How to judge celebrity venture case studies

This rubric for portfolio case studies 2027 tests evidence, not which study is best. A ranked list of the year's best case studies would be a list of the ones that were most fun to read, which is not the same as the ones that were right. So this page ranks nothing and hands you the rubric instead.

Use it on anything you are about to believe about a celebrity venture: a profile of a famous founder, a write-up of a brand deal, a thread that traces somebody's holdings. It works from the outside, without access to the author's notes, and most pieces fail it in the first two checks. No publication or author is named.

This site covers celebrity businesses and investments: the fame, ownership and ventures behind famous names. That makes the rubric here a celebrity venture rubric in practice. The claims you are checking are usually about who owns a brand, which company a famous person backed, when a trademark changed hands, or whether a launch was ever a real business.

The records are the same ones: filings, registers, trademark databases and dated announcements. Read the checks below with those ventures in mind. A study of a famous person's company deserves the same scrutiny as any other, and often gets less.

What to take away

Judge a study by whether each claim is matched to a record that could actually show it.
The strongest signal of quality is the presence of things the author could not find out.
A study with no dates on its evidence cannot be checked, and an uncheckable study is not evidence of anything.

The rubric

Passes when

Sourcing: Each factual claim names the record behind it
Dating: Every record is given with the date it was checked
Entity precision: Companies are named exactly, with jurisdiction
Inference marking: Guesses are labeled in the text as guesses
Absences: The piece says what it could not establish
Numbers: Figures are attributed to whoever produced them
People: Individuals appear only in roles filings give them
Corroboration: Sources are independent of each other

Fails when

Sourcing: Claims float free, or cite other coverage
Dating: Records appear as though permanent
Entity precision: The piece says "the company" throughout
Inference marking: Inference is written in the same voice as fact
Absences: Everything resolves neatly
Numbers: Figures appear unattributed
People: Motives and decisions are assigned
Corroboration: Several outlets repeat one announcement

Score nothing. Read the failures and decide what the piece can still be used for, which is often something, just not what its headline claims.

The three questions that do most of the work

Which record would show this? Take the study's strongest sentence and ask what document would establish it. If no document could, the sentence is inference regardless of how it is written. Match the conclusion to what the evidence can actually establish; the portfolio case studies list page sets out examples of study shapes and the conclusions they can support.

Three Questions That Do the Work

Which record would show this?

Yes

name the document that establishes it

the sentence is inference

When was this checked?

Registers are snapshots of the last filing. A study that says an entity is active without saying when it looked has told you about an unknown date. Filing duties and update schedules vary by jurisdiction and entity type, so check the relevant registry's own instructions rather than assuming a universal interval.

Does the ownership claim rest on a filing?

For a U.S. trademark claim, the USPTO Trademark Search is the public search interface for the USPTO database of registered trademarks and prior pending applications. A trademark record can test a claim about a mark, but by itself it does not establish who owns the company using it. Company ownership is checkable only where evidence actually establishes it; anything else is a connection the author drew.

What a good study contains that a bad one does not

A list of what was searched and found nothing.
Exact entity names with jurisdictions, not trading names.
Dates attached to each record rather than to the article.
At least one sentence saying what the author could not determine.
A correction note where something was changed after publication.

The fourth item is the tell. Real research in this category hits walls constantly. A piece with no walls either had extraordinary access, in which case it should say so, or filled the gaps.

Good Study vs Bad Study

Good study

Searches: Lists what found nothing
Entities: Exact names, jurisdictions
Dates: Attached to each record
Limits: Says what it could not determine
Corrections: Notes changes after publication

Bad study

Searches: Omits dead ends
Entities: Trading names only
Dates: Attached to article
Limits: Everything resolves neatly
Corrections: No correction note

Do not fill a gap with an unsupported line: label it unknown and say what record or access would be needed. The portfolio case studies portfolio page gives further guidance on distinguishing evidence from unsupported claims.

What professionals do that readers cannot

A buyer assessing the same business runs due diligence: they receive accounts, contracts and customer data under an agreement, and they spend months on it. Nothing published without that access can match it, and a study that reads as though it had that access is claiming something it does not have.

Public research can establish structure and dated claims when linked records support them; it does not, by itself, disclose private economics. The portfolio case studies page gives a fuller evidence ladder for separating documented facts from inference.

Applying it to the language, not only the sources

Coverage in this field goes wrong through verbs more than through facts. For example, test a study claiming Kylie Jenner owns Kylie Cosmetics: if it cites only a trademark result, that supports a claim about the mark's record, not equity ownership in the operating company. Treat the broader claim as inference unless the study supplies evidence that actually establishes that ownership. The celebrity companies examples page offers more examples of how claims can slide between different arrangements.

Why we publish no ranking

Ranking celebrity venture case studies would require judging accuracy against records that a ranker may not have. It could reward polish over evidence. The best celebrity companies 2027 page addresses ranking companies; this page instead gives readers a method for testing case-study claims.

Common questions

Is a study that fails several checks worthless?

No. It may still be a useful map of what has been claimed publicly. Use it as a lead list rather than as a finding.

What if a study cites documents I cannot access?

Ask whether the document is one a reader could obtain in principle. Registers and filings are; private agreements are not, and a citation to one implies access that should be explained.

Should a study name individuals at all?

Only in the roles that filings give them, and only where naming serves the reader. A person listed as a manager is a person listed as a manager, and nothing more follows from it.

How long does applying this rubric take?

About ten minutes for a long piece, most of it spent on the first two checks. Those two are where nearly everything fails.