Quick Answer
Betty Crocker is a fictional character created by the Washburn-Crosby Company in 1921 to answer consumers' baking questions, and owned by General Mills after a 1928 merger. By 1945, Fortune reported, a written policy required that anything said by or credited to Betty be literally true of a current member of the home economics staff. Applied to AI content, the rule is short: every claim must be true of a named person on your team today.
More of what your company publishes is now drafted by a model, edited by a marketer and posted under a founder's name. None of that is a problem on its own.
The problem is the sentence nobody on the team can stand behind. The lesson the founder never learned, the customer pattern nobody saw, the depth the company does not have. Most AI content policies ask if a draft was reviewed or disclosed, which is a weaker question than the one General Mills was already asking by 1945.
This covers what the 1945 record actually says about Betty Crocker, where the popular retelling drifts from it, and a five-step version of her rule you can run on AI-drafted posts this week.
How to Apply the Betty Crocker Rule to AI Content
You apply the Betty Crocker rule by tying every sentence a model drafts to a real, current person who could defend it on a call tomorrow. If no such person exists, the sentence does not ship, however good it sounds.
The five steps below turn the 1945 policy into something a content team can run on an ordinary Tuesday. Each one covers what it is, how it works, an example, how to run it, how you know it worked, and the mistake teams usually make.
1. Name the Person Each Claim Is True Of
The core of the Betty Crocker rule is attribution before publication. Every factual claim, lesson or opinion in a draft gets a name beside it: the person on your team of whom that sentence is literally true.
Take a founder post that says the team rebuilt onboarding after watching customers stall on the same screen. The name beside it is whoever sat in those calls, and if nobody did, the model invented a memory and the sentence goes.
Run it as a column in your review document, one row per claim. You know it worked when a reviewer can point at any sentence and name its owner in seconds. The common mistake is writing the founder's name against everything, which just moves the fiction up one level.
2. Say We When the Knowledge Belongs to the Team
The 1945 policy told copywriters that Betty should say "we" rather than "I" whenever possible. Presenting her as a single woman who wrote every recipe and answered every letter was, in Fortune's words, "unsound and perhaps illegal".
Your AI-drafted guide has the same shape. It was assembled from sales-call notes, support tickets, a product manager's doc and a model's draft, so "we recommend" is honest and "I have seen" usually is not.
Run it by searching every draft for "I", "my" and "me", then checking each one against the named person's actual experience. If someone writes in your voice, whether a ghostwriter, a model or Better Marketing, the "I" should cover only what you did, saw or believe.
You know it worked when the founder would say every "I" sentence out loud to a prospect. The mistake is switching everything to "we" to dodge the question, because "we" still has to be true of somebody on the team.
3. Keep the Voice Inside What Your Team Knows
Betty was told to stick to home economics and never discuss a private life. That is a scope rule, and every voice your company publishes in needs one: the founder, the company blog, the support persona with a human first name.
Content marketing for tech companies breaks here more often than anywhere else. A model will write a confident paragraph on security, compliance or where your market is heading, and your team may genuinely know two of those three.
Run it by writing one line of scope per voice and pinning it to the top of the brief the model receives. You know it worked when out-of-scope sentences get cut in review rather than softened with a hedge. The mistake is letting the model fill the gaps with plausible generalities, which read fine and teach the reader to distrust the rest.
4. Put a Named Byline on Anything a Reader Would Question
A byline is the modern version of a current staff member. Google's own guidance on AI-generated content says "You should consider having accurate author bylines when readers would reasonably expect it", and Google ties knowing who created a page to E-E-A-T.
An opinion piece on pricing, a teardown, a lesson from a failed launch: each is a page where a reader thinks "who wrote this?" and deserves an answer. A changelog is not.
Run it by bylining every opinion or experience page to the person who will answer an email about it. You know it worked when a reader's objection reaches that person and gets a real reply. The mistake is a brand byline on a personal opinion, which is a persona speaking without the policy.
5. Review the Archive When the Named Expert Leaves
The most overlooked word in the 1945 clause is "current". A claim that was true of an engineer who left last spring is no longer true of anyone on your staff.
AI drafting makes this worse, because a model fed your archive keeps repeating the departed person's judgment in your company's voice. The page reads the same, while the person who could defend it is gone.
Run it as a step in offboarding: list every page carrying the leaver's name, experience or method, then reassign, update or retire each one. You know it worked when no live page implies expertise the team no longer has. The mistake is leaving an ex-employee's "I" posts on the company blog because they still get traffic.
What the 1945 Fortune Profile Says About Betty Crocker
The April 1, 1945 Fortune profile of General Mills treats Betty Crocker as a composite, not a person. Fortune's own archive now carries the piece, noted as a version of the story from that issue.
Part of Betty, Fortune wrote, was Marjorie Child Husted, director of home service, whose forty assistants thought up recipes and advice. Another part was the company's chief nutritionist, another the radio voice in Chicago, and another the mail room.
That mail room opened the 4,000 letters Betty averaged each working day and answered about 80 percent with form letters. The rest went up to the home service department for individual attention.
Governing all of it was a Betty Crocker policy, drawn up by the advertising and legal departments as a constant guide for the copywriters who wrote her lines. Fortune lists its clauses:
- Betty must be dignified, always in the style of a gentlewoman.
- She should be friendly but not intimate.
- She must stick to home economics and never discuss her private life.
- She should say "we" rather than "I" whenever possible.
- Everything she says must be true of a real, current member of staff.
Fortune gives the last clause verbatim, as a rule adopted "for some esoteric legal reason". The clause reads: "anything said by Betty or credited to Betty must be literally true with respect to some current member of the company’s home economics staff." You can read it in Fortune's archived 1945 profile.
The same article puts a number on the asset. General Mills carried Betty on its balance sheet at one dollar, Fortune reported, against about ten million dollars spent making her what she was.
Where the Betty Crocker Legend and the 1945 Record Differ
The Betty Crocker legend and the 1945 record differ on how she started, who answered the mail and how famous she was. Those gaps matter here because they are exactly the kind of detail a model repeats with confidence.
| Claim | Popular Retelling | 1945 Fortune Profile |
|---|---|---|
| How it started | A jigsaw-puzzle ad, 30,000 puzzles returned | A recipe contest, several thousand entries |
| Early questions | Several hundred letters, or thousands | Not stated |
| Who answered | Each letter answered personally | By 1945, about 80 percent by form letter |
| How famous | Second best-known woman, ahead of Eleanor Roosevelt | Best known to housewives, except Eleanor Roosevelt |
| Evidence for fame | Surveys of homemakers | A company survey: 91 percent knew of her |
The puzzle version comes from General Mills' own 2024 history of Betty Crocker, which describes a Saturday Evening Post ad placed by Washburn-Crosby. A 2022 General Mills profile says the company was flooded with thousands of questions, and the 2024 page says several hundred.
The same 2024 page says Fortune named Betty the second best-known woman in America, "followed by First Lady Eleanor Roosevelt". Fortune's text says the opposite order: Betty was the best known "with the exception of" the First Lady, which makes Roosevelt first.
Fortune also published no poll to support the line. Its only figures on her fame came from a company survey, in which 91 percent of housewives knew of Betty and 56 percent correctly tied her to General Mills.
The hand-answered letters are not strictly wrong either. General Mills describes home service staff personally answering the radio mail of 1924, while Fortune describes the 1945 operation, two decades and many thousands of letters later.
Disclosure Alone Did Not Stop People Believing in Betty Crocker
Telling the public that Betty Crocker was not a person did not stop the public believing she was. Fortune explained her composite nature to a national business readership in 1945, and in the same article reported that ten men had written to her proposing marriage.
The pattern outlasted the war. The Minnesota Historical Society's MNopedia records that after the test kitchens moved to Golden Valley in 1958, tour guides often found themselves consoling visitors shocked to learn she was not real.
Disclosure tells the reader something. Substantiation protects them, because a claim that is true of a real person stays true whatever the reader believes about the voice delivering it.
That gap is wider in AI answers. When ChatGPT, Perplexity or Google's AI Overviews quote your page, the claim travels and the disclaimer at the foot of the post usually does not.
So an AI SEO service worth paying for starts by asking which claims on a page could survive being quoted alone, with your name attached and no context. A disclosure line is worth adding where readers would expect one, and it is never a substitute for the claim being true.
Modern Advertising Rules Already Echo the Betty Crocker Policy
American advertising rules now say in general terms what the Betty Crocker policy said about one persona. The Federal Trade Commission's Endorsement Guides state the principle in plain words: "Endorsements must reflect the honest opinions, findings, beliefs, or experience of the endorser."
The same section adds a time element that mirrors "current". An advertiser may keep using an expert endorsement only while it has good reason to believe the expert still holds the views presented.
In August 2024 the Federal Trade Commission went further with a final rule on fake reviews and testimonials. It targets reviews that claim to come from someone who does not exist, with AI-generated fake reviews named as the example. It also covers reviews from people with no actual experience of the product.
| Betty Crocker Clause | Modern Echo | Where It Lives |
|---|---|---|
| Literally true of a staff member | Honest experience of the endorser | FTC Endorsement Guides |
| A current member of staff | Endorser still holds the views | FTC Endorsement Guides |
| No superwoman answering her own mail | No reviews by people who do not exist | FTC rule, August 2024 |
| Say we, not I | Accurate bylines where readers expect them | Google Search guidance |
None of these rules demands everything General Mills demanded. They govern endorsements, reviews and search quality, and a blog post can clear all three while still claiming experience nobody at the company has.
The platform rules for AI-generated content, from Google Search to Amazon KDP, are laid out in what the published anti-slop policies actually say. For search, E-E-A-T is the closest echo, and it is why an AI SEO service reads a byline as evidence rather than decoration.
When the Betty Crocker Rule Is the Wrong Tool
The Betty Crocker rule is built for voices that make factual claims, and it is the wrong tool in a few ordinary cases. Applying it everywhere just slows the pages that never needed it.
- A clearly fictional mascot that makes no claim of expertise, where nobody is misled about what it knows.
- Product documentation and changelogs, where the product itself verifies each statement.
- A solo founder writing their own posts, where the rule shrinks to checking that you did what the draft says you did.
It is not a reason to relax outbound. A sequence that a B2B demand generation agency sends in a founder's name is the voice a prospect tests on the first call, so every line in it needs an owner.
The rule also costs something, and it is fair to say so. Expect a first pass to delete the confident, general paragraphs a model is best at producing, which is exactly why content marketing for tech companies so often skips it.
The same discipline applies to numbers. A figure nobody on the team can trace is a claim with no owner, a problem covered in what plausible-looking numbers teach about trust.
Write Your Own Betty Crocker Policy Before You Scale
General Mills wrote the Betty Crocker policy down, and yours fits on one page. Every claim true of a named, current person; "we" for team knowledge; a scope for each voice; a byline wherever a reader would ask; an archive review when someone leaves.
That page is what makes founder-led outreach believable, and it is where Better Marketing starts its work as a B2B demand generation agency: warm trust first, then first customers.
Trust is built in specifics, never in slogans.