Merchant Pool device fingerprint script (Microsoft DFP)
Loads mdt.js from merchantpool1.linkedin.com and runs a device fingerprint call.
In one line
Can collect device traits from a Microsoft fingerprint script on LinkedIn pages
Loads mdt.js from merchantpool1.linkedin.com and runs a device fingerprint call.
A second device fingerprinting layer, apparently Microsoft's fraud protection technology (inferred from cookie names and a doFpt call), keyed to LinkedIn's browser ID.
What it can and can't do
What the user does: Nothing. It happens when a LinkedIn page loads the script
It can
- Load a script from a LinkedIn payments host
- Collect traits like screen size, fonts and browser settings for a device fingerprint
- Send those traits to Microsoft for fraud checks
It cannot
- Show you anything on the page
- See your files, other tabs or other apps
- Run if the script is blocked
- Name you without a login
Browser support
| Browser | Status | First version | Note |
|---|---|---|---|
| Chrome | Supported | Not available | None |
| Edge | Supported | Not available | None |
| Firefox | Supported | Not available | None |
| Safari | Supported | Not available | None |
| Safari on iPhone | Supported | Not available | None |
| Chrome on Android | Supported | Not available | None |
| Samsung Internet | Supported | Not available | None |
How a site asks for it
| API | script https://merchantpool1.linkedin.com/mdt.js?session_id=<bcookie>&instanceId=<uuid>; window.dfp.doFpt(document) |
| How access is granted | none |
| What the user sees | None (no browser prompt) |
| Needs a secure context (HTTPS) | No |
| Needs a user gesture | No |
| How long it lasts | Cookie dfpfpt per LinkedIn table; Microsoft fptctx2 inferred |
| How to take it back | Not listed |
| Inside an iframe | Not listed |
Typical flow
- LIX flag pemberly.tracking.dfp.integration enables the load
- mdt.js is requested with session_id set to bcookie and a fixed instanceId (prod fb6bbd47-fa7c-4264-b4e9-b25948407586)
- Script calls window.dfp.doFpt(document)
Data you get
What comes back: Vendor fingerprint
Shape: object
How often: per page
| Field | Type | Unit | Range | Precision |
|---|---|---|---|---|
session_id | string | bcookie | Not listed | stable per browser |
Technical limits
- Link to Microsoft Dynamics 365 Fraud Protection is inferred, not stated by LinkedIn
- What mdt.js collects was not deobfuscated in this research
The story
What it is
A script from a LinkedIn server runs a device fingerprint check tied to your browser ID.
For you
You never see it. A script collects clues about your device, such as screen and software traits, to build a fingerprint. Public analysis suggests it comes from Microsoft's fraud tools.
For makers
A device fingerprint lets a platform tell if the same device creates many fake accounts or ads. It works even when cookies are cleared. The same trick can also track a person across visits, which is why privacy groups push back.
Where you see it
| Type | Example | Apps |
|---|---|---|
| security | A fraudster opens twenty fake ad accounts from one laptop. The fingerprint links them as one device. | LinkedIn Ads |
| payments | A stolen card is used to buy ads, and the fingerprint helps spot the device behind many cards. | LinkedIn Ads |
| social | A banned account owner comes back with a new email, and the device fingerprint links the new account to the old one. | |
| games | A person makes many accounts to cheat on a game leaderboard, and a fingerprint can reveal one device behind them. | LinkedIn Games |
| ads | Ad review teams use the fingerprint to remove clicks from the same device that clicks the same ad hundreds of times. | LinkedIn Ads |
| analytics | An analytics team counts unique devices rather than cookies, which stays steady after people clear cookies. | None named |
Risks
- Works without cookies
- Cannot be easily reset by you
Good uses and abuse
Legitimate uses
- Payment and account fraud screening
Known abuse
- Persistent identification
Scores
| Score | Value | Meaning |
|---|---|---|
| Sensitivity | 3 of 5 | How revealing the data is |
| Utility | 2 of 5 | How useful it is to honest apps |
| Friction | 0 of 5 | How much the user has to do |
| Fingerprint bits | 15 | Rough identifying entropy, when known |
A real example: LinkedIn
URL, instance id and doFpt call seen in the snapshot; the Microsoft DFP product link is an inference from names.
Related capabilities
Related
- APFC / DNA device fingerprint (abuse prevention features). Can collect about 48 device traits and send them in secret with later requests
Sources
How well checked: Inferred from related behavior.