Set up a website for SEO
The setup chain in order — connect Google, run the setup check, crawl the site, fill the brief, create a research target and set a budget.
Setup is a chain, not a checklist you can do out of order. Each link feeds the next: Google supplies the performance data, the crawl lets the AI read the site, the brief says what the business sells, the research target says which market to measure, and the budget caps what paid research may spend each month.
Work top to bottom once and the rest of the Command Center stops arguing with you.
Before you start
You need the website's URL on its Builder project, a Google account with access to the site's Search Console property and Google Analytics 4 property, and owner or admin rights to set the monthly budget.
Run the setup
Connect Google
Connect Search Console and Google Analytics 4, and select the properties, in Analytics settings. The Command Center only reports their state — it is not where you pick them.
Run the setup check
Select the website, open the kebab menu (More website actions) in the workspace header and click Setup check. It audits everything the Command Center needs and refreshes the readiness signals. The tooltip says No-spend setup check — it never charges research budget.
Read connection readiness top to bottom
SettingsData & trackingConnection readiness. Work down the list until every dot is green, using each row's own button.
Crawl the website
Open StrategyResearch, find the Site knowledge lane in the workspace bar and click Crawl website. This is free. It turns the site's pages into knowledge sources the AI can read.
Fill the Research brief
StrategyResearchBrief. At minimum, set Niche and Primary services. Draft from the website fills only the still-empty fields from the crawl, so run the crawl first.
Set the country
SettingsContent & page rules, then set Country. A non-country location code stored here blocks research-target creation, so fix it before the next step.
Create the research target
Open any research surface. The empty state reads Research <host> with
DataForSEO and names the market it will use. Click Create research target
for <host>. Creating it is free.
Set the monthly SEO budget
SettingsAutomation & budget. Enter the most paid research may spend each month. Left blank, the field reads Blank = $15 a month and paid research runs against $15 a month. On a workspace exempt from credit metering the field reads No cap: a blank budget has no monthly cap there, and only Command Center keyword data stops at $15 a month.
Capture the reporting baseline
SettingsData & tracking, then set the Baseline start date and click Capture baseline. Do this against the month you are about to report on.
What a blank monthly budget means
On a metered workspace a blank budget is a $15 cap: it no longer pauses anything, paid research runs against $15 a month, and Settings shows $15 (default). On a workspace exempt from credit metering a blank budget has no monthly cap. Only Command Center keyword data stops at $15, and Settings shows No cap · keyword data $15. Type a number to set your own cap.
Connection readiness reference
Every row has a coloured dot, a plain-English detail line and a fix button.
| Row | Green means | What fixes it |
|---|---|---|
| Search Console | A property is selected and returning data | Pick the property in Analytics settings |
| Google Analytics 4 | A GA4 property is selected and returning data | Pick the property in Analytics settings |
| DataForSEO evidence | The platform research credential is available to this workspace | Nothing you can change — it is platform-provided |
| Google connection | Connected, including the sitemap-submission scope | Reconnect Google and re-consent |
| Research brief | Niche and primary services are set and read as terms | Open Research brief and fix the fields |
| Site knowledge | The crawl read the site and stored a usable profile | Retry site crawl, or add sources by hand |
| Reporting baseline | A baseline date is locked | Capture baseline |
| AI provider keys (Advanced) | The writer and review providers are configured | Platform-level; ask support |
Two rows deserve a longer explanation.
Property does not match this website's host. This is a warning, not a block.
Many sites verify only the www variant, or only the bare domain. Data still
flows. Pick the matching property when you can.
Search Console write access needs attention. The Google connection was made before the sitemap-submission scope existed, so article submission to Search Console will fail. Click Reconnect Google on that row and re-consent.
The AI may not be able to read the site at all
If Site knowledge is amber, either the crawl was truncated or the site is a client-rendered shell with nothing for the crawler to read. A truncation is fixed by Retry site crawl on the Research brief. A client-rendered shell is not — no number of re-crawls will help, and the fix is server-side rendering or adding knowledge sources manually.
The Data menu
Once setup is done, the header's Data button is where you refresh anything by hand. Each lane shows its own Data as of line.
| Setting | What it changes | Default |
|---|---|---|
| Google evidence | Pulls Analytics and Search Console for the evidence window. Free, but uses your Google quota. | Free |
| Historical monthly totals | Re-pulls up to 12 closed months of Search Console and Analytics history and rebuilds monthly totals. Reports you already delivered stay frozen. Owner only. | Free |
| Deep Search Console pull | Replaces the evidence window's Search Console data with one coherent capture at the same widths as the Google evidence refresh — 1,000 queries, 400 pages, 1,000 page-query pairs — plus up to 250 countries. Free, but uses your Google quota. Owner only. | Free |
| Free sources | Site probe, schema audit and sitemap check, plus a Search Console and Analytics re-pull. PageSpeed runs on its own cadence. | Free |
| Search evidence | Paid DataForSEO pull of keyword rankings, search research and authority. Owner only, and it needs a second confirming click. | About $2 |
A lane reports its oldest clock, and says some sources never pulled when part of the lane has no data at all. Never pulled means exactly that. Running one lane temporarily disables the siblings that share its action family.
What good looks like
- Every connection readiness dot is green, or amber only on the host-mismatch warning.
- Site knowledge was crawled in the last 35 days. Older than that and the freshness pill turns into a stale badge.
- The Research brief reads as terms, not sentences, and nothing is flagged as auto-filled.
- A research target exists, with the right country and a sensible monthly ceiling.
- The monthly SEO budget is typed, or left blank on purpose: $15 a month on a metered workspace; on a workspace exempt from credit metering no monthly cap, with only Command Center keyword data stopping at $15.
- A baseline is captured.
Troubleshooting
| Symptom | Cause | Fix |
|---|---|---|
A research tab says Research <host> with DataForSEO | The website has no research target yet | Click Create research target for <host> — it is free |
| Target creation fails with a market error | Settings stores a non-country location code | Set a Country in Settings → Content & page rules first |
| Readiness says the monthly SEO budget is reached | This month's paid research used the budget — a typed budget, or on a metered workspace a blank one, which counts as $15 | Wait for next month, raise the budget in Settings → Automation & budget, or run research yourself as an owner and accept the overrun |
| Readiness says Command Center keyword data reached its $15 default | On a workspace exempt from credit metering a blank budget caps only Command Center keyword data, and this month's $15 is used. Strategy runs and drafting have no monthly SEO budget there | Wait for next month — each keyword-data pull you start asks you to confirm meanwhile — or type a monthly SEO budget in Settings → Automation & budget, which then caps every lane, strategy runs included |
| The Search Console row warns about a host mismatch | The selected property is the other www variant | Pick the matching property; data still flows meanwhile |
| The Google row asks you to reconnect | The connection predates the sitemap-submission scope | Reconnect Google and re-consent |
| Site knowledge is amber after a re-crawl | The site is a client-rendered shell | Add knowledge sources by hand, or render the site server-side |
| Capture baseline will not take a date | The date is in the future | Baselines cannot start in the future — pick today or earlier |