Skip to main content

Advanced SEO tools

The free audit reads your project's source code. The four paid tabs in Project settings → SEO & GEO — Site crawl, Keywords, Competitors and Rankings — do something different: they look at your published site and at the wider web.

That means three requirements:

  • A paid plan.
  • A published project — there has to be a live address to look at.
  • AI credits. Every run buys data from a search-data provider, so every run is charged — see What this costs.

On a free workspace, the four tabs show a lock and the tab bar carries a Paid plan link. A tab that can't run says why in place of its controls — Publish this project first — off-page checks resolve against its domain. on a project that isn't published, or Search-data checks are not configured on this server. when the search-data service isn't set up. The Keywords, Competitors and Rankings tabs also say it before you press anything: their header line ends Each scan uses AI credits.

Which domain is being analysed?

Every result on the Keywords, Competitors and Rankings tabs names the domain it analysed and when. On a freshly published Berrybell address, search engines have no history for it at all, so keyword, competitor and ranking results will be thin or empty. That's expected, not a failure — connect a custom domain before reading much into these numbers.

Site crawl​

On the Site crawl tab, press Crawl site (after the first crawl, the button reads Re-crawl). It crawls your deployed site page by page and reports defects it finds in the live HTML — the things a source audit can't see because they only appear once the page is rendered and served.

Findings are folded onto the same issue types as the source audit, so one defect found two ways shows up once rather than twice.

Worth knowing before you run one:

  • One crawl per project at a time.
  • A cooldown between crawls — six hours by default, counted from when the last crawl started, however it ended.
  • A page ceiling — 50 pages by default.
  • It's charged when it finishes, for the pages it visited. A site with fewer pages than the ceiling pays for its own pages, not for the ceiling.
  • A crawl only completes while the tab is open. The Site crawl tab checks on a running crawl every ten seconds, and that check is what collects the result. You can leave while it runs, but the tab has to be open at some point after the crawl has finished. If that hasn't happened within three hours of the start (by default), the crawl is marked failed: Crawl abandoned — the provider never reported it finished. Start a new crawl. A failed crawl isn't charged, but the cooldown still counts from when it started.

While a crawl you started is running, the button is disabled and the tab shows Crawling …. If you open the tab while a crawl is already running or the cooldown hasn't passed, the button is disabled and the tab says why: A crawl is already running for this project. or This project was crawled recently. Try again in … with the time left. Right after a crawl you watched finish, though, Re-crawl is clickable again; pressing it before the cooldown has passed shows that same crawled recently message instead of starting a crawl, and nothing is charged. A crawl is also refused when your balance is below the minimum to start, or when checks already running have set aside what's left — see What this costs. Crawls don't count toward the limit on checks a workspace can run at once.

Because of the page ceiling, an empty result means "no issues on the pages it visited" — never "your site is clean".

Keyword research​

On the Keywords tab, press Find keywords to get keyword opportunities for your published domain. The table shows search volume, organic ranking difficulty, and the search intent behind each term.

Nothing runs when the tab loads — you press the button, so you're never charged for a run you didn't ask for. The last result stays on the tab until you run it again, and a run that comes back empty replaces it — see Results stay until you scan again.

Competitor gap​

On the Competitors tab, press Analyze to find competitor domains and the keywords they rank for that you don't.

Rank tracking​

On the Rankings tab, add the keywords you want to track in the Add keywords, comma separated box — up to 100 at a time — and remove one with the × beside it. Check now stays disabled until at least one keyword is tracked; press it to get the current position of each tracked keyword. The last check stays on the tab until you press Check now again. The Studio shows the latest check only, not a history of past positions.

Checks run against a fixed default target — United States, English, desktop — set on the server rather than chosen per keyword.

Positions are read from the top 10 results by default. A keyword that ranks lower than that, or not at all, shows not in top 10.

A check covers at most 25 keywords by default — the 25 most recently added — and a long batch can also run out of time before it reaches them all. A keyword whose lookup fails, or that's refused (for example one containing a search operator such as site:), is left out as well. Whatever the cause, the tab tells you how many were actually checked against how many are tracked (Checked 25 of 40), so a short result never quietly looks like a complete one. The line says the rest ran out of time whatever the actual reason.

Each keyword is a separate lookup at the provider, so a check costs more the more keywords it covers. A keyword the check skipped was never looked up, and isn't charged.

Google AI Overview citations​

When you check rankings, each row also reports what happened with Google's AI Overview for that query — whether one was served, and whether your domain was cited in it. You'll see one of four answers:

AnswerMeaning
none shownGoogle served no AI Overview for that query.
not resolvedWe couldn't determine it this time.
citedAn AI Overview was served and your domain was cited in it.
not citedAn AI Overview was served and your domain wasn't in it.

The distinction between "none shown" and "not resolved" is deliberate: an answer we couldn't get is never displayed as a miss.

Results stay until you scan again​

The Keywords, Competitors and Rankings tabs each keep their last result. It's saved with the project, so it's still there after a reload, on another device and for your teammates, and viewing it costs nothing. Each tab keeps one saved result, replaced by the next scan on that tab that returns one. The Site crawl tab likewise opens on the project's latest crawl.

Each result names the domain it analysed and when — for example Analysed example.com on Sep 13, 2026, 10:42 AM. A kept result can be days old, or be about your Berrybell address from before you connected a custom domain, so check that line before acting on it.

On the Competitors and Rankings tabs, a scan that couldn't produce a result isn't saved, so it doesn't overwrite the saved one. That covers a comparison that couldn't reach the search-data provider or couldn't read a domain from the project's address, and a ranking check that came back with nothing: you see the notice at the time, and the last real result is what the tab shows after a reload.

The Keywords tab can't tell a failed lookup from one that found nothing, so an empty keyword run does replace the saved result and shows No keyword ideas returned for this domain.

A scan still running when you switch to another tab carries on, and shows as running when you come back.

What this costs​

Every Advanced check buys data from a search-data provider, so every run uses AI credits — the same balance a build draws on. It's priced the way a build is: what the provider charged for that run, plus Berrybell's margin, converted into credits. There's no flat price per button:

  • Keyword research and competitor gap cost what the data they return costs. A run is usually a fraction of a credit.
  • Rank tracking is charged per keyword checked. Looking up Google's AI Overview costs extra only on a query where Google loaded one separately; on every other query that extra is refunded, and you aren't charged for it.
  • A site crawl is charged when it finishes, for the pages it actually visited — not for the page ceiling. A crawl that fails isn't charged.

Running the free on-page audit and the GEO checks never uses credits, and neither does coming back to a result you already have. Drafting a fix from either tab does — see the free audit.

To start, a check or crawl needs a minimum in your balance — one AI credit by default. That's a gate, not the price. Below it, the check is refused before it starts with Not enough AI credits for this scan., and nothing is charged — top up and run it again. If auto-reload is on, the refusal also triggers it. Checks already running can hold part of your balance too — see Several checks at once. None of this draws on your hosting budget, so running low here can never pause a published app.

Charges appear under SEO in the Usage tab of Plans & usage, alongside AI-written SEO fixes. The crawl cooldown and page ceiling above still apply: they are limits, not the price.

Several checks at once​

A workspace can run at most three Advanced checks at a time by default, across all its projects. A fourth is refused with This workspace already has 3 SEO scans running. Try again when one finishes. Site crawls don't count toward the three.

While a keyword, competitor or ranking check runs, it sets aside one AI credit until its charge is recorded. That isn't a charge, and it isn't taken off your Remaining balance — but a new check or crawl starts only if your balance, less what running checks have set aside, still covers the one-credit minimum. Otherwise it's refused with Not enough AI credits to run this scan alongside the ones already running. Wait for one to finish, or top up to run more at once. That refusal doesn't trigger auto-reload: your balance covers the minimum, and it's only held by checks already under way.

A check that's interrupted before it finishes releases its place and what it set aside within five minutes. A check that has already run is never charged more than your balance holds.

Next steps​