We measured 32 businesses in Sapporo
Measured 21 October 2026
"We do not show up in AI answers" is a question we hear more and more. So rather than speculate, we fetched the homepages of 32 businesses across five industries in Sapporo, one page each, and measured them.
By industry
Where the sample is small, one site moves the average. Read these as tendencies.
| Industry | Sites | Average | No structured data | Address and phone not machine-readable |
|---|---|---|---|---|
| Child development / after-school day services | 8 | 71 | 7 | 7 |
| Care and welfare providers | 5 | 70 | 3 | 4 |
| Restaurants and cafés | 4 | 55 | 4 | 4 |
| Professional services | 9 | 68 | 7 | 9 |
| Manufacturing and construction | 6 | 73 | 3 | 6 |
What the measurements showed
This is the only machine-readable statement of what a business actually does. Without it, AI and search are left inferring from prose, so the business does not surface when someone asks for that service in Sapporo.
Printed on the page is not the same as stated in the data. Without it there is no link to maps, local search or Google Business Profile, and no way to tell one similarly named organisation from another.
Blocking would make a site impossible for ChatGPT or Perplexity to cite, but none of these 32 did. The door is open. The problem is that there is nothing machine-readable behind it.
A single plain-text summary for generative AI. Not yet required, but almost nobody in these fields has one, so being early is an advantage.
Shared into LINE or Facebook, the link appears as bare text. In fields that travel by word of mouth, this matters.
- All 32 ran on HTTPS. A few years ago this was the common failure; it is solved.
- All 32 had a meta description set.
- Only four lacked the mobile viewport tag.
How to read these numbers
- An average of 68 is not "bad". It describes sites that read fine to a person but cannot be read by a machine.
- The two highest scorers (both 93) simply had structured data. The gap is doing it versus not doing it, not rebuilding versus not rebuilding.
- The lowest average was restaurants and cafés (55). In food and beauty, many businesses have no site of their own and rely entirely on booking platforms, so that part of the market could not be measured at all.
- For the lowest scorers the fix is a few dozen lines, not a new website.
- These 32 are only the public sites we could find, not every business in Sapporo. Those with no site at all could not be measured, so reality is likely worse.
- When first published (at 13 sites), this report stated that 1 of 13 blocked generative AI crawlers in robots.txt. That was wrong: the correct figure is 0.
- The cause was a defect in our own checker. In robots.txt everything after a # is a comment, and we were not discarding those lines, so examples written as comments were read as live rules.
- The average was corrected from 70 to 71 at that time. The sample has since grown to 32 sites and the current average is 68.
- The checker is fixed and the same fix is live in our public free check.
How we measured it
- On 21 October 2026 we fetched one page — the homepage — for each business found via search, across five industries: child development, elderly care, food, professional services, and manufacturing and construction.
- For each we also fetched robots.txt, sitemap.xml and llms.txt, then scored 14 weighted checks.
- We identified ourselves (HagumusFindabilityCheck) and requested sequentially. The visit appears in each site's own logs.
- Three sites we could not reach (one domain does not resolve, one did not respond, one returned an error) are excluded from the totals.
- The same engine powers the free check published on this site, so you can measure your own the same way.
- We do not publish company names, because we did not ask permission to.
See your own score, right now
A URL and 30 seconds. No sign-up, no email. Everything shown on screen.
