Rising average position with falling traffic usually means you are gaining ground on low-value or zero-click queries while losing it on the high-value ones, so the average climbs even as the clicks that matter walk away. Position up does not mean winning, and this paradox is the proof. An average can improve through a mix shift, where you start ranking well for lots of terms that never earned many clicks, while slipping on the few terms that did the real work, and the net is a prettier number sitting on top of a worse business outcome.
The first read is the mix shift: segment your queries by value and you may find the gains concentrated in cheap, low-intent terms and the losses in the money queries. A second read is SERP-feature and AI-answer capture. You can hold or even improve your blue-link position while a featured snippet, a knowledge panel, a People Also Ask block, or an AI overview sits above you and absorbs the click before the searcher reaches your result, so your ranking looks healthy while your share of clicks shrinks. These SERP layouts and AI answers change fast, so the exact behavior is worth confirming against current results. A third, simpler read is seasonality, where demand itself dipped while your positions held, which a year-over-year check will reveal.
What ties these together is that average position can hide a losing trade. You can be “winning” on paper, your aggregate position better than last month, while the specific queries that drove revenue have quietly slipped or been intercepted on the results page. The average rewards breadth of presence, but traffic rewards winning the clickable, high-intent moments, and those two can move in opposite directions. So a rising position number is not reassurance; it is a prompt to find out which queries actually moved.
The action is to segment by query value and diagnose the trade. In Search Console, separate your high-value, high-click queries from the long tail and look at position and clicks for each group, not the blended average. If your top queries dropped, work on those directly. If they held but lost clicks, inspect the live SERP for snippets, AI answers, and other features eating the click, and adjust your snippet, format, or strategy accordingly. Read the mix, not the average, to see whether you are actually winning.