The project I talked us out of
I spent about a year designing a feature, then found out 498 people had used it out of 334,512 sessions and argued we should throw it away.
We did. What I proposed instead shipped a few months later, and people use that.
Nobody questioned it, me least of all
A heating zone is one section of a building with its own radiators. It can be freezing while the rest of the building is fine.
In our product every zone was its own page. Manage a four-zone building and you had four pages with no way to look across them.
Nobody interrogated it because it already looked like a feature.
It came from our head of product, it described a real problem, and it named a real customer. That is usually enough. I had seen the four pages myself.
My role
I ran the research, drew every screen, and made the argument to stop.
Not mine: the decision. That was our head of product’s, and I didn’t have the authority to cancel anything.
So I did the work
I interviewed people internally. It went fine, which in hindsight was the first thing I should have noticed.
Then I found somebody outside the company who managed multi-zone buildings for a living and asked him three questions:
- What is your checklist when you look at a zone?
- What do you want to know first?
- With a hundred zones, how do you pick which one to open?
He was great. I got a desktop overview, mobile navigation between zones, and zone stats we had never thought to show anybody.
It went through the usual public rounds. People liked it. Honestly, I still think the designs were good.
Then I pulled the numbers
Nobody had ever checked the denominator.
I only checked because I’d gotten into the habit, not because I suspected anything. Ninety days of data:
- 334,512 sessions on the platform. 498 touched multi-zone.
- 100 users out of 4,699.
- 39 of those 100 opened it exactly once, ever.
Not once a month. Once. They looked at the thing I’d spent nine months on and never came back.
Then I reread my interview notes, and the answer had been in them the whole time.
People weren’t describing a zone problem. They kept saying “I can’t see everything I care about in one place.”
Nursing homes together. Schools together. Customer success wanting their biggest accounts in one view.
Turns out zones were the smallest version of a much bigger problem. I had spent nine months making the smallest version really nice.
The conversation I didn’t want to have
I had to tell our head of product that something he’d asked for repeatedly wasn’t worth building.
If that came out wrong it lands as me defending my own taste, which is the fastest way to lose an argument you’re right about.
So I took it to Kelly-Ann on support first. I gave her the numbers and asked her to help me rank the problems and sharpen how I said it.
She reordered two of my three points. She was right both times, and I’d rather find that out from her than in the meeting.
Introducing Portfolio Groups
Let people group any buildings they want, name it, and save it. Then give the group a savings view, one sensor list, and bulk edits.
A four-zone building becomes one group with four members. The original request gets handled as a special case instead of its own feature.
I use that as a check now on anything that arrives already shaped like a feature.
What happened
Multi-zone stopped. Portfolio Groups shipped a few months later.
By then I was advising on rollout, including pushing us to message every support person individually instead of trusting a note in the app nobody reads.
The zone stats went into the grouped views, so the design work got redirected more than thrown away.
The most useful thing I did that year was find out none of it mattered.
The part I got backwards
I should have pulled that data in week one. It was one request to our data team and I made it nine months in.
Everything I argued in February was sitting on a server the previous March.
The instinct was right and the order was wrong. The order is the part that costs money.