Three data sources. No join. No map.
Every marketer managing a website operates across three disconnected systems. When their signals drift apart, a page gaining impressions while it bleeds sessions, the leak stays invisible until it shows up in quarterly revenue.
Search Console
How pages appear in search: impressions, clicks, position.
Can't see: What users do after they click.
GA4
What users do on the site: sessions, flow, conversions.
Can't see: Why they entered from those URLs in the first place.
Crawl tools
What each page actually contains: links, structure, content.
Can't see: Which findings have any traffic impact at all.
The tools that exist sit on either side of a cliff.
under $600 / year
Desktop crawlers
Screaming Frog, Sitebulb. They can pull GA and Search Console figures in as per-URL columns, but not correlate a journey or a session path, and their AI integration exports media, not a fix. The analysis is still your job.
Pricing on request
Enterprise platforms
Botify, Lumar. They correlate all three sources, and they need a data engineering team to configure.
Small to mid size businesses and marketing agencies have no tool in between. NavSignal exists to fill that gap: join Search Console, GA4, and keyword data at the URL level, render the compound signal as a navigable journey map, and generate the fix, not just the finding.
Three things changed that made this buildable.
01
GA4 BigQuery Export got exact
In July 2024 Google added fields that allow exact reconstruction of the page-to-page sequence within a session. Before that, journey-map edges were an approximation.
02
LLM costs collapsed
An AI action brief across a full audit now costs pennies instead of dollars, which turns per-URL briefs from a research experiment into a product feature.
03
Universal Analytics sunset
The 2023 cutover left millions of businesses sitting on GA4 data they still do not fully understand, and no tool that interprets it for them.
What the agent will and will not do.
NavSignal's agent can open a pull request against your site's repository. That is the point of it, and it is also why its limits are worth stating plainly.
- ReadsSearch Console and GA4 through read-only OAuth scopes, keyword data, and a crawl of your own site.
- DecidesWhich pages are leaking, which pattern explains each one, and what change would close it. Every brief carries a confidence rating.
- ProposesA pull request. It never merges, never force-pushes, and never touches your live site directly.
- CannotWrite anywhere except your workspace’s own connected repository. The target allowlist is enforced in code before any network call, and an environment override can only fail that check, never satisfy it.
- ReversiblePreflight checks run before any write, and a deployment can be reverted from the history panel.
Built on real data, against a live site.
The pipeline is real. It was built and run against a live home-services business, with permission, correlating that site's own Search Console, GA4, and keyword data. The demo canvas walks a fictional sample store so the tour stays stable and no client data is exposed.