We have a wide variety of keywords for each image to make sure we capture all the ways someone might look for it. We also employ a fuzzy search to capture variations of words and typos. We're erring on the side of returning more, but less relevant, results than not enough. We've been testing this a lot, but we're expecting there will be more fine tuning to do. File a support ticket and give us your specific examples of major misses and we'll keep tweaking the search service in the background.
I'm using search but it's returning a lot of stuff that doesn't seem relevant.
Share on social: