What Site Brain improves.
Seven playbooks. Search visibility is being tested on my own sites and the other six are being built; as each one is ready, you choose whether it runs on your site. A playbook is a file: what to look for, what to propose, what evidence to attach, what risk to declare. You can read mine, switch them off, and add your own.
What each one looks for
A proposal from any playbook arrives in the same form: the page, the exact change, the evidence, what it expects, the risk and how to undo it.
Search visibility
Testing- Looks for
- Pages that get impressions in Search Console and no clicks.
- Proposes
- A rewritten title and description. Clearer headings, redirects and merges for pages that compete with each other are being built.
- Evidence
- The Search Console rows for the page.
- Typical risk
- Low to medium - a new title on a busy page deserves a second look.
Visibility in AI answers
Building- Looks for
- Whether AI assistants cite your pages when they answer questions in your field, and which of your pages are hard for them to read.
- Proposes
- Clearer summaries, direct answers to the questions people ask, and markup that makes a page easy to quote.
- Evidence
- Sampled answers, with the sources each one used.
- Typical risk
- Low - the page gets clearer for people too.
Page speed
Building- Looks for
- Slow pages, heavy images, scripts that block the page from drawing, and layout that jumps around.
- Proposes
- Right-sized images, a better loading order, and removing what the page doesn't need.
- Evidence
- Core Web Vitals and lab measurements for the page.
- Typical risk
- Low to medium - scripts that other tools depend on are flagged first.
Structured data
Building- Looks for
- Pages with no markup, or markup that's wrong or out of date.
- Proposes
- Product, article, FAQ, organisation and breadcrumb markup that matches what's on the page.
- Evidence
- Validator results before and after.
- Typical risk
- Low - nothing visible changes on the page.
Stale content
Building- Looks for
- Pages that lost traffic, state facts that have changed, or link to pages that have gone.
- Proposes
- A list of what to refresh, the facts that changed, and a draft of the update.
- Evidence
- The traffic trend, the date of the last change, and the broken links.
- Typical risk
- Medium - it's your content, so a person reads each draft.
Conversion paths
Building- Looks for
- Forms that are hard to find or never reach the CRM, and calls to action nobody clicks.
- Proposes
- A clearer call to action, a shorter form, or a fix for a broken handoff.
- Evidence
- Reconciled form submissions, leads and conversions.
- Typical risk
- Medium - it changes what visitors see, so it waits for your approval.
Accessibility
Building- Looks for
- Missing alt text, weak contrast, unlabelled form fields and headings out of order.
- Proposes
- Alt text, labels, heading fixes and contrast changes.
- Evidence
- Automated checks on each page, listed by page.
- Typical risk
- Low for alt text and labels; design changes wait for your approval.
What no playbook does
None of them publish opinions as facts, change pricing or legal text, or touch anything outside the access you gave them. When a check fails, it stops and asks.