Original data · transparent pilot benchmark
2026 GEO readiness benchmark for service-business websites
A dated, reproducible snapshot of 17 publicly accessible business websites in Ireland, United Kingdom, and United States, scored with the same deterministic Generative Engine Optimization rules used by the free AI visibility checker.
Results at a glance
What did the pilot scan find?
This is first-party observational data from a defined convenience sample. It is not a population estimate, a ranking study, or evidence that any single technical signal causes AI citations or organic traffic.
Crawler-specific observations
| Crawler and purpose | Allowed | Blocked | Unknown / no robots.txt |
|---|---|---|---|
| OAI-SearchBot — ChatGPT Search crawling | 94% | 0% | 6% |
| PerplexityBot — Perplexity indexing | 94% | 0% | 6% |
| GPTBot — OpenAI model-training crawler | 88% | 6% | 6% |
Why the distinction matters: GPTBot controls OpenAI model-training access; OAI-SearchBot is the crawler associated with ChatGPT Search. “Unknown” means the standard robots.txt could not be verified and is not silently counted as either allowed or blocked. See OpenAI’s publisher documentation.
Eight-category view
Median score by measured category
Each category is scored from 0 to 100 before its published weight is applied to the overall result. A category median describes this sample only.
| Category | Weight | Sample median |
|---|---|---|
| Crawl access and indexability | 15% | 100/100 |
| Site discovery and architecture | 10% | 100/100 |
| Entity and offer clarity | 15% | 80/100 |
| Answer-ready content | 15% | 60/100 |
| Evidence and citation quality | 15% | 80/100 |
| Author and ownership trust | 10% | 40/100 |
| Structured data and consistency | 10% | 50/100 |
| Experience, freshness and measurement readiness | 10% | 80/100 |
Sample and method
Exactly what was measured?
Sample frame
- 17 included service-business websites; 3 excluded scans.
- Markets represented: Ireland (6), United Kingdom (7), United States (4).
- Sites came from a pre-existing public-business research list; no site was selected because of its expected score.
- Only aggregate results are published. The benchmark does not expose company names, email addresses, or individual scores.
Technical scope
- Run date: 2026-08-28; deterministic rule version: geo-readiness-v1.0.0.
- Homepage plus up to six representative same-site pages.
- Robots rules for OAI-SearchBot, PerplexityBot and GPTBot; standard sitemap availability; visible content and schema signals.
- Only sites with at least one readable public page and at least 50% measurable coverage were included.
Interpretation limits
What these numbers do—and do not—mean
Use this as a diagnostic baseline, not a promise. Public websites change, crawler policies can be user-agent specific, some signals cannot be verified without analytics or server logs, and AI platforms make independent citation decisions. The scan measures whether useful foundations are observable; it does not claim direct access to ChatGPT, Google, Perplexity or their proprietary indexes. This dataset does not promise traffic, rankings, citations, leads, or revenue.
The most useful comparison is therefore within one website over time: document the baseline, fix the highest-impact gaps, rescan with the same rule set, and combine the result with Search Console, analytics and real AI-answer monitoring.