The fastest way to find out whether a diagnostic tool is any good is to point it at itself. So we did. We ran citescan.ai through its own Discovery Readiness audit, took the result on the chin, worked the fix list, and re-scanned.
The homepage went from 43/100 (Grade C) to 69/100 (Grade A) — a 26-point gain — across one focused session of changes. Below is the honest version: what scored low, what we did about it, and what actually moved.
The starting line: 43, Grade C
The first scan was not flattering. Here's where the three pillars landed.
| Pillar | Before | What it measures |
|---|---|---|
| Visibility | 38 | Whether AI engines can find your content and recognize your brand as a real entity |
| Credibility | 34 | Whether AI engines trust your content enough to quote it |
| Technical | 55 | Whether your site is structured so AI engines can cleanly extract your content |
The sub-scores told the real story. Structured data sat at 0/100 — the page carried no JSON-LD at all, so AI engines had to guess what our content was rather than being told directly. Content freshness was 15 — no visible dates, no dateModified in schema, so every engine that uses recency as a tiebreaker quietly deprioritized us. And our author signal was a name with no credentials: "built by Sameer Sadiq," full stop. No role, no bio, no link. To an E-E-A-T filter, that reads as anonymous.
The report's blunt summary of us: "The foundation is there. Most AI-specific signals are missing." Hard to argue with your own product.
What we changed
We worked the fix list in priority order — the same order the report hands every user, ranked by impact on the score. The high-leverage changes:
Structure for machines
The single biggest gap was zero structured data. We added Organization JSON-LD, WebPage schema, and a Person block. Structured data went from 0 to 52 in one push. Technical jumped 32 points.
Dates, everywhere
Added a visible "Last updated" line and datePublished / dateModified to the WebPage JSON-LD. Freshness was at 15 — every engine using recency as a tiebreaker was quietly passing us over.
A real author
Expanded the attribution from a bare name to role plus a linked bio, and added a Person schema block with jobTitle and sameAs. This one fix lifts two pillars at once — it's both a Credibility signal (E-E-A-T) and a Visibility signal (entity authority).
A clean definition
Rewrote the opening so the first thing under the H1 is a standalone, CTA-free definition of what an AI search visibility audit is — the 40–60 word answer paragraph that AI Overviews preferentially extract. Answer structure went from 24 to 68.
On-page basics
Fixed the title tag length and keyword order, added a meta description, and added the canonical tag that was missing entirely.
Every one of these had a ready-to-paste code block in the report. We pasted our own.
The finish line: 69, Grade A
| Pillar | Before | After | Δ |
|---|---|---|---|
| Visibility | 38 | 52 | +14 |
| Credibility | 34 | 68 | +34 |
| Technical | 55 | 87 | +32 |
| DRI (composite) | 43 | 69 | +26 |
And the grade moved two full bands:
A word on what an "A" means here, because 69/100 sounds modest until you see the scale. CiteScan's grades are calibrated against a reference corpus of real pages — and on that index, Wikipedia's strongest articles score in the 61–67 range. The bar for an A is deliberately set against the best-cited pages on the internet, not against an abstract 90%. A 69 isn't a near-miss; it's top-band.
Top of the reference band. Wikipedia's strongest articles cluster at 61–67. The A grade is the same bar applied to your page.
One honest note: the before and after scans also coincided with a correction to how the three pillars are weighted, so the 26-point composite delta reflects both the page improvements and a more accurate model underneath. What we can say cleanly is that the pillar movements — +14 Visibility, +34 Credibility, +32 Technical — came from real changes to the page, and a stable re-scan on June 15 confirmed the scores held.
Why this is the only honest sales pitch a scoring tool can make
Most AI-visibility tools tell you what's wrong. The gap in the entire category is that almost none of them do the work — they hand you a list of homework. We built CiteScan to hand you the finished fix: the actual JSON-LD, pre-filled with your domain and brand, ready to paste.
The proof that it works is that we used it on ourselves, in public, with the caveats attached. If you want to see where your own site stands against the same reference bar Wikipedia is measured on — and walk away with the code to improve it, not just a to-do list — that's the whole product.
Methodology note: Discovery Readiness Index is a 0–100 composite of Visibility (35%), Credibility (30%), and Technical (35%) pillars, each scored 0–100. Grades are calibrated against a reference corpus spanning best-in-class reference pages to parked domains. Scores are point-in-time snapshots; re-scan to measure change.