Reading results

What to do when image-search engines disagree

Different indexes and ranking methods can produce different results for the same file. Treat disagreement as a reason to compare sources, not as a vote that one engine must win.

Compare equivalent inputs

Confirm that each search used the same file and that you did not accidentally compare a full photograph with a tight crop. Note account tier and any test-mode restrictions as well.

Sort by evidence type

  1. Separate exact copies from similar images.
  2. Group results that point to the same source.
  3. Open unique candidate pages.
  4. Record what each page actually establishes.

One engine might find a creator's portfolio while another finds a retailer's licensed copy. Both can be valid results even though their rankings differ. Conversely, a large group of similar thumbnails may all trace to one irrelevant source.

Write the combined finding

Summarise the strongest source-backed conclusion and the unresolved gaps. Do not average similarity scores from different providers; their scales need not mean the same thing. Keep your original question visible so the comparison does not expand into aimless searching.

Continue with how image search works.

Published by Facehunch with AI assistance. How we use sources and maintain these guides.
KEEP EXPLORING

Related guides

FROM READING TO EXPLORING

See how a photo search comes together.

Try the fictional example, or run a consented API test. Paid searches are disabled.

Open Facehunch