Browser and Device Fingerprinting Explained: How Accounts Get Linked

Your accounts can be linked without a single cookie. A fingerprint is the sum of dozens of small facts about your browser or phone, and together they can point to one device. Here is what's in it, how unique it really is, and what it means for agencies.

Daniel Reyes11 min readЧитать на русском
Part of the guide: Multi-Account Management: How to Run Many Social Accounts Safely
Browser and Device Fingerprinting Explained: How Accounts Get Linked

Browser and device fingerprinting is how a platform recognizes a device from the details it reveals, with no cookie needed: the browser and its version, the operating system, screen, fonts, graphics hardware, timezone, language, the way it opens encrypted connections and the network it comes from. Each detail is shared by millions of people. Together they narrow a device down to a small group or a single machine. Platforms combine that fingerprint with IP addresses, account details and behavior to decide which accounts belong together, which is how accounts get linked even when they never share a login.

What is browser fingerprinting?

Browser fingerprinting is identifying a browser, and so usually a device and a person, from the characteristics it shows to websites. The W3C's fingerprinting guidance defines it as "the capability of a site to identify or re-identify a visiting user" through "configuration settings or other observable characteristics."

The W3C splits it into two kinds:

  • Passive fingerprinting uses what arrives with every request anyway: HTTP headers such as the User-Agent string, the IP address and other network-level information. The device doesn't run anything special, so it can't notice.
  • Active fingerprinting is when a site "runs JavaScript or other code on the local client" to learn more: window size, installed fonts, connected devices, performance, sensors and how it renders graphics. It's "potentially detectable on the client," but only if you go looking.

Unlike a cookie, a fingerprint isn't stored on your device, so deleting things doesn't reset it. This article is part of our guide to running many social media accounts safely, where fingerprints are one layer among network, security and behavior.

What does a fingerprint contain?

A fingerprint contains every detail about the browser or device that varies between people and stays stable for the same person. The most commonly measured ones:

SignalPassive or activeWhat it revealsHow much it varies
User-Agent and client hintsPassiveBrowser, version, OS, sometimes device modelModerate
IP address and network typePassiveLocation, ISP, home vs mobile vs datacenterHigh, but changes
TLS handshake (JA3/JA4)PassiveWhich client software opened the connectionLow among common browsers and apps
Language and timezoneActive (some in headers)Region and settingsModerate
Screen size and pixel ratioActiveDisplay hardwareModerate
Installed fontsActiveOS and installed softwareHigh on desktops
Canvas and WebGL renderingActiveGraphics hardware and driversHigh
CPU cores, memory classActiveHardware tierLow to moderate
Automation flagsActiveWhether the browser is controlled by softwareBinary, but decisive

Two of these deserve a word. Canvas and WebGL fingerprinting asks the browser to draw an image or a 3D scene and hashes the result; tiny differences in graphics hardware, drivers and font rendering make the output vary between devices. TLS fingerprints describe how a client opens an encrypted connection. Cloudflare's documentation says they "identify TLS clients based on how they initiate connections," and that "mobile application traffic will produce the same JA3 fingerprint across devices and users," which is why a script claiming to be Chrome but connecting like a Python library is easy to spot.

The automation flag is the bluntest signal. MDN notes that navigator.webdriver "indicates whether the user agent is controlled by automation," and Chrome sets it when automation or headless flags are used.

How unique is a browser fingerprint?

Unique enough to matter, though less than the most quoted figures suggest. The classic number comes from the EFF's Panopticlick study by Peter Eckersley: of 470,161 browsers, 83.6% had a unique fingerprint, and the distribution held at least 18.1 bits of information, meaning a randomly picked browser would share its fingerprint with at most one in 286,777 others.

That sample was privacy-minded volunteers, who tend to have unusual setups. When researchers measured 2,067,942 fingerprints from one of France's 15 largest websites in Hiding in the Crowd, only 33.6% were unique. Among the 251,166 mobile fingerprints, just 18.5% were. The authors also found non-unique fingerprints are fragile: change a few features and a common fingerprint often becomes unique.

What we take from both studies:

  • Desktops are more distinctive than phones. Custom fonts, varied hardware and extensions give desktops more to measure.
  • Phones of the same model look alike. Fixed hardware and system fonts put millions of devices in the same bucket.
  • Rarity is the risk. A setup that nobody else has identifies you on its own.

Uniqueness isn't the whole picture, though. Platforms don't need a fingerprint to be globally unique; they only need it to match between two accounts on their own service.

How is device fingerprinting in apps different?

In an app, the platform doesn't get to run new probing scripts in a web page; it sees what its own app sends. That usually includes the device model, OS version, app version, language, an installation identifier and the events the app logs during a session. It's less varied than a browser fingerprint, because every copy of the app on the same phone model reports the same kinds of values the same way.

Mobile platforms also give apps ways to check the device itself. Apple's App Attest lets an app generate "a hardware-based, cryptographic key" for each user account on each device, certified by Apple as belonging to "a valid instance of your app." Google's Play Integrity API does similar work on Android, including recognizing a device flagged before. How Reddit, Instagram or Threads use these checks isn't public.

The practical difference for anyone running accounts: a browser exposes many measurable details that can drift or contradict each other, while app traffic is uniform. We explain what that means for automation in mobile app API vs browser automation.

Platforms treat a shared fingerprint as one piece of evidence that two accounts are related, and they weigh it with other signals before acting. A family laptop shared by two siblings produces two accounts with one fingerprint, and that alone isn't wrongdoing.

Reddit has been open about how this works for one feature. Its ban evasion filter "doesn't filter based on post or comment context" and is instead "informed by a variety of signals that imply certain accounts are related," including "how users connect to Reddit and information they share with us." Moderators choose between filtering levels, and Reddit warns that high filtering catches more suspected evaders with "possibly more false positives." That's what fingerprint-based linking looks like in practice: probabilistic, threshold-based, and combined with other data.

The signals platforms combine fall into four groups:

  1. Device: the browser or app fingerprint described above.
  2. Network: IP address, its type and location, and how often it changes.
  3. Account data: email and its pattern, phone number, recovery details, payment details.
  4. Behavior: login times, communities touched, posting rhythm, accounts interacting with each other.

One match in one group is weak evidence. Matches across several groups, repeated across many accounts, are strong evidence. We cover the Reddit side in more depth in how Reddit connects accounts to each other.

A worked example: two accounts, one laptop

Say a social media manager runs two clients' Reddit accounts from her own laptop. Client A's account uses her home Wi-Fi. For client B, she opens an incognito window and turns on a VPN set to another city.

Here's what the platform sees:

SignalAccount AAccount BLinked?
IP addressHome ISP, her cityVPN provider, another cityNo
Browser and OSChrome on macOS, same versionChrome on macOS, same versionYes
Fonts, canvas, screenHer laptopHer laptopYes
Timezone and languageHer settingsHer settings (mismatched with VPN city)Yes, and inconsistent
Login timesWeekday morningsSame mornings, minutes apartYes

The VPN changed one row out of five. The device and behavior rows still say "same person", and the timezone now contradicts the IP. Nothing here breaks a rule by itself: running two clients' accounts is legitimate. But if one client's account is ever suspended, the other now shares its fate. The fix is separation by design: each client's account gets its own stable network and its own device identity, every time.

What does this mean for agencies running many accounts?

It means every account that represents a different person or brand needs its own device identity and network, and both should stay the same for the life of the account. A real person's phone and home connection change rarely; an account whose fingerprint shifts every week looks less like a person, not more.

The setups agencies use, and how they handle fingerprints:

  • Physical phones: a real device fingerprint per account, at the cost of hardware and staff.
  • Cloud phones: virtual Android devices, each with its own identity, rented per minute or month.
  • Anti-detect browsers: one isolated browser profile per account with its own fingerprint values, which must be configured to stay consistent with the proxy.
  • Mobile API: software that presents each account as the official app on a phone, with a fixed device identity.

Viralis is built on the last approach. It runs Reddit, Instagram and Threads accounts through an iOS-based API, so each account appears as the official iPhone app with its own consistent device, app version and app sessions, and each account has its own proxy. There's no browser fingerprint to manage, because there's no browser.

Can you change or hide a fingerprint?

You can make a fingerprint less distinctive or keep separate ones apart, but you can't make a device show no fingerprint at all, and some attempts backfire. Privacy-focused browsers take two approaches: making every user look the same, or adding small random noise to measurements. Eckersley's paper warned that anti-fingerprinting tools "can be self-defeating if they are not used by a sufficient number of people": a rare protection becomes its own identifier.

The legitimate reason to manage fingerprints in multi-account work is separation, keeping one client's account from being linked to another's. It doesn't make prohibited behavior acceptable. Reddit defines ban evasion as using an alternative account to keep participating in a community after being banned there, and says it can lead to a sitewide suspension. A new fingerprint for a banned account is exactly that, and it's the case platforms build these systems to catch.

How to check and clean up your own setup

Start by listing, for every account you manage, which device or profile it logs in from and which network it uses. Any device or network that appears next to more than one owner is a link you didn't mean to create.

  1. Map accounts to devices and networks. One row per account: device or profile, proxy and region, timezone, language.
  2. Remove shared devices between different owners. Two clients on one browser profile, one phone or one laptop is the most common link we find.
  3. Align the layers. Timezone and language should match the proxy's region.
  4. Stop rotating. Pick one stable identity per account and keep it.
  5. Separate behavior too. Different login times and no interactions between accounts that belong to different owners.

Then make it routine: check the map whenever someone adds an account or a device, and review it monthly. For the network half of the setup, our comparison of residential, mobile and datacenter proxies covers which type fits which account.

Frequently asked questions

What is browser fingerprinting?

Browser fingerprinting is identifying or re-identifying a device from the details its browser reveals, without storing anything on it. The W3C describes it as a site's ability to identify a visiting user through configuration settings and other observable characteristics. Headers, IP address, screen size, fonts, graphics rendering and timezone each say little alone; combined, they can single out one browser among hundreds of thousands.

Does incognito mode change my fingerprint?

Mostly no. Private mode starts without cookies and history, but the browser, operating system, screen, fonts, graphics hardware and timezone are the same as in a normal window, so the fingerprint stays largely the same. That's why logging two accounts in from the same laptop, one in a normal window and one in incognito, still links them.

Does a VPN or proxy hide my fingerprint?

No. A VPN or proxy changes the IP address the platform sees, which is one network signal, but leaves the device fingerprint untouched. Worse, a mismatch between the two (a US residential IP with a browser set to a European timezone and language) can itself stand out. Network and device are separate layers, and both need to be consistent.

How unique is a browser fingerprint?

It depends on the population. The EFF's Panopticlick study found 83.6% of browsers in its privacy-minded sample were unique. A larger 2018 study of about two million fingerprints from a popular French website found only 33.6% unique overall, and 18.5% among mobile devices. Desktop browsers with unusual setups stand out most; phones of the same model look much alike.

Can platforms link accounts by fingerprint alone?

They rarely rely on one signal. Platforms weigh several: device fingerprint, IP and network, account details like email and phone, and behavior. Reddit says its ban evasion filter uses signals that imply accounts are related, including how users connect to Reddit, and lets moderators choose a filtering level that trades accuracy against catching more suspected evaders.

Is it legal to use anti-detect browsers?

Using a browser that manages separate profiles is generally legal, and agencies and ad teams use them to keep clients' accounts apart. Platform rules are a separate question: using a new fingerprint to return after a ban is ban evasion, which Reddit's rules prohibit, and faking engagement across accounts breaks every major platform's terms whatever browser you use.

Sources

  1. W3C: Mitigating Browser Fingerprinting in Web Specificationsw3.org
  2. Eckersley: How Unique Is Your Web Browser? (EFF)coveryourtracks.eff.org
  3. Gómez-Boix, Laperdrix, Baudry: Hiding in the Crowd (WWW 2018)hal.science
  4. Cloudflare Docs: JA3/JA4 fingerprintdevelopers.cloudflare.com
  5. MDN: Navigator.webdriverdeveloper.mozilla.org
  6. Apple Developer: Establishing your app's integrity (App Attest)developer.apple.com
  7. Reddit Help: Ban evasion filtersupport.reddithelp.com
  8. Reddit Help: What is ban evasion?support.reddithelp.com