Example audits: before and after
These three worked examples show what a Zitatix audit finds and what fixing it does to the scores. They are composite case studies — each is built from real failure patterns we see constantly, with details anonymized and simplified so the fixes stay generalizable. Every score shown is a real Citation Readiness Score™ computation under methodology version 1.2 (see how scoring works).
Example 1: B2B SaaS with a JavaScript-only homepage
First audit: . Re-audit: .
A project-management SaaS with strong Google rankings scored far worse than the team expected, for one dominant reason: the site was a client-rendered single-page application. AI crawlers received an empty <div id="root"></div> — no headline, no feature copy, no pricing. GPTBot was also blocked in robots.txt by a rule copied from a blog post.
| Pillar | Before | After |
|---|---|---|
| Findable | 12 / 25 | 25 / 25 |
| Quotable | 4 / 25 | 21 / 25 |
| Understandable | 15 / 25 | 23 / 25 |
| Trustworthy | 10 / 25 | 17 / 25 |
| CRS | 41 | 86 |
What changed: the marketing pages were pre-rendered to static HTML at build time (the app itself stayed a SPA); GPTBot, ClaudeBot, and PerplexityBot were unblocked; an FAQ section with question-shaped headings was added; Organization and FAQPage JSON-LD went in. Six weeks later, Perplexity cited the product's own pricing page for "how much does [product category] software cost" — a query previously answered from a third-party review blog.
Example 2: local services firm invisible to AI
First audit: . Re-audit: .
A regional electrical contractor had a hand-built site with decent HTML but almost no machine-readable identity: no structured data, no meta description, images without alt text, no sitemap, and an about page consisting of one sentence. Off-site, the AI Visibility Index was 18 — the business existed in almost no directories and had a handful of unlinked reviews.
| Pillar | Before | After |
|---|---|---|
| Findable | 14 / 25 | 23 / 25 |
| Quotable | 8 / 25 | 17 / 25 |
| Understandable | 6 / 25 | 19 / 25 |
| Trustworthy | 5 / 25 | 15 / 25 |
| CRS | 33 | 74 |
What changed: LocalBusiness JSON-LD with address, hours, and service area; a sitemap; descriptive titles and meta descriptions per page; a real about page naming the owner and licenses; an FAQ answering the questions customers actually ask ("what does a fuse-box replacement cost?"). Off-site, the firm claimed its profiles on the four directories that mattered locally — the AVI moved from 18 to 44 over the same period, and the re-audit was the baseline for tracking it further.
Example 3: e-commerce content hub with unquotable articles
First audit: . Re-audit: .
An online retailer's buying guides were long, well-researched — and structurally useless to answer engines. Articles opened with 400 words of scene-setting before the first fact, headings were clever rather than descriptive ("The plot thickens"), there were no tables or lists, and no dates. The content deserved citations and could not get them.
| Pillar | Before | After |
|---|---|---|
| Findable | 22 / 25 | 23 / 25 |
| Quotable | 9 / 25 | 23 / 25 |
| Understandable | 16 / 25 | 22 / 25 |
| Trustworthy | 11 / 25 | 22 / 25 |
| CRS | 58 | 90 |
What changed: each guide got a two-sentence summary answer directly under the H1, question-shaped subheadings, a comparison table, Article JSON-LD with author and dates, and outbound citations to the standards and manufacturers referenced. The prose stayed — it was reorganized, not rewritten. The Quotable pillar jump from 9 to 23 accounts for most of the 32-point gain.
The pattern across all three
- The biggest single killer is content that is not in the HTML — either blocked in robots.txt or rendered only by JavaScript.
- The cheapest points are metadata: titles, descriptions, JSON-LD, dates, alt text. These are hours of work, not weeks.
- Quotability is an editing problem, not a writing problem. Existing content usually contains the answers; it just doesn't lead with them.
- On-page fixes move the CRS within one re-crawl cycle. The AI Visibility Index moves slower, because it depends on independent sources — start the off-site work early.
Want the same breakdown for your site? Run a free scan — the report shows exactly which checks failed and how many points each fix recovers.