact. The cause is a hypothesis, and the record should know the difference.
The same rule kills the adjectives. "Poor." "Unacceptable." "Shoddy." They tell the trade how you feel and tell the record nothing, and they read very badly indeed when somebody reads them aloud in a tribunal. The measurement does the work the adjective was trying to do, without the tone.
Mark the photo instead of describing it
Half of a bad description is a sentence trying to do a photo's job. "To the left of the second tile from the top, about a third of the way along." A pin on the photo at the fault does that in a tap, and does it unambiguously. Circle the stain. Put an arrow on the lifted edge.
Then use the words for what the photo cannot show. The sound the door makes. The smell under the sink. What happened when you ran the tap for ten minutes. A photo of a dry floor with the words "ponds 5 mm after water test, drains in 4 min" is evidence. A photo of a dry floor with the word "drainage" is a photo of a dry floor.
Where the tool helps, and where it cannot
IssuesId is built so the description only has to describe. The room comes from the location tree or the QR label on the door frame. Trade, priority, type and classification are fields. The photo is taken in the app and pinned where the fault is. There is a mic on the description field, and what you say is added to what you typed rather than replacing it, so "grout missing, shower niche" typed and "left return, about 300 mil" said into the phone come out as one description. The search box matches the description along with every other field, which means a word used consistently is a word you can find across four hundred items.
Since October the form also has a go at the words for you. Start with the photo and within a few seconds it has drafted one short line, what is wrong and where, and suggested the classification and the priority. Each one is marked Suggested with a reason underneath. Change anything and it is yours, and nothing is saved until you press Create.
I will say the uncomfortable part, since we built it. The draft is very good at the first two of the three things, and that is exactly what makes it dangerous. A line that reads "grout missing along the left return of the shower niche" looks finished. It is two thirds of a description. It can only see what is in the frame, so it knows nothing about the 300 mm unless the photo shows the whole run, nothing about the tap you ran for ten minutes, and nothing about the noise the door makes. Point the camera at a bathroom from the doorway, the way item 214 was taken, and you get a sentence about a bathroom.
So the habit has not changed, it has moved. You used to write the nine words. Now you read the draft as the person in the van and add what only you know: the extent, the test, the thing you heard. Typing into the box refines the other suggestions as you go. If you would have to ring yourself to understand it, it is not finished.
And the email the trade gets is built in the order the trade reads: site and room first, then the description, then the photos, then who to call. The nine words land exactly where the tiler is looking for them.
Item 214, run again
Photo from inside the shower, close enough to see the run. Before the phone is back in a pocket the form has drafted "Grout missing along left return of shower niche", suggested tiling and a standard priority. Pin on the niche. "300 mm run" said into the mic, added to the end. Trade tiler, due in ten days to suit the unit's handover. Well under a minute, standing in the ensuite, and the only words anyone had to supply were the ones the photo could not.
Thursday the tiler opens the email in the car park, reads it, takes grout and a float out of the van, and goes up to level 3. Twenty minutes later it is marked complete with a photo of the niche. Nobody rang anybody.
The short version is at how do you write a defect description a trade can act on, how the suggestions work is in Defects, and the same rule is the Description column on the defect register template, where it says: write it for a stranger.
