Audits you can trust: sharper, more consistent reports

A report is only useful if you can trust it. This cycle we went through the audit end to end and fixed the small inconsistencies that quietly erode that trust — plus added a few signals so you always know how much weight to put on a given result.

Reports that don't contradict themselves

The audit blends deterministic checks with AI-written analysis, and occasionally the two could drift apart. Now they're reconciled before you ever see them:

• Recommendations match the checks. If our AI design review flips a check from failing to passing — say it confirms you do have a CTA above the fold — any recommendation built on that check is dropped automatically. No more "add a CTA above the fold" sitting above a green checkmark.
• Comments can't contradict their score. A category scoring 95 will never be described as "needs significant work," and a low score won't be called "strong."
• The AI sticks to your business. Recommendations can no longer describe an e-commerce store as a restaurant or a SaaS as a blog — the resolved site type is now enforced in the write-up, and re-generated if AI vision changes its mind about what you are.

A confidence signal on every audit

Some audits run on a full set of data (organic keywords, multiple inner pages, AI vision, AI search visibility); others have less to work with. Every report now shows a high / medium / low confidence badge next to the Growth Score — and explains why — so you know whether you're looking at the full picture or a first read. It's in the PDF too.

No more audits wasted on dead URLs

Before doing any work, Webmatik now runs a quick pre-flight check. If a URL is unreachable, returns a 404/5xx, is a parked "domain for sale" page, or redirects to a completely different domain, you get an instant, clear message instead of a hollow report — and no audit from your quota is spent. (Plain www ↔ non-www redirects are treated as the same site, as they should be.)

AI Readiness and AI Search Visibility, clearly separated

These are two different things, and we now name them consistently everywhere — in the report, the PDF, and across the site:

• AI Readiness (GEO) — the on-page generative-engine signals you control: an llms.txt file, FAQ/Q&A structure, E-E-A-T signals, AI-parsable markup. This is one of your eight scored categories.
• AI Search Visibility — the outcome: whether ChatGPT and Gemini actually mention and cite your brand when buyers ask. We measure it with real queries (on the Growth plan), and it's shown as its own metric rather than folded into your score.

A few smaller refinements

• Faster audits. We moved our infrastructure to US-East, right next to our database — less latency on every run.
• Cleaner category navigation in the report, with full category names that no longer get cut off.
• Accessibility notes now describe accessibility specifically, instead of reusing the visual-design summary.

There's nothing to switch on — all of this applies automatically to new audits. Run one and the report should simply feel a little more honest.