Anqor Studios improved its AI search visibility score from 42 to 75 by making three categories of change: fixing crawler access so AI indexing bots could actually read the site, repairing entity data so the business was unambiguously identifiable to AI knowledge systems, and rewriting key page sections to contain standalone factual claims that an AI answer engine can quote directly without additional context. Those three levers, applied in that order, produced a 33-point lift documented in the audit log published on this site. No single change moved the number alone. The crawler fixes were necessary but not sufficient. The entity repairs raised the floor. The claim-density work is what pushed the score into the 70s. If you want to improve your own AI search score and website GEO standing, the honest answer is: fix what blocks AI readers, establish who you are, then give AI systems something specific and citable to say about you.
---
Why an AI Search Score Is a Different Problem Than SEO
Most businesses treating GEO as an SEO variant are solving the wrong problem. Traditional search optimization is about ranking in a list of ten blue links. Generative engine optimization is about being the answer, or being cited as the source of the answer, when someone asks an assistant a question directly.
Perplexity and Google AI Overviews do not return lists. They return a synthesized response with citations. The citation selection is not purely link-authority-driven. It depends on whether the AI system can read your site, identify who you are with confidence, and find a chunk of your content that answers the question in a self-contained way.
The shift matters because a business that ranks third in a link list still gets clicks, but a business that is not cited in an AI-generated answer gets nothing from that query. According to research published by Ahrefs, AI Overviews already appear for a significant share of informational queries, and that proportion has increased through 2025 and into 2026. The queries where AI Overviews appear are often the highest-intent ones, which is exactly where you want to be named.
---
The Baseline: What a Score of 42 Actually Looks Like
When Ranqr, the AI search visibility audit product Anqor built and runs on its own site, produced a score of 42, the audit surfaced three distinct failure categories. Understanding them is more useful than the number itself.
Crawler Access Failures
AI crawlers are not identical to Googlebot, and many sites that perform well in traditional search have robot directives, JavaScript rendering requirements, or authentication gates that block AI-specific crawlers entirely. At baseline, parts of the Anqor site had structural issues that meant AI indexing bots were encountering barriers that standard SEO audits would not flag as problems.
The fix was not complex but it required knowing which crawlers to test against. robots.txt directives written to manage Googlebot crawl budget were inadvertently catching other crawlers. Adjusting those directives and confirming access with targeted crawler tests was the first remediation step.
Entity Data Gaps
AI knowledge systems work by building associations between entities: a business name, its location, its category, its principals, its products, and the relationships between them. When that data is incomplete or inconsistent across the sources AI systems ingest, the system either attributes information to the wrong entity or simply lacks the confidence to cite you.
For a single-person studio operating in Dubai under a name that does not appear in major business directories with consistent structured data, the entity problem was real. The remediation involved ensuring that every page carried coherent structured data markup, that the business's name, location, and category were consistent across every surface AI systems can read, and that the site's own content referenced the business in third-person, named, quotable form rather than relying on implied context.
Quotable Claim Deficit
This was the hardest problem to see before doing the audit. AI answer engines extract specific, factual, self-contained statements. A sentence like "we build AI systems" is not extractable in a useful way. A sentence like "Anqor Studios raised its AI search score from 42 to 75 by fixing crawler access, entity data, and claim density" is extractable, attributable, and citable.
The single most impactful content change was rewriting the opening sections of key pages so that the first 150 to 200 words contained at least one standalone factual claim that could be lifted out and quoted without losing meaning. This is the same principle behind why this article opens the way it does.
---
The Three-Stage Remediation in Sequence
The order matters. Running entity repairs before fixing crawler access is wasted effort if the crawlers that would read your updated structured data cannot reach your pages. Rewriting content for quotability before fixing entity data means your quotable claims may be attributed to the wrong entity or no entity at all.
The re-score step is not optional. Without measuring against the same framework before and after, you have no way to know which changes moved the number and which were neutral. This is why the audit log matters more than the final score: the log records which change category produced which movement.
---
Before and After: What Changed at the Page Level
| Area | Score of 42 Baseline | Score of 75 After Remediation | |---|---|---| | Crawler access | AI-specific crawlers blocked by broad robot directives | Directives scoped precisely; confirmed open access for target crawlers | | Structured data | Partial and inconsistent across pages | Coherent, consistent markup on every key page | | Entity disambiguation | Business name and category not consistently co-located | Name, location, category, and service scope present together on every page | | Claim density | Opening sections written for human narrative, not AI extraction | Every key page opens with a standalone, attributable factual claim | | Third-party entity signals | Minimal directory presence | Consistent data across external sources AI systems are known to ingest |
---
The score moved from 42 to 75. The audit log recording the specific changes is published on the site and is verifiable against the page you are reading right now.
---
What This Means for Businesses Trying to Improve Their Own GEO Standing
The Anqor case is useful precisely because it is not a hypothetical. The site it describes is the site you are reading from. The score it claims is checkable. That specificity is itself a demonstration of the principle: AI systems cite sources that make verifiable, concrete claims, not sources that describe things in general terms.
For operators in sectors like real estate or hospitality, where inbound enquiries increasingly originate from AI-assisted research, appearing in the cited sources of an AI answer is not a content marketing nicety. It is a lead source. A property buyer in Dubai who asks an AI assistant "which AI automation studios work with real estate agencies in the UAE" will see a synthesized answer. That answer will cite two or three sources. Whether your business is one of them depends on whether AI systems can read your site, identify you, and find something specific to say about you.
The three-stage process described above applies regardless of industry. The scoring framework Anqor uses through Ranqr exists because the studio needed to measure its own progress, and the same framework is now applied to client audits. If the methodology was not good enough to move this site's own number, it would not be offered to other businesses. That is a constraint Anqor operates under by design.
---
The Constraint That Makes This Credible
One thing worth saying plainly: the 42-to-75 movement is not the ceiling. The audit log records where the score plateaued and why. Some factors that influence AI citation are not within a site owner's direct control, including the decisions made by individual AI systems about which sources to weight in which query contexts. What is within control is the three-stage foundation. A site that scores in the 40s on a GEO audit has structural problems that no amount of content production will solve. A site that scores in the 70s has earned the right to compete for citations on the basis of content quality and topical relevance.
The audit log is not published as a trophy. It is published because a claim without a verifiable record is the same as no claim at all, and because the act of publishing a checkable audit log is itself an example of the quotable, specific, attributable content that GEO rewards.
---
FAQ
How long does it take to improve an AI search score?
The Anqor Studios score moved from 42 to 75 over the course of the GEO audit and remediation cycle documented on the site. The timeline depends on how many crawler access issues, entity gaps, and structural problems exist at baseline. Sites with clean technical foundations but thin entity data tend to move faster than sites with both categories of problem.
What is a GEO score and how is it measured?
A GEO score measures how well a site's content, structure, and entity data position it to be cited by AI answer engines like Perplexity, ChatGPT search, and Google AI Overviews. Anqor measures this against a repeatable audit framework that checks crawler accessibility, structured data completeness, entity disambiguation, and the presence of standalone quotable claims. The score is not a third-party platform metric but an internally defined, consistently applied benchmark.
Is GEO the same as SEO?
They overlap but are not the same. Traditional SEO optimizes pages to rank in a list of links. GEO optimizes content to be extracted, quoted, and cited by AI systems that answer the question directly rather than listing pages. Some fixes, including correcting crawler access and structured data, help both. Others, like writing standalone factual claim blocks, are specific to GEO.
Can I verify that Anqor's score actually moved?
Yes. The score, the methodology, and the audit log are published on the Anqor Studios site. The homepage states the before figure (42) and after figure (75) directly, and the audit log records the specific changes made. It is not a case study with anonymized numbers; it is a live, checkable record.