Social Media · linkedin.com
13 out of 20 Compliance-bound
A findable, readable policy in front of an AI training setting that is switched on until you go and switch it off.
How many clicks to find the privacy policy from the homepage?
One click, from the homepage footer. Opens as a navigable web page with a clear title and the effective date shown prominently.
Is there a meaningful choice about cookies and tracking before data is collected?
No banner, and consent folded into signing in. Nothing appears on first load; agreement is implied through small print in the sign-in flow. A guest opt-out of targeted advertising does exist under “Guest Controls” in the footer, but it is two clicks deep, cookie-based, and opt-out rather than opt-in, so it disappears with the cookie that records it.
How hard is the policy to read, how much of it is there, and can you get it in one piece?
Grade 11.6 across 6,863 words, in one place. Difficulty 2, length 2, access 4. The introduction runs harder, around Grade 14, under the weight of defined terms; the sections that actually describe what is collected are the more readable part. Measure it yourself.
How many clicks to reach data settings, and can you delete your account yourself?
Deletion in three clicks, and a default worth knowing about. Me, then Settings & Privacy, then Close and delete account. Self-serve deletion works for standard accounts. The finding that matters is next to it: “Data for Generative AI Improvement” is toggled on by default, so personal data and posts train generative models unless a person finds the setting and turns it off.
Find the support page. How many clicks, and is there a way to speak to a human?
A help centre and no way through it. One click from the footer, and identical logged in or out. No phone, no chat, no email, no human at any point.
Overall finding
13 out of 20 Compliance-bound
The Transparency score of 4 flatters the rest. LinkedIn publishes a findable, well-presented policy and then structures the choices behind it to discourage use: consent implied rather than asked, an advertising opt-out buried in guest controls, and AI training on by default. A platform whose entire product is professional trust scores lower on Consent than Facebook does.
- Scored
- 11 April 2026
- Revised
- 6 August 2026
- Region
- Canadian site, private window
- Session
- Logged out, except Control
Scored from public evidence only. No company can pay to change a score, and paid relationships are disclosed here. Read the independence statement.