How we audit

By Vishal · Published 13 October 2026

1. Crawl

We crawl every reachable URL with a desktop and a mobile user agent, respecting robots.txt, up to 2,000 URLs (larger sites are quoted). The crawl records status codes, redirects, canonicals, indexability directives, titles, descriptions, headings, word counts, structured data, images and alt text, internal links in and out, and response times. Orphan pages are found by comparing the crawl against the XML sitemap and, when shared, Search Console's page list.

2. Performance

Core Web Vitals come from the Chrome User Experience Report (field data from real users) for the origin and for each key template where the site has enough traffic. Lab diagnostics (Lighthouse) are used only to explain field numbers, never to replace them. We report LCP, INP and CLS against Google's thresholds — 2.5 s, 200 ms, 0.1 — and set tighter internal targets for the fix list.

3. Search Console and keywords

With read-only access we export 16 months of queries and pages. From this we find: queries at positions 5–20 with impressions (the fastest wins), pages whose clicks or position fell more than 20% over eight weeks (the refresh queue), and cannibalisation — two or more URLs alternating for one query. Without access we approximate with third-party data and the SERP, and say so in the report.

4. The 60-point checklist

The checklist is public. Each check is marked pass, warn or fail with evidence (URL, screenshot or number). Checks are weighted: critical (indexing, HTTPS, mobile, rendering) ×3, on-page and content ×2, authority and nice-to-haves ×1. The score is weighted passes over weighted total. It is a diagnostic reading, not a ranking prediction.

5. AI visibility panel (Pro)

We write 20 buyer prompts for the category — never including the brand name — and run each in Google AI Mode, ChatGPT with search, Perplexity and Gemini from fresh, logged-out sessions with an India locale. For each answer we record: brand mentioned, brand cited with link, page cited, and competitors named. Citation share per engine is cited prompts over total. We also record which domains the engines cite for the category, because that tells you where mentions must come from. The same method, at 200 brands, produces the India AI Visibility Index.

6. Review and the fix list

Vishal reads the crawl and the checklist results page by page for the money pages and by pattern for the rest, then writes the verdict and orders the fix list. Ordering uses impact (expected traffic or conversion effect, from experience with similar sites and the Search Console data), effort (hours, from the developer's point of view), and dependency (what unblocks what). The top five items are always things that can be done in the first week.

Sources we rely on

• Google Search Central documentation — technical requirements, spam policies, the AI features guide, and the Search Console generative AI reports.
• Chrome User Experience Report for field performance; Lighthouse for diagnostics.
• Google Search Console and Google Ads Keyword Planner for query data; third-party indexes for links when a client has no access of their own.
• The engines themselves for AI visibility, by hand, every time. No scraped proxies.

What we do not do

We do not inflate scores to sell a retainer, invent volumes, pad fix lists with trivia, or recommend link buying, cloaking, doorway pages or scaled AI content. If something on the fix list seems wrong, email us and we will show the evidence or correct it.