The Botworthy Score™: a research-weighted score for whether AI models will quote your content

The Botworthy Score™ is a 0 to 100 score for how likely AI models are to quote a page, across 10 criteria weighted by published research into what AI models actually cite. I built it, I coined the name, and as of this month it lives at botworthyindex.com with a free scoring skill, a £20 a month membership and a sector index on the way.

This is the news update on all three: what the score measures, why I turned a client rubric into a product, exactly what members get, and where to start.

Key points

• The Botworthy Score measures what is on the page: whether it answers first, whether a passage can be lifted out cleanly, whether there is evidence behind the claims, and whether it is current.
• Every criterion carries a weight set by citation research, not opinion. When the research changes, the weights change with it.
• The free scoring skill stays free. It is a fixed snapshot of three studies and it will not be updated.
• Membership is £20 a month plus VAT. You get the live score, the evidence behind every weight, a monthly research brief, the current skill and Index data as it lands.
• The Botworthy Index will rank the content of the leading brands in one sector at a time.

What is the Botworthy Score?

The Botworthy Score is a measure of how quotable a page is to an AI model. When ChatGPT, Claude, Gemini or Google’s AI Overviews answer a question, they build the answer from passages that make sense on their own. The score rates how well a page is written for that job, on a scale of 0 to 100.

It asks four questions of every page. Each question is an area of the score, and each area holds several criteria.

Area	What it asks	Example criteria
Answerability	Does each section answer its heading in the first sentence?	Answer-first sections, question-led headings, a summary at the top
Extractability	Can a passage be lifted out and still make sense?	Self-contained passages, short sentences, lists and tables
Evidence	Is there something worth citing?	Statistics and specifics, named sources, named expertise
Freshness	Has it been updated recently?	Date of last substantive update

The weights come from the largest citation studies available: the Princeton GEO paper (Aggarwal et al., KDD 2024), Cyrus Shepard’s Zyppy meta-analysis of 54 studies, and Ahrefs’ analysis of almost 17 million AI citations. Where the evidence for a criterion is strong, it carries more weight. Where it is thin, it carries less. The full criteria, the scoring rules and the statistical checks behind them are in the rubric write-up.

Why “Botworthy”? Because the reader of your content is now, very often, a bot. The question is no longer only whether a page is worth a person’s time. It is whether a page is worth a bot’s quote.

Why I built it

I built the Botworthy Score because the GEO advice I kept reading had no numbers behind it, and the one measurement I did make kept changing underneath me.

The first version was a rubric I wrote to answer a client’s question. A major UK insurer asked how AI-ready their home insurance knowledge centre was, and I did not want to hand over a list of opinions. So I scored 32 articles, double-coded the judgement calls, ran significance tests and stress-tested the weights. The average article scored 45.6 out of 100. 43% of sections buried the answer. Question headings, the most repeated piece of GEO advice going, made it worse: sections under a question heading answered directly 38% of the time, against 67% under statement headings.

I published the method, and the response on LinkedIn told me two things. Content teams want a score they can defend to a sceptical director, and most of them do not have one. That made the case for a product rather than a one-off audit.

The second reason is the research itself. New studies on what AI models cite come out every month, and they do not all agree. A fixed checklist written in the spring is already behind by the autumn. A score only stays useful if someone reads each new study, tests whether it changes anything, and re-weights the criteria to match. That is a monthly job, and it is the job the membership pays for.

What you get

There are two versions of the score, and the free one is not going anywhere.

	Free version	Live version (members)
What it is	A fixed snapshot of the score, built from three published studies	The score as it stands today, re-weighted as new research is published
Where it runs	A Claude skill, from GitHub	The current member skill, scoring 25 to 40 articles at a time
Updates	None. The weights do not change	Every month, with a record of what moved and why
Price	Free	£20 a month plus VAT, cancel any time

Members get five things.

1. A monthly research brief. Every new study on what AI models cite from the past month: what it tested, how big the sample was, what it found, and whether it should change how you write. Plain English, with links to the papers. It arrives in the first week of each month.
2. The live scoreboard. Every criterion, the weight it carries and the studies behind it. When a weight moves, the scoreboard shows what moved, by how much and which research moved it.
3. The latest scoring skill. The current member version, which runs in Claude and scores a whole content library against the live weights.
4. Patterns across your team. The skill groups weak sections by the writing habit behind them, so you can see which habits cost citations across your content team, build the training to fix them, and rescore to prove it worked.
5. Index data as it lands. Sector scores from each Botworthy Index study, so you can benchmark against the leading brands in your market.

Membership costs £20 a month plus VAT, so £24 for UK members. Businesses outside the UK can enter a VAT or tax ID at checkout. If you cancel, you keep everything you have downloaded.

What is coming next

Three things, in order.

• The Botworthy Index. I am scoring the content of the leading brands in one sector at a time and publishing the rankings. You can sign up to get the first Index when it is out.
• Rescoring after rewrites. The next client studies will rescore pages after rewrites and training to measure the change, which turns the score from a diagnosis into evidence.
• Linking scores to citation data. In the first study, the highest-scoring articles were also the ones AI models cited most. Matching scores to AI visibility data across more libraries is how the weights get sharper.

Where to start

• Read about the Botworthy Score at botworthyindex.com, including the writing habits that cost the most citations.
• Get the free scoring skill and run it on a handful of your own pages.
• Join Botworthy for £20 a month if you want the live score and the research behind it.
• Read the original study if you want the method, the statistics and the eight habits that bury the answer.

Botworthy is a service of Work That Limited. If you have questions about the score, the membership or the Index, email jessica@didgeheads.com.