During a page walkthrough this week, I slowed the phone’s connection to see what happened after submission. The button faded slightly when pressed, and little else changed. I waited, pressed it again, then realised on the third attempt that I was no longer sure whether the first touch had registered.

The page eventually moved on. The route technically worked, but I did not want to mark it as passed. Would that hesitation make someone leave, go back or start again elsewhere? I put the phone on the table and repeated the path, this time without watching the button continuously.

The event table could wait

Repeated clicks usually send me straight to deduplication: event names, user identifiers and time intervals are familiar territory. This time I recorded the screen first. Watching it back, the uncomfortable part was the silence before an unmistakable response. I did not trust that the action had been received.

Only then did I open the records. A single touch might produce several events; several touches might produce just one request. A recorded interface action and a completed submission are different things. I paused the video during the wait and lined up the touches, request start, visible feedback and outcome separately.

There was no tidy finding. A slow connection was not the only condition worth testing: touching the edge of the button or missing a notice outside the line of sight deserved attention too. I kept the reproducible clip without filling the remaining gaps into a complete explanation.

For now I have not filed that recording under accidental user input. The finger repeatedly pressing the button was mine.

Before leaving, I tried again on a normal connection. The wait was shorter and I did not keep tapping. Both recordings went into the same folder; neither replaced the other.