Business owners keep asking us the same question: are we behind? We looked at our own scan records to offer a starting point, not a verdict on business quality.
AI readiness here measures how easily the AIready scanner can access and interpret a website. It does not measure how often AI assistants cite or recommend a business, and it is not a standard shared by every AI platform.
What did 158 submitted websites score?
Across 158 unique websites scanned with AIready between 4 September and 17 October 2026, the median score was 55 out of 100 and the mean was 53. Of those sites, 33 scored below 40 (21%), 81 scored 40–69 (51%), and 44 scored 70 or above (28%).
One honest caveat, stated once: these are recorded scores from a self-selected sample of submitted websites, not a representative survey of New Zealand businesses. An AIready score reflects how accessible and clearly structured a website is to the scanner; it does not prove, and does not predict, whether ChatGPT, Claude, Perplexity or any other assistant will cite or recommend the business.
How were the figures calculated?
The snapshot contains 363 scan records. To avoid counting repeat submissions as separate websites, we used the latest recorded score for each unique domain, leaving 158 websites. The median is the middle score; the mean is the arithmetic average, rounded to a whole number.
Each scan returns five scored measures. For this article they group into three pillars: bot access and crawlability (can an automated visitor retrieve the page), structure and discoverability (is the important information present and easy to find), and content clarity and specifics (does the page state concretely what is offered, for whom, and on what evidence). The snapshot reports aggregate scores only, so it cannot isolate how many points any single pillar contributed to a total.
AIready is based in New Zealand, but the submitted domains were not restricted to verified New Zealand businesses, so these figures should not be described as a national benchmark. The below-40, 40–69 and 70-plus ranges are reporting groups, not product ratings, and a low recorded score can reflect a temporary scan error as well as a genuine access problem — investigate and repeat the scan before drawing conclusions. See the aggregate figures and full sample limitations on the AI Readiness Benchmark page.
What does the middle score range tell us?
Just over half of this sample scored 40–69. That tells us where their recorded totals sit, but not why. Without a published per-check analysis, we cannot say that all 81 sites share the same gaps. The useful next step is to inspect each site’s results rather than assume the same fix will work for everyone.
Do big brands automatically score better?
No brand-size comparison was performed for this snapshot. A well-known business can still be difficult for a particular scanner to access: automated-visitor challenges, script-dependent content or redirects can interfere with retrieval. These are possible explanations to investigate, not measured findings about the brands in this sample.
What information makes a website easier to interpret?
Specific service descriptions, the locations served, clear contact details and verifiable evidence give readers and automated tools more concrete information than phrases such as “quality service” or “tailored solutions”. This is a recurring pattern seen across our scan reviews rather than an isolated variable measured in this sample, so treat it as a practical check to make, not proof that vague wording caused a particular score in this dataset.
Why does this matter for New Zealand businesses?
1News reported on 4 September 2026 that OpenAI said New Zealanders send almost 10 million ChatGPT messages a day. That is an OpenAI-provided usage figure reported by 1News, not an independent measurement from AIready. It provides context for AI adoption; it does not show how often AI recommends businesses or validate this sample as NZ-only.
What can you do today? Three steps.
- Check. Run a free scan on your website and see your score and the individual findings behind it.
- Investigate. Look at the specific checks that held the score back. If a scan reports an access error, fix the cause and repeat the scan rather than immediately rewriting the site.
- Compare. Ask the questions your customers actually ask across ChatGPT, Claude and Perplexity, and record which businesses are named and which sources are cited. That is the separate test of whether AI recommends you.
Run a free scan on your website and one competitor. Inspect the individual findings, fix one confirmed gap, then scan again.
AI readiness: common questions
What is a good AI readiness score for a website?
In this AIready snapshot, 70 or above is the upper reporting group: 44 of 158 websites (28%) reached it. It is not a universal pass mark or a guarantee of citations. Inspect the individual scan findings alongside the total score.
Why can a well-known brand score poorly on AI readiness?
Access challenges, script-dependent content, redirects or temporary scan failures can limit what a scanner retrieves. This snapshot did not compare brand sizes or quantify those causes. Check the specific findings and repeat an error-affected scan before judging the website.
Does an AI readiness score show whether AI recommends a business?
No. The score describes access and interpretation by the AIready scanner, not actual mentions, citations or recommendations. Assess those separately using consistent customer questions across AI platforms and verify the sources each response cites.
Written by Denvor Snowden, Co-Founder of AIready. Figures come from AIready scan records.