The filter that deleted the answer it was looking for

The filter that deleted the answer it was looking for

An analyst wanted to compare mobile and desktop behavior, so they applied a hard mobile filter to the whole report and started drawing conclusions. The numbers were clean and wrong.

A filter permanently removes data from the view, so you lose the baseline you're comparing against. A segment overlays a subset on top of the full picture. The clue: the filtered report couldn't show the desktop side at all, so 'mobile is worse' had nothing to be worse than.

The change: same question, segments instead of a filter, mobile and desktop side by side over the same dates.

The real finding flipped: mobile conversion was 2.1% versus desktop 2.4%, a gap of 0.3 points, not the catastrophe the filtered view implied.

Filter when you want to permanently clean noise. Segment when you want to compare. Use a filter for a comparison and you throw away the other half of the answer.