When reverse image search is not working, first work out which of four things is happening: the file is rejected, the upload or search never finishes, the search finishes with no results, or the results are unrelated. Each has a different cause. Rejections are almost always format, size or a bad link; stuck searches are connection, browser or queue problems; empty or unrelated results mean the engine's index or search type does not fit your question.
The table below maps the common symptoms to their likely cause and the fix to try first. The sections after it explain each one in more detail.
Symptom, cause and fix at a glance
| Symptom | Most likely cause | Try this first |
|---|---|---|
| "Unsupported file" or the file is greyed out in the picker | HEIC, AVIF, GIF or a non-image file renamed to .jpg | Export a real JPEG or PNG copy |
| "File too large" | Over the site's size limit, often a full-resolution camera photo or a long screenshot | Export a smaller copy at around 2,000 px on the long side |
| Upload bar stalls or spins forever | Cloud-only file, weak connection, or the browser blocked the request | Download the file locally, check the connection, try one small test image |
| Search accepted but stays "in progress" | Provider queue or rate limit | Wait for the status or timeout; do not resubmit |
| Image URL fails but uploading the file works | Wrong URL, a link that needs sign-in, or one that has expired | Save the image and upload the file |
| CAPTCHA appears again and again | Traffic checks on your network, VPN, blocked cookies or rapid retries | Turn off the VPN, allow cookies for the site, slow down |
| Search completes, "no results" | Image is not in that engine's index, or you used the wrong kind of search | Try a different engine type or a better crop |
| Results are lookalikes or unrelated | Engine matched colours or composition, or the crop is dominated by an interface | Crop to the subject, use a copy-oriented engine |
| Results differ from a friend's search | Different country, account, settings or engine version | Compare like for like and note the conditions |
| Result link opens a dead or changed page | The index is older than the live web | Treat it as unverified and look for another copy |
The file is rejected
Every search service lists the formats it accepts. Google Lens is generous; many others accept only JPEG, PNG and sometimes WebP. Facehunch, for example, accepts JPG, PNG and WebP up to 10 MB and checks the actual file contents, not just the name.
Common culprits:
- HEIC from an iPhone. iPhones save HEIC by default. Export a JPEG copy instead (the iPhone and Android guide lists four ways).
- A renamed file. Changing .heic or .webp to .jpg does not convert anything. The site reads the real format and rejects it.
- A saved web page. "Save Page As" in a browser can produce an .html file even though you saw a photo. Right-click the picture itself and choose Save Image As.
- Animated GIFs, AVIF and RAW files. Support varies widely. Export a still frame or a JPEG.
- A damaged download. If the file will not open in your own photo viewer, it will not upload either. Download it again from the source.
To check what a file really is, open it in Preview (Mac) or Photos (Windows) and export a fresh copy as JPEG or PNG. Always export from the original, not from an earlier converted copy, so you do not stack up compression.
The file is too large
Size limits vary by service, and the upload screen or help page usually states the limit. A full-resolution phone photo can run to several megabytes, and a PNG screenshot of a long page can be much larger.
- Check the actual limit on the upload screen or help page.
- Crop first. Removing borders and interface often halves the size and improves results at the same time.
- Export a copy as JPEG at a moderate size; around 2,000 pixels on the long side keeps plenty of detail for search.
- Look at the result before uploading. Faces, small text and logos should still be clear.
If a tiny file also fails, size was not the problem. Go back to format, connection or the service itself.
The upload or search is stuck
Uploading and searching are two separate steps. A file can upload fine and then sit in a queue while the engine processes it. Face search providers in particular can take a few minutes when busy, and a well-built tool will show progress rather than a fake timer.
Work through these in order:
- Wait for a status or an error. Opening the same search in three tabs does not make it faster. It can create duplicate jobs and trigger rate limits.
- Check the file is local. A photo stored only in iCloud, Google Photos or OneDrive may still be downloading when you pick it. Open it fully first.
- Check the connection. Load an ordinary web page. Switch from mobile data to Wi-Fi or the other way round.
- Try a small test image you do not mind sharing, such as a photo of a mug. If it works, the problem is your original file. If it fails too, it is the browser or the service.
- Try another browser or a private window, which runs without your extensions. Ad blockers and privacy extensions sometimes block upload requests.
- Write down the exact error message before retrying, so you can report it to support.
Do not send the same private photo to a string of unrelated services just to see which one responds. Each upload is another copy of your image on someone else's server. Our guide to photo search privacy explains what to check first.
An image URL fails but uploading works
Search sites that accept links need to fetch the image themselves, from their own servers. Your browser can see things they cannot.
- You pasted the page URL. The address bar points to the page, not the picture. Right-click the image and choose Copy Image Address (or Copy image link).
- The link needs a login. Images in private albums, social media DMs and many cloud drives only load for signed-in users.
- The link has expired. Some platforms generate image links that work for minutes or hours.
- The site blocks hotlinking or automated fetches.
The fix is the same in all cases: save an authorised copy through the site's normal controls and upload the file. Do not make a private album public or strip access controls just to satisfy a search tool. And avoid pasting links with long access tokens into public notes or unfamiliar tools, because those tokens can grant access to more than one picture. Facehunch only accepts uploaded files, so this problem does not arise there.
The CAPTCHA keeps coming back
Search engines show CAPTCHAs when traffic from your connection looks automated. Google explains the "unusual traffic" page on its help centre.
Typical triggers and fixes:
- VPN, proxy or Tor. Many people share the same exit address. Turn it off for the search or switch server.
- Shared networks. Offices, universities, hotels and mobile carriers put many users behind one address.
- Blocked cookies. If the site cannot store the cookie that records a solved challenge, it asks again. Allow cookies for that site.
- Rapid repeats. Many uploads or refreshes in a short time look like a bot. Close duplicate tabs and wait a few minutes.
- Extensions. Try a private window or another browser.
Do not install "CAPTCHA solver" extensions or disable your browser's security settings. If a site remains locked, use a different legitimate engine. A CAPTCHA has nothing to do with your photo; it does not mean the image was flagged.
The search finishes with no results
"No results" means that engine, with that input, returned nothing. It does not mean the image is original, that the person is real, or that the photo has never been online.
Reasons it happens:
- The image is not in that index. Every engine crawls a different slice of the web. Private accounts, most of Instagram and Facebook, dating apps and messaging apps are largely not crawlable.
- Wrong kind of search. A general engine looks for copies or similar pictures. If you are looking for the same person in different photos, that is a face search. The comparison of search types explains the difference.
- Too much has changed. Heavy crops, filters, mirroring, text overlays or compression can push a copy below the match threshold.
- The source is new or was removed. Indexes lag the live web in both directions.
What to try: one cleaner crop, one wider crop, then a different engine type. The screenshot and crop preparation guide covers flipped, compressed and text-covered images. Then stop. Ten near-identical retries add no information.
The results look unrelated
A visual engine compares patterns. Two portraits with the same lighting and pose, or two beach photos with the same colours, can score as "similar" without any shared source.
- Crop to the subject. If half the screenshot is a chat app, you will get chat app screenshots.
- Use a copy-oriented engine such as TinEye or Bing's "Pages with this image" when you want exact reposts.
- Check results on the source page. Compare the background, clothing, crop and small details. A match is only useful if the page itself supports it.
- Read similarity scores carefully. In face search, a high score means the faces look alike to the algorithm, not that the identity is confirmed. See how accurate face search is.
Results differ by country, account or device
Two people searching the same image can see different results, and that does not mean one of them did it wrong.
- Google Lens features and result types vary by country and language.
- Signed-in searches, SafeSearch settings and search history can change ranking.
- Some engines are unavailable or limited in certain regions.
- Free or preview tiers of paid tools often hide sources that a full search shows.
- Mobile and desktop versions of the same engine can show different tabs.
If you are comparing tools, use the identical file, note the date, engine, account type and visible settings, and compare those conditions before drawing conclusions. Do not use VPNs or fake locations to get around a service's regional restrictions.
The result page is dead or has changed
Search indexes keep thumbnails after pages change or disappear. If a result opens to an error, a different image, or a login wall, record it as unverified rather than treating the cached preview as evidence. Try the page in a web archive or look for another live copy before relying on it.
If you are collecting evidence of a fake profile or stolen photos, take a dated screenshot of the live page when you find it, with the URL visible.
When to stop and change approach
Reverse image search answers one narrow question: where else does this picture, or something like it, appear? If you have tried two crops and two engine types and still have nothing, the answer for now is "not found in these indexes". That is a result.
If your real question is whether someone you are talking to is genuine, a face search can look for other photos of the same face rather than the same file. Facehunch does this for $4.99 per search, with no subscription, and shows each result's source link and similarity score so you can check it. It is meant for people you are already in contact with or for finding your own photos, never for identifying strangers. Check the help centre if something in the app is not behaving as described.
Written by Facehunch with AI assistance and checked against the cited sources. Our editorial policy.