Audits

Audits are one-off reviews of your website. The AI reads your site (or, for AI search, the web), lists problems and opportunities by severity, and suggests a fix for each. You can add any finding to your roadmap with one click.

Audits are in two places:

Audits	Where
Landing page and Onboarding	The Roadmap page, in the Audits card below the board
SEO and AI search	SEO & GEO → Opportunities, in the AI audits card below the list

The SEO and AI search audits used to be on the Roadmap too. They moved to Opportunities, next to the ranked fixes from Site health, Search and AI visibility. They work the same in both places, and the Roadmap’s re-plan still reads all four.

Before you start

Audits need your product’s website. Set it in Settings → Products with the pencil next to the product (the Website field). Without it, an audit fails with “Add the product’s website in Settings first”.

Audits also use your product intel (brief, personas, notes) for context, so build intel first for better results.

The four audits

Audit	Label in the app	What it checks
Landing page	Landing page	Promise, proof and call to action
Onboarding	Onboarding	The path from first visit to first meaningful action
SEO	SEO	Technical basics and missing pages
GEO	AI search	Whether AI assistants and search recommend you

Landing page

The AI reviews your homepage and key pages as a landing page for your audience. It checks:

• whether the promise is clear in five seconds, in the audience’s words;
• whether there is one obvious call to action;
• proof: examples, numbers, testimonials;
• whether objections are answered (price, privacy, “is anyone using this?”);
• clutter and competing links;
• mobile issues visible in the page text.

Onboarding

The AI reviews the path from first visit to the first meaningful action. It can’t sign up, so it judges from your pages, FAQ, help and docs: signup friction, what a new user sees first, whether the first action is suggested, invite-only or waitlist confusion, and what would make someone come back.

SEO

Before the AI looks, Startica measures some facts directly from your homepage:

Fact	What’s recorded
Homepage status	The HTTP status code
Title	The text and its length
Meta description	The text and its length
H1	How many, and the first three
Canonical	The canonical link, or missing
Open Graph	Whether title, description and image are set
robots.txt	Present or not, with up to five Disallow or Sitemap lines
sitemap.xml	Present or not, and how many URLs it lists

The AI combines these with your page text to look for indexing problems, weak titles and descriptions, thin pages, missing pages that answer your audience’s searches (comparisons, how-tos, “best X for Y”), and internal linking.

Only /sitemap.xml at your site’s root is checked. A sitemap at another path is reported as missing.

AI search (GEO)

GEO (generative engine optimization) is about showing up in AI assistants’ answers. This audit doesn’t crawl your site. Instead, the AI searches the web the way your audience would, with several phrasings of their questions and of “best X for Y”. It notes which products come up, whether yours appears and how it’s described, and which sources (lists, forums, reviews) drive those answers. It then reviews those notes and suggests what content or mentions would get your product included.

For Fieldnote, that means searches like “best note-taking app for field research” or “offline notes app for ecologists”.

How an audit runs

Click Run next to an audit (or Run again if it has run before), on the Roadmap or on Opportunities. The button shows Running… while it works. Each audit takes a minute or two.

Step	Landing page, Onboarding, SEO	AI search
1. Gather	Crawl up to 8 pages: your homepage, then the most relevant links from it. Respects robots.txt.	Web research with up to 8 searches and 4 page fetches, citing URLs.
2. Measure	SEO only: the facts listed above.	—
3. Review	The AI writes a summary and 5–12 findings.	Same.

Crawled text is capped at about 50,000 characters. Pages that time out or return errors are skipped. If nothing can be crawled, the AI is told so and says what it couldn’t check.

The review uses the strategy model task (Claude Opus or GPT-6.1 Sol by default). The AI search research uses the research task (the same models by default) with the provider’s web search. See AI models.

Audits run once per click. They aren’t retried automatically; if one fails, the error shows on the card and you can run it again.

Findings

When an audit finishes, its section shows the date, a two- or three-sentence summary with the single most important fix, and a collapsible list of findings, high severity first.

Severity	Meaning
high	Likely costing you signups now
medium	A clear improvement
low	Polish

Each finding has a title, a detail (what the AI saw, often quoting the page) and a Fix.

Only the latest run of each audit is shown. The roadmap re-plan reads the high-severity findings from your latest audits.

Add a finding to the roadmap

Click Roadmap on a finding. Startica adds it as your own roadmap item, so re-plans won’t change or drop it:

Roadmap field	Comes from
Title	The finding’s title
Why	The finding’s detail
Next step	The finding’s fix
Stage	Activation for Onboarding; Acquisition for the others
Priority	P1 for high, P3 for medium, P4 for low

The success metric is left empty. Edit the item to add one. See Roadmap.

Costs

The app describes each audit as costing well under $1. The AI search audit costs a little more than the others because of the web searches. Actual cost depends on your site and the model you chose; check System for real spend. See Costs.