Incuca Intelligence · Site health
Website health check: how to know if your site is healthy
Most teams find out their website got worse through a symptom: the contact form stopped bringing leads, a page disappeared from Google, a customer called to say the site won’t load. Site health exists to flip that order. The check runs from the outside, through the site’s address, the same way a visitor or a search crawler would reach it. Nothing needs to be installed to get started.
What a website health check covers
Each point answers a question a site owner would ask if they knew where to look, and every finding comes with what to do about it:
| Point | The question it answers |
|---|---|
| Mobile speed | Does the page load fast for people arriving on their phone, which is how most people arrive? |
| Desktop speed | And on a computer, where many purchases are completed? |
| Indexing | Can Google index the pages? Are robots.txt, the sitemap, canonical tags, meta tags and structured data right? A page outside the index doesn’t show up in search. |
| Tracking | Are the measurement tools (Google Analytics, Tag Manager, ad pixels, cookie consent) installed? Without them, no other number tells you where a sale came from. |
| User experience | Does browsing and reading the page create friction? Frustrated visitors don’t complain; they leave. |
| Accessibility | Can people who navigate by keyboard, use a screen reader or struggle with low contrast use the site? |
| Security | Is HTTPS enforced, is the certificate valid, and are the security headers in place? |
| AI presence | Can AI assistants read the site and cite it when someone asks about your market? |
The 90-point bar isn’t ours. It’s the Lighthouse convention, Google’s auditing tool, for telling good apart from needs improvement. The screen says so next to the score, so nobody mistakes the number for an opinion.
How often the site is checked
- Whether the site responds: every ten minutes, on the tracked address. That interval is the difference between “the site went down” and “a customer told us the site was down”. The screen shows how many times the site went offline in the last 30 days, the average response time and the list of outages.
- Indexing, tracking and security: once a day, overnight.
- Speed and user experience: a full Lighthouse run once a week, on a fixed day for each site.
- Which AI crawlers can read the site: once a week, testing each AI crawler’s user agent one at a time.
After the checks, Intelligence compares today with the previous reading. If a point dropped a band or an AI crawler can no longer read the site, a warning is created. The same situation never becomes two warnings, and the warning clears itself once the problem is gone. Most days there is no warning at all, and that’s how it should be: an alert that fires every day teaches people to ignore the screen.
Not measured is not the same as zero
A check can fail: the site took too long, blocked the crawler, the page didn’t load. When that happens, the screen says the point wasn’t measured in that run and shows the last measured value with its date. It doesn’t turn a failed reading into a zero or into “broken”. That sounds like a detail until someone presents a problem to the board that the site never had.
The same goes for INP, the Core Web Vitals metric for responsiveness. It only exists with real visitor interaction, so the lab reading reports it as not measured instead of making up a value.
Security: what you can see from outside, and what only shows from inside
From the outside, the check covers the basics for any site, whether WordPress, an online store or another platform: whether HTTPS is enforced, whether the certificate is valid, whether HSTS and the security headers are set, and whether cookies are flagged as secure. It also shows when the server says too much about itself, such as the exact software version it runs, which helps anyone looking for a target.
On WordPress sites with the Incuca plugin installed, the check goes further: the versions of what’s installed are compared against the public catalog of known vulnerabilities, the NVD, and the screen also shows what the site’s defenses blocked. A site without the plugin isn’t treated as insecure, only as a site we can’t see from the inside yet.
An llms.txt file built from your own pages
On the same screen, the “Generate llms.txt” button builds the summary file for AI readers from the site’s own pages, ready to publish at the root of the domain.
How it differs from PageSpeed Insights and Search Console
PageSpeed Insights measures one page at the moment you ask. Search Console shows what Google indexed and which searches the site appears for. Both are official tools and still worth using; Search Console can even be connected to Intelligence as a data source.
What Site health adds is monitoring over time: the same check on a schedule, with history, a warning when something gets worse, and every point on one screen. It also lives in the same account that shows where visits come from and how much became revenue, in the buyer journey. A site that slowed down on a Tuesday can be compared with what happened to that week’s leads.
How to get started
Start with a one-time reading: the free website diagnosis shows speed, how Google and AI assistants read the site, structured data, tracking and security, with a fix for each finding. To follow it every day, with history and warnings, book an Intelligence demo with the Incuca team.
References
- [1] Chrome for Developers. Lighthouse performance scoring (0 to 49, 50 to 89 and 90 to 100 bands). 2026. https://developer.chrome.com/docs/lighthouse/performance/performance-scoring
- [2] web.dev. Web Vitals (LCP, INP and CLS thresholds at the 75th percentile). 2026. https://web.dev/articles/vitals
- [3] NIST. National Vulnerability Database (NVD). 2026. https://nvd.nist.gov/
FAQ
Frequently asked questions about website health
What usually comes up before a demo.