Evidence: proof the drafts can cite
Evidence under Hubs holds the claims your drafts may stand on. Proof as claims, not pages: each one is a single statement, the words on a page it rests on, and the page it came from.
Where the claims come from
- Find sources lists what you have already published: your site and the documents on your positioning (the pages Positioning has read), and your live hub articles. It reads nothing new.
- +Add a page adds a third-party page: a study, a regulator's page, an analyst figure, a peer's public case. Claims from it count as third-party evidence.
- +Add a document takes a PDF, Word or PowerPoint file from your Library. Say whether it is yours (a case study, a whitepaper) or a third party's (a study, an analyst report); a third party's claims count as third-party evidence. It is converted to text once and read straight away. A scan with no selectable text cannot be read; add a version with real text.
- Read (or Read new for every source not read yet) pulls the checkable claims out of a source. The AI suggests them; the code keeps a claim only when the exact words it rests on are on that page. A claim it cannot point at is dropped, and the Dropped column says how many.
What a claim carries
- Grounds, its kind: a mechanism (how something works), a number from your own work, a named customer, a testimonial, a third-party figure, or a perspective (a position you take, with its reasoning). Evidence proves results; perspective proves judgment.
- Rests on: the words on the page.
- Named: the customer, person or company it names, if any.
- Pillar: which pillar of your positioning it supports. Tag it, and pools on that pillar see it first.
- ICP: which target it speaks to, or every ICP. Baseline, Coverage reads it.
Approve, reject, name
Nothing reaches a draft until you approve it. Rejected claims are not offered again when the source is read again; neither are approved ones, which stay as they are.
A named customer stays unnamed in every draft until you tick May be named. Tick it only when the customer has agreed. Until then the drafts see "a customer" in its place.
Where approved claims go
- Composer: each pool's drafts see the approved claims, strongest first and their pillar first. A row's one true thing rests on one of them, or on your material, or the row says less. A claim's link becomes a link the row may use. Perspective claims carry the teach and reframe beats and are never offered as proof of a result. Each draft shows which claims it used.
- Article AI Assist: drafts get an evidence brief: what you can prove, with links to cite, and the positions you have already taken, so a new article stays consistent with them.
An article you write is not proof of your own claim. It is how proof gets to the reader. Proof that only a customer can give cannot be written into existence; ask the customer.
What adds the most next
At the top of Evidence, Suggest reads your approved claims against your positioning and lists the three pieces of work that add the most credibility now.
Credibility is proof times perspective. Evidence proves results; perspective proves judgment; a pillar with one and not the other is weaker than it looks. So when proof is thin the suggestions lean to proof, and when you have proof but no stated position they lean to a position. Nobody asks you to choose. Each suggestion says why it is next, ranked by what it adds on the pillars your live campaigns write on, and a batch is never all one kind.
- Outside source: a third-party page found for a pillar where outside proof adds the most. Add and read pulls its claims in for you to approve; approved, they are citable the same day. Searching needs the Exa key set up for your account; until then this kind does not appear.
- Article to write: an article that extends what you have already said: what it does for a reader (teaches, reframes, proves), the position it takes, the claims it should cite and any outside data to find. Draft opens the article composer with that brief and those links filled in.
- Perspective series: when no pillar has real proof yet, five pieces over a quarter: the problem named, the mechanism, the contrarian position, the predictions, the method. Positions earn credibility before results exist.
- Ask a customer: proof only a customer can give, for pillars your campaigns use.
An article you write is not proof of your own claim, so no suggestion is a piece that only praises. The drafts flag the thought-leadership register ("the future of", "paradigm shift") so it can be edited out.