Querywright documentation
How to operate Querywright: connect the AI you already use, then research, decide, write, check and watch the results, with the App showing all of it.
Querywright is the SEO operating layer for Claude and ChatGPT. Your AI does the reasoning and the writing. Querywright keeps the evidence, the plan, the work and the results, so the AI can pick up where it left off, and the App lets you see and manage all of it. The decisions that matter stay yours.
- 01UnderstandYour AI
- 02DecideYou
- 03CreateYour AI, then you
- 04CheckQuerywright
- 05ObserveQuerywright
- 06ReconsiderYour AI, then you
Start with how Querywright works with your AI, then get set up. Looking for a prompt to try? See prompts you can copy.
Getting started
Create your account, connect Claude or ChatGPT, and run your first scan.
- How it works with your AI The loop Querywright runs with Claude or ChatGPT: who records, who reasons and who decides at each step, and where each step shows up in the App.
- Getting started Create your Querywright account, choose your plan, add your first site and connect Claude or ChatGPT, then copy the starter instruction for your first scan.
- Connect Claude Add Querywright to Claude as a custom connector, sign in and authorize it, confirm it under Settings → Connected AI, and turn it on in a conversation.
- Connect ChatGPT Add Querywright to ChatGPT as a developer-mode app with OAuth, sign in and authorize it, then confirm the connection under Settings → Connected AI.
- Your first site scan Send the starter instruction, let your AI scan your site, say yes or no before it measures search demand, and find the saved scan and pages in the App.
- Understand the App Find your way around the Querywright App: the navigation rail, the six site sections, Search, Refresh, the Acquire menu, shortcuts and the phone layout.
Core concepts
Evidence and your AI’s judgment, the decisions that are yours, and what Querywright never does on its own.
- Evidence vs AI judgment What Querywright observed, measured and checked, what your AI reasoned, and what you approved. How the App keeps them apart, and why a sequence is not a cause.
- Your decisions Every decision only you make in Querywright, from approving a plan to enrolling rank tracking, and what counts as saying yes to your AI.
- Measured zero vs no data Why a measured zero, no data, not checked and a failed check are different answers, and how the same rule covers Search Console and Analytics.
- What it does not do What Querywright leaves to you and your AI: no publishing, no Business Profile management, no outreach, no attribution, no approvals on its own and no model.
Research
Scans, search demand, keyword research, Google’s results, competitors and links.
- Site and page scans What a scan of your whole site or of one page records, how coverage is reported, when a page is read in a browser, and where saved scans show in the App.
- Search demand How Querywright measures how often a search is made, in a named location and language, what each demand label means, and where measurements show in the App.
- Keyword research How keyword research finds and measures new searches around a topic, from your AI or the App, what the results are and are not, and which allowances it uses.
- Google results How Querywright observes Google's top 50 organic results on desktop and mobile in a named location and language, what it records and what it never claims.
- Competitors Which domains appear in Google's results for your plan's searches, what Querywright records about competitor pages, and why those facts are never endorsements.
- Links Who links to your site and to competitors, from dated link snapshots, the link sources you choose to pursue, and the Site Intelligence units snapshots use.
Work
Suggested plans, approval, writing briefs, approved copy, checks and the live comparison.
- Suggested plans and approval What a suggested plan holds, how exact-wording conflicts and Not pursuing work, and how your explicit approval through your AI makes it the plan in effect.
- Writing briefs What a writing brief is built from, how it differs for an existing page and a new page, and when Querywright issues it again with newer page text.
- Approved copy What accepting copy covers: the exact text you saw, saved with your AI's findings, why a later edit needs its own acceptance, and Choose which version to keep.
- Checks How Querywright checks approved copy before you publish: what Ready to publish and Needs fixes mean, check items, statements you confirm, and decision rechecks.
- Live comparison After you publish, a new scan and a live comparison show whether your approved copy is on the live page, what Approved copy is live means, and its limits.
Performance
Rankings, Search Console, Google Analytics, Local SEO and page experience.
- Rankings Weekly Google desktop and mobile checks of the pages in your plan, top 50, in the market the demand was measured in: check states, page ownership and baselines.
- Search Console Reading the finalized clicks, impressions, CTR and position imported from Google Search Console, partial coverage, missing rows and the Google index line.
- Google Analytics What organic visitors did after the click: sessions, engagement, key events with the meaning you give them, revenue and landing pages, read-only from GA4.
- Local SEO Local pack observations for an exact search, location and device, which listing is yours, name, address and phone by source, and listing ratings as reported.
- Page experience Core Web Vitals field data at p75 for real visitors and Lighthouse lab runs, kept apart and as Google returned them, and the Page experience reads allowance.
App guide
Every screen in the App, region by region.
- Overview Each panel on a site's Overview in screen order: what needs your decision, site status, pages by stage, search and traffic, and this month's records.
- Pages and page detail The Pages table and map, every page work state with its next step, and a page's detail: the journey strip, its four tabs and its exact records.
- Work The Work section's five tabs: Decisions, Plan in effect, Suggested plans, All plans and Not pursuing, and what each shows about your plans.
- Performance The Performance section's six tabs: Overview, Rankings, Search Console, Analytics, Local and Page experience, and the About these numbers note.
- Evidence and Acquire The Evidence section's six tabs, from Scans to Google index, and the Acquire menu: what each action gathers and which ones use your included allowance.
Working with your AI
What to ask, when your AI asks first, and prompts you can copy.
- Working with your AI What to ask Claude or ChatGPT, what Querywright does on its own, what is saved, how to pick up where you left off, and how your AI labels each claim.
- When your AI asks first Which requests use an included allowance and what your AI asks before them, what uses none, and the decisions your AI never assumes on your behalf.
- Prompts you can copy Prompts for each common job in Claude or ChatGPT, from the first scan to rankings, with who does what and whether your AI will ask before using allowance.
Integrations
Connect Google Search Console and Google Analytics.
- Connect Search Console Connect read-only Google Search Console access, choose the Domain or URL-prefix property for each site, and follow the first import of 90 days of history.
- Connect Google Analytics Connect read-only Google Analytics 4 access, choose the GA4 property that measures each site, and see how each day imports once it is 72 hours old.
Account
Your subscription, allowances, sites and AI connections.
- Your subscription The Founding Beta terms, starting and canceling through Stripe, what happens when a payment fails, and how deleting your account differs from canceling.
- Allowances Every allowance included in your plan each billing period, what uses one, what uses none, what happens when one runs out, and how held allowances work.
- Sites Add, switch, remove and restore sites in Querywright, and how a site uses one of the ten site slots the first time a scan of it succeeds in a billing period.
- AI connections See which Claude or ChatGPT connections can use your Querywright account, connect another one, and disconnect any connection you no longer use.
Troubleshooting
What a message means and what to do about it.
- Connector problems Fix problems connecting Claude or ChatGPT to Querywright: expired connection links, missing or out-of-date tools, setup screens, and old Account wording.
- Google connections Fix Google connection problems for Search Console and Google Analytics: unfinished Google screens, missing properties, reconnecting and lost property access.
- Imports and backfills What each Search Console and Google Analytics import state means, why a period shows partial coverage or blank totals, and when you need to act on an import.
- Common messages The messages the App shows and the ones your AI passes on from Querywright, what each one means, and what to do, grouped by sign-in, billing, sites and more.
Reference
Every label the App shows, in one place.
Two things worth knowing first
Querywright does not include a model. You bring Claude or ChatGPT, connect it once, and work in that conversation. The App is where you see the evidence, every page’s state, your plans, the copy you accepted, the checks and the results.
The decisions are yours. A suggested plan becomes your plan only when you approve it, and a draft becomes approved copy only when you accept that exact version. Querywright records those decisions when you make them, never before. See your decisions, and the commercial terms on the pricing page.