SEO reporting with a plan for what comes next

Import a Keyword Planner CSV or TSV file, add manual research, or bring in queries from a connected Search Console property. Source, retrieval date, geography, language and period stay with the research. Unknown values stay unknown, an actual zero stays zero, and Planner bands are kept apart from Search Console metrics. You can assign reviewed related-keyword suggestions to existing or proposed pages in a batch, and you choose each destination before applying. Planning works before Search Console returns any data. Direct Keyword Planner API research is not included.

Prepare content from your saved business profile, page brief, sourced facts and keyword evidence. AI connections are private to your signed-in account, with model and thinking settings in Workspace integrations. WeaverData has no separate daily draft quota; provider limits, model access and billing rules still apply. Use a Claude Code subscription through the paired Windows companion on an awake computer. Its allowance is shared with normal Claude usage, and usage credits must stay off. OpenAI and Anthropic API keys are separately billed alternatives, with no automatic fallback. ChatGPT subscription connection is not available yet. Raw customer inquiries and contact details are excluded. Autopilot analysis can use aggregate inquiry counts alongside website and search evidence.

An admin or owner can enable an Autopilot policy for a selected website, choosing the AI connection, goal, supported content sections, batch size and review cadence. Automatic mode approves validated content under that policy and records its run and revision. Review mode waits for approval of the prepared revisions. The manual workflow remains available: generate or write copy, preview it, save it and approve the exact revision before publication. Manual edits invalidate conflicting AI work. Editing approved copy clears that approval while the existing live snapshot and publication history remain available. Autopilot never executes website code changes. It supplies a proposal and handoff prompt for a chat model with access to the relevant website code integration.

The connected v2 Next.js managed publisher supports single-slug pages under /features, /blog and /guides on WeaverData. Older v1 adapters remain feature-only. Publication activates the exact approved snapshot and then checks the public HTTPS page, including its title, canonical, heading, indexing directives and structured data. Later draft edits leave live content unchanged until another approved revision is published. An external website needs a developer to install a supported adapter; a tracking connection alone does not grant publishing access. The reviewed Next.js export handoff remains available. General CMS publishing and arbitrary website code changes are separate integrations. A successful live check confirms observed HTML. Google crawling and indexing are separate.

Results connect page-level search evidence with qualified organic inquiries when tracking, landing-page and CRM evidence match. Missing rows remain unknown, and incomplete coverage is labeled. The analyst receives aggregate recorded-source inquiry counts separately; those totals do not establish page attribution or causation. Autopilot schedules an observation check and a later results review. The default cadence is three days and seven days, while newly published pages normally wait at least 28 days before automatic rewriting. A check can wait without another AI request when there is no meaningful new evidence. A justified next action becomes another recorded run under the current policy. Manual threshold-based recommendations still offer review tasks. You can inspect evidence, edit pages, pause or cancel pending work, and handle code proposals yourself. Scheduled Claude work waits if its paired companion is offline.