Professional networks: how to judge profile claims

Profiles on professional networks such as LinkedIn are self-published claims. Nothing more. They are written by the subject, edited by the subject, and shown to you by a company with no obligation to check them. Treat one as you would a CV handed across a table: useful, informative, and unverified.

That is not a criticism of the category. It is the single most important thing to hold in mind when you use one of these services to research a person, a company, or a hire, because almost everything else people get wrong about them follows from forgetting it.

What to take away

Rank sources by how independent they are of the person making the claim.
If you are using a professional network to be found rather than to find, a different set of questions applies.
For research, keep a short record per claim rather than per person.

What a profile is evidence of

The profile shows	It is good evidence of	It is not evidence of
A job title and dates	What the person chooses to be known for	What they did, or that the role existed as described
An employer name	An association the person is willing to assert	Current employment, or that the employer agrees
Endorsements from contacts	That people are willing to click	Competence, and usually not first-hand knowledge
A recommendation written by a named person	That a named person wrote something positive	Independence: most are reciprocal or solicited
A verification badge	That the service ran some check	Whatever you assume the check covered
Connection count	Time spent on the network	Influence, seniority, or reach

Verification badges need context. A badge's meaning depends on the network's own definition and checks, which may change over time. On LinkedIn, a workplace verification tied to a work email supports access to an email associated with the workplace; it does not establish a person's job title, employment dates, or performance. Before relying on any badge, look for the service's current description of what it verified.

What a profile proves

Profile shows

Job title and dates: Chosen identity
Employer name: Willing assertion
Endorsements: Willingness to click
Recommendation: Someone wrote it
Verification badge: Some check ran
Connection count: Time on network

Good evidence of

Job title and dates: Actual role
Employer name: Current employment
Endorsements: Competence
Recommendation: Independence
Verification badge: Your assumption
Connection count: Influence or reach

Not evidence of

Job title and dates
Employer name
Endorsements
Recommendation
Verification badge
Connection count

The distinction in NIST Special Publication 800-63-3, Digital Identity Guidelines between establishing a person's identity and authenticating someone who is signing in to an account is useful here. Rely only on what the network says its specific check establishes.

Ranking sources by independence

Move from the profile itself toward sources with greater independence from the person making the claim. Not every claim has all five rungs; use sources that speak to the exact claim and stop when it is corroborated or further checking is not worthwhile.

The profile itself:evidence of what the person claims, not independent corroboration.
Another person-controlled source:a resume or personal website can show consistency, but is not independent confirmation.
An organization-controlled source:an employer's or school's own page can confirm what that organization publishes, such as a listed role.
An independent record or report:use one that directly bears on the claim. Examples include the SEC's EDGAR database for public-company filings, the relevant state's Secretary of State business-entity records for company registration, the relevant state licensing board's lookup for a professional license, or Crossref metadata for a DOI-linked publication. A publication record can support the publication claim, but does not by itself confirm a person's employment or identity.
Confirmation from the responsible organization:for an ordinary private-employer job-title or dates claim, ask the employer to confirm the specific fact at issue.

The verification ladder

Profile itselfself-asserted starting point
Consistency across the person's own outputs
Third-party mentions predating your interest
Institutional registersregistries, licensing, courts
Direct confirmation from a source you chose

Some research stops at the profile itself and calls it verified. The important gap is between a self-published claim and confirmation from a source independent of the person.

Searching well

For example, LinkedIn's internal search is designed for networking and recruiting, not independent research, and results can be shaped by your connection graph. Two people searching for the same name may get different lists. Professional networks vary in how much profile information they expose to search.

Search techniques that work

Search from outside the network domain
Search the claim, not the person
Try former names and variants
Find a unique identifier field
Note what you cannot see

Practical technique:

Search from outside as well as inside.For LinkedIn, a general web search for public profile pages can surface results its internal search buries and show you what a logged-out visitor sees.
Search the claim, not the person.If someone says they led a project, search the project. Independent traces are more likely to exist for the work than for the individual.
Search former names and name variants.Transliteration, married names, and initials break exact matching.
Watch for the near-match trap.Common names, similar employers, and the same city produce confident false identifications. Before you conclude two records are one person, find a field that works as a unique identifier: something only one person could hold.
Note what you cannot see.Some visibility is set by the subject, some by the network, some by whether you are logged in or paying. An absence of information is not information.

Write down which of these you did. A month later you will not remember whether you checked a registry or only meant to.

The structural questions, once you are building rather than researching

If you are using a professional network to be found rather than to find, a different set of questions applies.

Structural questions when building

Is the graph symmetric?

Yes

slow growth, obligation per link

faster growth, less meaning per link

Is the graph symmetric?

Networks built on mutual connections behave differently from networks built on one-way following. Asymmetric growth is also what turns a directory into a feed, and the pushed and pulled demand split describes what happens next. Symmetric graphs grow slowly, cap out, and carry an obligation: you are agreeing to be in someone's network as much as they are agreeing to be in yours. Asymmetric graphs grow faster and mean less per link. Many professional services now run both at once, which is worth understanding before you interpret either number.

Who decides who sees your posts?

Almost certainly a ranking system, not your connection list. That means the network is a publishing platform wearing a directory's clothes, and the same caution applies as anywhere else: reach is granted, not owned. The catalog's discussion of major social networks considers the same question.

Can you export the relationship?

This is the decisive one. Most of these services will let you download a list of your connections in some form. What varies is whether it includes anything you could actually use to contact a person outside the platform. A list of names is a souvenir. The distinction is the one that decides portability in messaging services too: what you hold is either an address you can write to or a number that says how many people could write to you. Test the export before you need it; the same practical question applies to photo platforms.

What does the platform sell?

Recruiting products, advertising, and data access all shape what the service optimizes for. A network that sells access to its members will design for the buyer's convenience, not the member's.

Recording what you found

For research, keep a short record per claim rather than per person. Note the exact claim, the source and where to find it, the date checked, what the source supports, what it does not establish, any unresolved question, and your conclusion.

Field	Note

Record per claim

Claimthe assertion in the subject's words
Where it came fromself, third-party, institutional
Independencedid the source predate your enquiry
Corroborationthe ladder step you reached
Unresolvedwhat would confirm it

Illustrative record (hypothetical):

Claim	Source	Date checked	Conclusion
A profile says the person held a stated role at a private employer	Employer confirmation	Date the employer responded	Corroborated only for the role and dates the employer confirmed; otherwise unresolved

Common questions

Is an unverified profile a red flag?

No. Most people never bother. A missing badge tells you almost nothing; an inconsistency between the profile and an independent record tells you a great deal. Spend your attention on the second.

Can I rely on employment dates?

Only loosely. Dates on self-published profiles are rounded, tidied, and sometimes wrong without any intent to deceive. If exact dates matter (for a background check, a legal question, a reference period), get them from the employer, not from the profile.

Someone's profile contradicts what they told me. What now?

Establish which one is more recent and which one had more at stake. People update profiles late and describe roles differently to different audiences. A contradiction is a prompt to ask a direct question, not a finding on its own.

Is it worth paying for the network's research tools?

Whether a paid tier is worth it depends on its current price and whether you need more searching or messaging; those features do not make claims better verified. Compare the price with the volume you need. For a small number of specific claims, prioritize independent records, such as confirmation from the employer or another relevant record holder.