Search Console setup
Clara reads Google Search Console to learn how your pages actually perform: which posts sit on page two, which lost positions, which earn impressions but no clicks. Connect it properly and the dashboard has real data within a day. Skip it and every recommendation she could make waits on it.
Work through this page once per site. The site's Recommendations tab tracks the same list in a card called Get Search Console ready for Clara, and ticks the steps it can verify by itself — so you can start here, or start from the card and follow the How link on any step.
Make sure your site has a sitemap
A sitemap is the list of URLs you are asking Google to read. Clara reads it too: it is how she knows which pages exist before Search Console has any history for them.
Most platforms already publish one. Open it and check:
| Platform | Where it usually is |
|---|---|
| WordPress (Yoast / Rank Math / All in One SEO) | /sitemap_index.xml or /sitemap.xml |
| Shopify | /sitemap.xml |
| Webflow | /sitemap.xml |
| Squarespace | /sitemap.xml |
| Next.js | whatever app/sitemap.ts returns |
If it loads, point Google at it from robots.txt by adding one line:
Sitemap: https://yourdomain.com/sitemap.xml
If it does not exist, turn on the sitemap feature in your SEO plugin, or ask whoever built the site — a sitemap is the one file a search engine always looks for first.
When you add a site, PenGraph reads robots.txt and the usual paths and stores what it finds. If your setup card says no sitemap was found, that is the honest answer from a live scan — fix the file, then press Check sitemaps on the site's Visible everywhere → Google tab.
Verify your domain in Search Console
Google will not report anything about a site it does not consider yours. Verifying is how you prove the domain is yours — normally by copying a code Google hands you into your DNS, which only the domain's real owner can do.
- Open search.google.com/search-console and sign in with the Google account you want Clara to read through. Use the account that will own this site long-term: access follows the account, not the browser you happen to be sitting in.
- Click the property picker in the top-left (the dropdown beside the Search Console logo) and choose + Add property.
- Two boxes appear side by side. Pick the Domain box on the left; the right one is URL prefix. Add it as a Domain property explains the difference and why the left box is nearly always the right one.
- Type your naked domain:
yourdomain.com. Nohttps://, nowww., no trailing slash, no page path — Google rejects anything else in this box. - Click Continue. Google shows a TXT record starting
google-site-verification=. Copy the whole string, that prefix included. - Add that record at your DNS host — the menu paths are in Where the TXT record goes below. This is DNS, not your website admin panel.
- Keep the value exactly as Google gave it: not quoted, not shortened, and not pasted onto an existing TXT record. A second TXT record is correct; one long edited record is not.
- Back in Search Console, click Verify. DNS can take a few minutes, and up to 48 hours in the worst case, so if it fails, wait and click Verify again rather than starting over.
- Leave the record in place. Removing it later can un-verify the property and cut Clara's data off.
Verification is one-time per property, and the same record covers that domain for every Google product. You keep full control: verifying ownership gives Google permission to report data, and nobody else gains access.
Where the TXT record goes
| DNS host | Menu path | Host / name |
|---|---|---|
| Cloudflare | DNS → Records → Add record | @ |
| GoDaddy | My Products → your domain → DNS → Add New Record | @ |
| Namecheap | Domain List → Manage → Advanced DNS → Add New Record | @ |
| Amazon Route 53 | Hosted zone → Create record (AWS adds the quotes around a TXT value for you) | leave the name blank for the root |
| Vercel | Project → Settings → Domains → DNS Records → Add | @ |
| Squarespace (including former Google Domains) | Domain → DNS → Custom records → Add record | @ |
The record type is always TXT, the value is always the string Google showed you, and the TTL can stay on its default.
Once you connect, the card's Your domain is verified in Search Console step closes off Google's own property list — you never tick that one by hand.
Add it as a Domain property
What this step is really about: telling Search Console that the whole domain is one property, so a single report holds http, https, www and non-www together. Clara reads one property per site, so a narrower property leaves part of your traffic in a report she never opens.
There are two kinds of property, and choosing the wrong one quietly hides half your data:
| Property type | Covers | Use it when |
|---|---|---|
| Domain | yourdomain.com, www.yourdomain.com, http://, https:// and every sub-domain, in one report | Almost always — this is the one to pick |
| URL prefix | Only the exact prefix you type, e.g. https://www.yourdomain.com/ | You only ever care about one sub-domain |
So how do I actually add one?
Adding a Domain property is not a separate job in front of the steps above — it is step 3. Both types come out of the same + Add property button; only the box, and the text you put in it, differ:
- URL prefix box: the full address with its scheme, e.g.
https://www.yourdomain.com/. It then reports on that address alone — your site also served athttps://yourdomain.comis a property you do not have. - Domain box: the bare name,
yourdomain.com. It then reportshttp,https,wwwand every sub-domain of it as one — one row per site, which is what Clara reads.
So the click path is + Add property → Domain → yourdomain.com → Continue → copy the TXT record → Verify, and the record goes at your DNS host, not in your website admin panel. Verify your domain in Search Console walks through each of those clicks and the DNS side.
Which type did I actually create?
Open the property picker in the top-left again. A Domain property is listed as the bare name — yourdomain.com, no scheme, no www. A URL-prefix property carries its scheme — https://www.yourdomain.com/. That listing is exactly what this card's tick asks you to confirm, and it is the one step nothing can check for you: Google's API lists your properties but never their type, so Clara cannot verify it the way she verifies the others.
I already added a URL-prefix property
Nothing you did is wasted, and you do not have to start over:
- Add the Domain property too — + Add property → Domain, naked domain, Continue. Google usually verifies it on the spot, because the same account already proves control of that domain; if it hands you a TXT record instead, add that and click Verify as above.
- Check which one Clara matched. She picks a property from the connected account by hostname and uses an exact URL-prefix property before falling back to the Domain property, so a leftover
https://www.yourdomain.com/can be the one she reads — and that property only ever covered that one host. The card's Your domain is verified in Search Console step names the property she matched, so that is where to look. - If it names the narrower one, either delete the URL-prefix property in Search Console so only the Domain property can match — a fresh property starts its own history, which is the one blunt edge here (see Pending is normal below) — then sync again; or keep both and change the site's Website address in Settings to the host you actually want read, then sync.
Connect Search Console to Clara
- Open the site in the dashboard and go to Performance.
- Click Connect Search Console.
- Choose the Google account that owns the property — the same one you verified with. A different account will connect and then find no data.
- Read the permission screen and click Allow. You come straight back to the dashboard.
What you are granting: Google's Search Console scope, which lets Clara read your performance data and add your sitemaps to the property. She never edits your settings, your pages, or anything on your website. You can revoke access from your Google account permissions at any time, and the dashboard will tell you when a connection needs re-authorising.
The button says Reconnect instead of Connect, and it runs the same flow — approve it again. A connection made before PenGraph started adding sitemaps still reads your data perfectly well but cannot hand Google a sitemap, and Google never widens a permission that was already granted, so reconnecting once is the only fix.
Sitemaps are added for you
You do not have to paste a sitemap URL into Search Console. When a site is added, PenGraph scans and stores its sitemaps; the moment the connection above completes, they are submitted to the property automatically, and the daily loop keeps Google's copy of them current.
Submitting is not uploading. Google records the URL and fetches the file on its own schedule, so a freshly added sitemap shows as Pending until Google gets to it. That is normal, and the setup card counts the step as done as soon as Google holds it — you are not waiting on Google's crawl queue to finish setting up.
If a sitemap is refused, the card says so in plain words. The usual reason is that the sitemap belongs to a different property than the one you connected — for example a www URL submitted to a www-only property while your sitemap lists the naked domain. Fix the property, not the sitemap.
Fix sitemap errors
Accepting a sitemap and being able to read it are two different things. Google takes the file, tries to download it, and reports back how many of its URLs it could not process — as errors (Google could not use the sitemap) or warnings (it worked, but something in it needs a look). That report is the same thing the notification bell shows as "Sitemap has errors", and it is the one Search Console message PenGraph can read directly: the Search Console API exposes the sitemap's error and warning counts.
When Google reports a problem, the setup card adds a step called Search Console reports no sitemap problems and keeps it open until the next daily check comes back clean. To see exactly which URLs Google is unhappy about:
- Open Search Console and pick the property.
- In the left menu, open Sitemaps.
- Click the sitemap PenGraph submitted and read the Errors / Warnings rows — Google lists the offending URLs underneath.
The usual causes, in order of frequency:
| What Google says | What it means | Fix |
|---|---|---|
| URL not allowed / blocked by robots.txt | A page in the sitemap is disallowed for Googlebot | Loosen the robots.txt rule covering it, or remove the URL from the sitemap |
| 404 / not found | A URL in the sitemap no longer exists | Redirect it, restore it, or drop it from the sitemap |
| Couldn't fetch / could not be read | Google could not download the sitemap file itself | Confirm the sitemap URL loads in a browser, then re-submit |
| Wrong host / not on this property | The sitemap lists URLs on a different host than the property | Use a Domain property covering www and non-www |
Once the sitemap is fixed, the step closes itself on the next daily check — you do not tick it by hand.
Indexing alerts for pages Google will not index
This is the closest PenGraph gets to reading Search Console's notification bell. Google publishes no API for the messages themselves — manual actions and security issues can never be read this way, and no amount of integration changes that. What is available is the URL Inspection API, which reports, per page, the same coverage state the "Page indexing" notifications are built from.
Every day PenGraph asks Google about the site's published pages and turns the answers into Indexing alert cards. A card appears when Google is genuinely refusing to index a page — a stray noindex, a robots.txt block, a soft 404, a broken redirect, a server error, a 404, a canonical Google overrode, or a page it crawled or discovered but chose not to index. Each card names the page, quotes Google's own words for it, and says what to change.
You do not have to wait for the daily run. The site's Performance page has a Check indexing button that asks Google right away, stores the answers, and files any new cards on the spot — useful right after you have fixed something and want to confirm it is clean.
Two things keep it honest:
- It is an allow-list, not "anything that is not indexed". Benign states — a page Google treats as an alternative to its canonical, a redirect, a URL it simply has not crawled yet — do not raise a card. A notification you cannot act on is noise.
- It never claims to see manual actions or security issues. If your traffic collapses site-wide, check those two reports yourself at Search Console → Security & Manual actions — they are the one part of the bell nothing here can reach.
Fix the page on your site, mark the card done, and it reappears only if a later inspection still finds the problem.
Keep your sitemap complete
A page nothing links to and no sitemap mentions is invisible: Google has no way to find it. Clara can see your sitemap, but only you know what belongs in it.
- Anything you want ranked should appear in the sitemap.
- Tag pages, internal search results and thin archives are usually better left out.
- Publish a new section (a blog, a new product line) and its sitemap entry rarely writes itself — check that it did.
The setup card asks you to confirm this yourself, and marks the step as your word rather than a verified fact.
Check nothing important is noindexed
noindex tells Google to keep a page out of search results entirely. It is the single most common reason a perfectly good page never ranks — and it is usually left over from a staging site, a "coming soon" launch, or an SEO plugin's default settings.
- View the page's source and search for
noindexin therobotsmeta tag:content="noindex". - Also check the HTTP header
X-Robots-Tag, which some hosts add without any trace in the HTML. - Older WordPress installs love to keep "Discourage search engines from indexing this site" switched on. Settings → Reading turns it off.
PenGraph does not crawl every page, so this is another step the card leaves to you.
Pending is normal
After you connect, a few things take a day or two and none of them are broken:
| What you see | What it means |
|---|---|
| Sitemap shows Pending | Google has it and has not downloaded it yet |
| Performance charts still empty | The first sync has not run, or the property has less than a day of history |
| Recommendations tab looks quiet | Quick wins need position 5–20 data; Clara finds them after the first sync with enough impressions |
| A page shows no impressions for its target keyword | Normal for new content — give it a few weeks before judging it |
If Search Console itself has never reported the site (Search Console → Overview shows zero), nothing downstream can fill in, and the fix is always upstream: property, verification, connection, in that order.
Troubleshooting
| Symptom | Cause | Fix |
|---|---|---|
| Card says no sitemap was found | The live scan saw nothing at robots.txt or the usual paths | Create the sitemap, then Check sitemaps on the Visible everywhere → Google tab |
| Connected, but the site has no data | Signed in with a Google account that does not own the property, or verified a different domain/host | Reconnect with the owning account; check the property in Search Console |
| Google refused the sitemap | It is not on the property you connected | Use a Domain property covering both www and non-www, then resubmit |
| "Reconnect required" | The token was revoked, or it predates sitemap submission | Click Connect Search Console again and approve |
| Numbers stopped moving | Google reports a few days behind; a dead connection is flagged in the dashboard | Check the site's Visible everywhere → Google tab for a connection warning |
Related: Clara (Autopilot) covers what Clara does with the data once it arrives.