Designing clearer communication around uncertain timelines

After all the work of selling a project, a homeowner could go weeks without hearing from us. Behind the scenes, we were organizing labor, sourcing materials, and working with the city. From their side, it could look like nothing was happening.

With Uber or DoorDash, the answer might be 18–22 minutes. A construction project might start in three days, three weeks, or—in unusual cases—three months. We couldn’t reliably calculate that window from the information we had. The contractor was closest to the work.

Estimated Start Week was our way to bridge the gap: ask the contractor for a week, then text it to the homeowner. It also helped us prioritize work, but the immediate need was giving people something to plan around.

We put the estimate inside “Review and accept work order.” To a contractor, it looked like another field to clear. We did a poor job of explaining that this answer would leave the task and become a text to the homeowner.

Contractors entered weeks without giving them enough consideration, and estimates weren’t consistently kept up to date. We were hitting the estimated week only 27% of the time. Homeowners now had a date to expect—and another reason to feel let down when it passed. It became a major source of frustration and escalations.

We tried the smallest intervention first: a checkbox immediately before completing the task. “I understand the estimated start week March 17 - March 24 will be shared with the homeowner.” It explained the consequence without showing the text itself.

Accuracy increased by 8 percentage points, from 27% to 35%. That was a modest improvement, and we were also talking to contractors about these messages. It was hard to separate the effect of the checkbox from the effect of those conversations.

PostHog sessions showed contractors clicking the checkbox fast enough that they may not have read it. Contractor feedback confirmed they still didn’t understand what we were texting homeowners. Accuracy remained low, alongside homeowner escalations. We needed more than disclosure.

I wanted the connection to be crystal clear. I considered a small chat bubble beside the task and putting the text next to the checkbox. Both brought the message closer. The phone was the most delightful of the iterations, and it made the destination concrete: this is what appears on the homeowner’s phone.

The contractor could see the message take shape on a simple phone mockup: the phone fades in, typing dots appear, then the text arrives with their estimated week. The subtext still explained what would happen, but the interface did more of that work. They were looking at a text and choosing whether to send it.

After introducing the preview, estimated start-week accuracy reached 62%, up 35 percentage points from the original 27% baseline. A hit meant construction actually started within the estimated week. Homeowner escalations also decreased. Together with the changes I observed in PostHog, these results suggested the preview was helping contractors reconsider what they were about to communicate.

It still left roughly four projects in ten outside the estimated week. Permits, material availability, crews, and weather can change even a carefully considered estimate. The preview made the communication more visible; it didn’t make construction predictable.

A more considered first estimate was only the beginning. When a permit, delivery, or schedule changed, the homeowner needed an update—not another missed expectation.

As a fast follow-up, we carried the same message-preview approach into a reschedule flow. The contractor selects a reason and a revised estimated start week, and sees the homeowner’s text take shape as they make those choices. They can review what they’re about to communicate before sending it.

The goal was to keep people informed as the plan changed. This follow-up is separate from the accuracy results above.