CiteUnseen

Rubric

Topical coverage and authority

Rubric v0.8 · 4 checks · 15 of 100 points

Do you answer the questions a buyer actually asks?

Coverage decides whether you are a candidate answer at all. A question with no owned page cannot be answered from your site by anything.

Last updated · Written by Chris Brock, who wrote the rubric and signs the audits that use it.

Why this dimension exists

Coverage is where a site goes from retrievable to worth retrieving.

An engine answering a question about your category fans out: one typed question becomes five or ten related retrievals. A site that answers only the question its homepage was written for misses most of them. The remedy is pages that answer what a buyer asks before, during and after that question, including the comparisons you would rather they skipped.

The four checks look for first-hand material as well as breadth, because anyone can generate forty pages on a topic. What cannot be generated is what you learned doing the work: the number you measured, the case that went badly, the tradeoff you made. That is the reason to cite you over other sites saying the same thing more fluently.

The 4 checks

1 of these 4 is deterministic, which means the free scan settles it without anybody reading the site.

TA-01 — Priority-question coverage

4 points · Proposed by machine, rated by a person

What is tested. Build a bounded question set from products, audience, sales objections, and customer journeys, then map each question to a substantive page.

What is recorded as evidence. Question-to-page coverage matrix.

What fails it. Major buying questions with no owned answer; multiple thin pages competing for the same question.

How to fix it.

  1. Take the priority questions recorded at intake and check each against the site.
  2. For any question with no page that answers it, write one.
  3. Where a page half-answers it, add the missing part rather than starting again.
  4. Link related questions to each other so a reader following one lands on the next.

How to confirm it is fixed. Search your site for each priority question and see whether one page answers it.

Reference.

TA-02 — First-hand experience

4 points · Proposed by machine, rated by a person

What is tested. Identify implementation detail, original research, public repositories, case evidence, operator lessons, demonstrations, or reproducible examples.

What is recorded as evidence. URLs and the specific first-hand contribution.

What fails it. Generic summaries that add no experience, method, data, or artifact.

How to fix it.

  1. Add what you have actually seen: the cases, the numbers, the things that went wrong.
  2. Say who did the work and when.
  3. Show the artefacts where you can: screenshots, extracts, before and after.
  4. Cut the passages that could have been written by somebody who had never done the work.

How to confirm it is fixed. Ask whether a competitor could publish the same page with their name on it. If yes, it carries no experience.

Reference.

TA-03 — Comparative and decision content

4 points · Proposed by machine, rated by a person

What is tested. Inspect honest alternatives, tradeoffs, fit boundaries, limitations, and selection guidance.

What is recorded as evidence. Comparison pages, decision tables, limitation statements, supported criteria.

What fails it. Competitor-name pages with no substantive comparison; universal claims of fit.

How to fix it.

  1. Write the comparison a buyer is already making, including the alternatives you lose to.
  2. State the criteria before the verdict, and use the same criteria for every option.
  3. Say plainly who each option suits and who it does not.
  4. Keep it current and put the date on it; a comparison a reader cannot date is one they cannot trust.

How to confirm it is fixed. Give the page to somebody outside the company and ask them which option it recommends and why.

Reference.

TA-04 — Internal topic relationships

3 points · Settled by machine, from the evidence

What is tested. Map internal links among product, educational, evidence, comparison, and trust pages. A URL whose declared canonical names another sampled page counts as that page: its inbound links are credited to the canonical and it is not a page in its own right.

What is recorded as evidence. Sample graph or relationship table.

What fails it. Orphaned authority pages; circular navigation; important evidence disconnected from commercial pages.

How to fix it.

  1. Link related pages to each other with anchor text that says what the target is about.
  2. Make sure no substantive page is reachable only from the sitemap.
  3. Group pages that belong together under a hub page that links to all of them.
  4. Remove links that exist only for volume; they dilute the ones that mean something.
  5. If a URL you expected in this finding is missing, it declares a canonical naming another page you were shown: it is counted as that page and needs no links of its own.

How to confirm it is fixed. Start at the home page and try to reach each important page in three clicks. Query-string variants of a page you can already reach do not need their own route.

Reference.