How to Correlate Accounts Across Platforms Using Usernames and Images
It starts with a name. Or maybe just a handle - like “mooncat85.” No real name, no photo. Just that. You type it into a search box and hit enter. And then it happens: a Reddit profile, an old blog comment, maybe an Etsy shop or a long-forgotten Flickr account.
Suddenly, you're not dealing with one post anymore. You’re dealing with a network.
Correlating social media accounts across platforms is one of the oldest tricks in open-source intelligence (OSINT) - but it’s still one of the most effective. Whether you’re investigating a public figure, documenting disinformation, or trying to understand digital behavior, account correlation gives you a map of the person, not just a dot.
And it doesn’t take spy tools or expensive software. Just a sharp eye, a few free resources, and the patience to look sideways when everyone else is looking straight.
Here’s how I do it.
Start with the Username, And Variations Matter
Most people aren’t creative when choosing usernames. They reuse them across platforms, maybe adding a digit here or a punctuation mark there. That’s the first link in the chain.
Let’s say someone uses “urbantracker” on Twitter. Try variations:
urban_tracker
urban.tracker
urbantracker123
urbantrackerphoto (common for image-based platforms)
To automate this search, use the excellent open-source platform WhatsMyName. It scans hundreds of social platforms - large and obscure - and shows you where a given username appears.
Be mindful of false positives. Just because the name matches doesn’t mean the person does. But if the profile pic, bio style, or content feels familiar, you’re probably on the right trail.
Use Reverse Image Search to Catch Cross-Posting
Many users share the same profile picture across platforms - or post the same selfie, meme, or event photo in multiple places. That’s your next breadcrumb.
You can drop any image into Google Reverse Image Search or TinEye and see where else it appears. These tools don’t just show duplicates - they help you spot:
Earlier posts (who really posted it first?)
Alternative accounts using the same photo
Reposts on forums, marketplaces, or blogs
For deeper searches, especially for faces, PimEyes offers facial recognition against billions of indexed web images. Use this with extreme ethical care - it’s powerful and can surface private or unexpected results. Always verify and contextualize.
If you’re working with an old profile photo from a defunct site, and want to extract its surrounding text or metadata from an archived page, Smartial’s extractor for archived content is a good place to start. You can pull the surrounding page text and check for usernames, links, or contact information that may not be visible at a glance.
Read the Writing Style - And the Metadata
Correlating accounts isn’t just about photos and names. People reveal themselves in how they write.
Compare tone, formatting, spelling quirks, emoji usage, hashtags, or punctuation. A person who writes “okie dokie” on Discord is likely to say it elsewhere too. Someone who ends every sentence with ellipses on Reddit probably does the same on Mastodon.
Look at timestamps. If two different accounts post about the same event at the same time, using the same phrasing, you’ve got a strong signal. Check post frequency, profile creation dates, even timezone patterns.
And if you’ve archived content from one of the platforms, metadata can help bridge the gap. Extracting this consistently - and responsibly - isn’t always easy, but we covered the essentials in our detailed post on understanding social media metadata. It’s worth reviewing if you're trying to piece together a broader identity from small clues.
Look for Linked Accounts (and Forgotten Footprints)
Sometimes people leave the door wide open.
They link their Twitch to their Twitter. Their Etsy shop lists their Instagram. Their blog mentions their LinkedIn. In some cases, you’ll find an old YouTube channel that still has an “About Me” with a dead Tumblr URL - and that Tumblr links back to their current TikTok.
The web doesn’t erase things evenly. You just need to follow the echoes.
Don’t overlook common aggregator sites, like:
These are often used to collect links, and many people forget to update or hide them even after they pivot to new platforms or usernames.
You can even plug domains or usernames into public search engines like Google Dorks or site-specific queries (site:medium.com username) to surface accounts that don’t show up via regular searches.
Use Archived Snapshots to Spot Identity Shifts
One of the more advanced tactics is correlating accounts over time. Someone may have changed their username, deleted posts, or shifted to a new platform - but the old material is often still archived.
This is especially useful for documenting:
Account rebrands
Shifts in political or professional identity
Deleted bios, handles, or photos
Past affiliations or collaborations
To correlate across time, check older versions of their profiles using archive.org. If you’re unsure how to build these checks into your workflow, Smartial’s guide on integrating archive.org data into OSINT workflows walks you through it step by step.
Snapshots don’t just freeze what someone said - they preserve how they linked accounts, which friends they tagged, and what services they promoted. That’s gold for correlation.
Correlation Is Not Proof, But It’s a Map
Remember: you’re building a profile, not pressing charges. Just because two accounts look similar doesn’t mean they’re the same. Correlation gives you a direction, not a verdict.
Treat each match as a hypothesis. Stack evidence. Be transparent about uncertainty.
The goal is not to unmask - it’s to understand. Especially if you’re documenting social media behavior for research, history, or security. Respect the lines. Don't reach further than the evidence supports.
When in doubt, slow down.
Digital identities are like shadows. You can’t always follow them directly. But you can read their shape, their movement, and where they point. That’s the art of correlation - and the craft of paying attention.