Images can rank in image search yet do little for the host page because image search and web search are separate result systems with separate judgments. An image earns its place in Google Images on its own signals, what it depicts, its filename and alt text, its surrounding context and quality, while the page that hosts it competes in web search on an entirely different basket of factors. Visibility in one system does not transfer into ranking in the other. The two are decoupled, which is why a photo can be everywhere in image results while its page sits nowhere in the blue links.
The mechanism is that Google evaluates the two surfaces independently. For image search, it asks which image best matches a visual or descriptive query and ranks images against each other. For web search, it asks which page best satisfies an informational or navigational query and ranks pages against each other. An image performing well answers the first question; it does not answer the second. Hosting a high-ranking image does feed a small amount of context to the page, but it is not a lever that lifts the page’s web position. This is observed behavior, the way the systems have been seen to operate, rather than a guaranteed rule, since how Google surfaces images can shift over time.
So the gap is real and expected. The same asset can be a winner in one ranking system and a non-factor in the other, and seeing your image traffic climb tells you nothing reliable about where the page stands in web results. The two scoreboards do not talk to each other in the way people assume.
Treat image search as its own channel and measure it as one. If images bring you visitors through Google Images, optimize for that surface deliberately and value the traffic on its own terms. Do not expect those image rankings to pull the page up the web results, and do not read a strong image-search showing as evidence the page itself is ranking. Work the page’s web ranking through the page’s own signals.