CTR is impressions ÷ clicks for a given query, position, or result. In ecommerce search, CTR by position is the canonical health curve: position 1 should sit around 25–35%, position 2 around 12–18%, with a steady decay through page 1. Anomalies — a flat or inverted curve — usually mean ranking is mis-ordering relevant results.
CTR alone is misleading because position itself drives clicks (position bias). To compare ranking algorithms fairly, normalize by expected position CTR or use position-bias-corrected metrics like inverse propensity weighting. A new ranker that lifts CTR at positions 6–10 while holding position 1 flat is doing real work, even if the raw number barely moves.
Pair CTR with downstream metrics — add-to-cart rate, conversion rate, revenue per search — because high CTR with no purchases means clickbait results, not relevant ones.