Weekly Updates / October 11 to October 17, 2026
We built a 10-city local network, 91 herbal guides and a public AI search dataset.
- BellyUp! now helps people find sports, specials and local creators in ten cities.
- Natural Herb Library now connects a condition to 55 formulas and 36 patterns.
- The public AI search case study lets anyone inspect four months of portfolio data.
Audio Summary
Listen to the Weekly Summary
This 4-minute-57-second Kodi readback explains how the week turned collections into useful systems, what the new 205-property baseline says, and which checks still failed.
Read by Kodi, AI co-architect
Read transcript
This week, October 11th through October 17th, was about turning collections of information into useful systems people can actually use.
The portfolio grew from 127 public websites and properties to 205. That's 78 new properties added.
Over the latest 28 days of Google data, the network received nearly 173,000 search impressions and 720 clicks. The average Google search position was about 41.
The biggest developments came from BellyUp!, Natural Herb Library, and AISymantix.
BellyUp! expanded to 10 cities, with Birmingham becoming the newest addition. Birmingham launched with 252 venues, 31 neighborhoods, and 18 guides.
Sports and game day coverage is now available in all 10 cities. The bar network also expanded, with 207 venues featuring current happy hours or specials.
The Local Creators and Down the Rabbit Hole projects expanded beyond Jacksonville into every city. There are now 63 creator profiles, with videos, social posts, and local stories appearing directly on BellyUp! websites.
All 81 BellyUp! properties now have verified Google Search Console access and submitted sitemaps.
Natural Herb Library made major progress. The library already contained more than 1,000 herb guides and 140 condition pages.
This week, it added 55 traditional Chinese herbal formulas and 36 traditional Chinese medicine patterns. That means visitors can now start with a health condition, learn about different patterns, explore related formulas, and see which herbs are used.
The receding gums guide became the first complete example of this system. All 2,437 sitemap pages passed the website crawl tests.
The formulas and patterns still need qualified review before being presented as anything beyond educational information.
AISymantix published a public study using four months of Google Search Console information and 35 days of server logs. The study examined how structured SEO data relates to AI crawler activity and search visibility.
About 21,700 requests reached the dataset system across 190 websites. Seven dataset pages received 226 Google impressions, but no clicks.
The site's average search position improved from about 88 to 30, although actual traffic remained very low.
The important part is that the study includes the underlying data, so other people can inspect the results instead of simply accepting the claims.
Other projects also moved forward. AIHealthcareNow connected 93 news stories to its company directory, creating more than 800 internal links. The directory now includes 63 companies.
Midland published 12 IBM Power 7 model pages, while other IBM Power websites received additional buying and planning information.
LeverageBuilder launched a working six-question assessment and six real portfolio examples.
There were also some important problems. BellyUp! continues to experience excessive crawler activity on filtered pages.
On October 14th alone, five crawler types made more than 1.45 million requests to filtered BellyUp! pages. Many of those pages contained no matching venues, but still returned successful responses.
A partial fix reduced the problem, but it did not eliminate it. The next repair needs to prevent search engines and AI crawlers from following endless combinations of empty filters.
Natural Herb Library also launched with a crawler restriction that accidentally blocked part of its public data. That restriction was removed, but the publishing process still needs improvement.
AINowGuide passed its technical checks but failed its editorial review because its purpose was too similar to Krisada.com. The project needs a clearer identity.
Looking ahead, the priorities are fixing BellyUp!'s crawler problem, watching whether local creators begin participating, reviewing the herbal formulas, and measuring whether the AISymantix study produces real search traffic or AI citations.
The portfolio now has 205 active public properties, but because so many were added recently, the next reporting period will be important for establishing a fair performance comparison.
The biggest takeaway from the week is that the network is moving beyond simply publishing more pages.
BellyUp! is becoming a connected local discovery system. Natural Herb Library is becoming a structured educational resource. And AISymantix is making its research available for anyone to inspect and test.
The systems are growing. The next challenge is proving that people can find them, use them, and benefit from them.
This was a week of turning lists into systems people can actually use. BellyUp! became a ten-city network with sports, current bar deals and local creator pages. Natural Herb Library gained the missing middle between a condition and its herbs. Then four months of portfolio data became a case study anyone can inspect instead of a victory claim nobody can test.
2026-09-17 to 2026-10-14
The numbers at publication
This frozen baseline contains 205 active public websites, 78 more than the October 10 edition. Search Console is complete for September 17 through October 14 with 720 clicks and 172,917 impressions. Ratio-based click-through rate is 0.416 percent and impression-weighted average position is 40.701. The separately validated server-request cache covers the same 28 dates with 10,687,641 successful content requests, most of them training requests to BellyUp! filter combinations. The embedded snapshot records 234 portfolio events dated October 11 through October 17 and stops at event 1915.
Why 10.7 million requests did not mean 10.7 million readers
The portfolio grew by 78 websites in one week
That is 78 more websites in one week, so this week starts with a new baseline.
What Google recorded
0.416% click-through rate average position 40.701, where lower is better
Why the server-request total looks enormous
- Search crawlers
- 346,966 3.25%
- AI search
- 17,696 0.17%
- User retrieval
- 7,505 0.07%
Weekly work record: 234 events across 62 primary domains, frozen at event 1,915.
- Public websites in this snapshot
- 205
- Portfolio work-log records
- 234
- Google impressions
- 172,917
- Google clicks
- 720
- Click-through rate
- 0.42%
- Average position
- 40.7
This snapshot is saved with the edition. Work-log dates and search-data dates are shown separately; later dashboard updates do not rewrite this history.
BellyUp! became a ten-city network with sports, specials and local creators
Birmingham became the tenth BellyUp! city this week. Its hub and six focused sites opened with 252 venues, 31 neighborhoods and 18 guides.
Then Jacksonville stopped being the place where the new ideas lived alone. Sports & Game Day reached all ten cities through nine new sports sites. The Bars format reached all ten too, with 207 venues carrying a current happy hour or bar special from the venue's own pages.
The Local Creators and Down the Rabbit Hole pilot also moved from Jacksonville to the other nine cities. The network now has 63 creator profiles and one credited first story in every city. Creator videos, Instagram posts and real page pictures play on BellyUp! instead of sending the visitor away.
A person can now choose a game-day stop, find a real bar deal or meet someone documenting the city without leaving the local network. That is the change. Ten city lists became ten connected publishing systems.
Natural Herb Library can now connect a condition to a formula and pattern
Natural Herb Library already held 1,054 herb monographs and 140 condition indexes. It did not have the layer that answers the question a reader is really asking: which Chinese herbal formula fits this pattern?
The site now has 55 classic formula pages and 36 traditional Chinese medicine pattern pages. Every formula lists its ingredients and links each supported ingredient to the matching herb monograph. Each formula also has the same future product slug reserved on Supplements Apothecary.
The receding gums page is the first proof of the full system. It explains three different patterns and connects each one to a formula, daily care and signs that need professional attention. The site now moves from condition to pattern, formula and herb instead of stopping at a long ingredient library.
All 2,437 sitemap pages passed the rendered Dataset crawl. The next limit is clear too: the formula and pattern records still need qualified review before the library should present them as more than researched educational guidance.
Four months of AI search data became a public case study
AISymantix published a case study built from 35 days of server logs and four months of Google Search Console data. The article carries seven tables. Its full data file is public beside it.
The numbers did not produce a clean victory story. The dataset layer received about 21,700 requests across 190 sites. Seven AISymantix dataset pages earned 226 Google impressions and no clicks. The site's impression-weighted position moved from 88.2 to 30.4 while weekly clicks stayed between zero and two.
The raw AI crawler rise was left out of the result because BellyUp! filter pages dominated it. That choice matters. A large number is not useful when the mechanism behind it is a crawler walking combinations that no person asked for.
The case study gives us a dated measurement we can rerun. It also gives the reader the tables and JSON needed to challenge the conclusion. That is a better DataSetSEO asset than a polished claim with no receipt behind it.
The useful systems kept spreading
AIHealthcareNow connected its company news to the company directory. Ninety-three stories now create 806 links to 50 profiles. The directory reached 63 companies, while documented ownership research attached 22 older FDA records to the companies that own them today.
Midland published twelve Power 7 parent model pages with specifications, room planning and three researched parts panels on each page. Power 5, Power 9 and Power 11 sites also received new decision pages this week.
LeverageBuilder stopped describing a framework from a distance. It now has a working six-question assessment and six public portfolio cases that show the system being used.
BellyUp! Search Console setup caught up with the network. All 81 websites now have full service access and a confirmed sitemap submission. That does not create visibility by itself, but it gives the next comparison a complete place to start.
The public portfolio grew from 127 websites to 205, so the baseline resets again
This issue freezes 205 active public websites. The previous edition froze 127. Seventy-eight websites joined the public portfolio and none left, so the rolling totals are a new baseline rather than a like-for-like performance change.
Google Search Console is fresh through October 14 and contains all 28 dates from September 17 through October 14. Those websites recorded 720 clicks and 172,917 impressions. Click-through rate is 0.416 percent because it is total clicks divided by total impressions. The impression-weighted average position is 40.701, where lower is better. The parent-child overlap check found zero duplicate page-days.
The separate server-request cache covers the same 28 dates under algorithm version 2. It recorded 10,687,641 successful requests to content routes. Training crawlers account for 10,315,474 of them. BellyUp! filter combinations dominate that total, so it is not a measure of readership or demand.
The editorial review covered all 262 optimization events after the prior cutoff and stopped at event 1915. The embedded snapshot contains 234 portfolio events dated October 11 through October 17 across 62 primary domains. Twenty-eight later records dated October 10 stayed in the candidate review but outside the dated weekly count.
The biggest number of the week was also the easiest one to misunderstand
Five crawler names made 1,455,629 query-string requests to BellyUp! lens sites on October 14. The server returned a successful page for 1,454,675 of them. Fifteen of seventeen plausible filter combinations showed no matching venue while still returning HTTP 200.
The selective three-filter cap reduced the depth of the crawl. It did not close the trap. No new preventive check passed this week, so this is a problem found again rather than a lesson learned. The next repair must stop publishing crawlable filter states and return a real not-found response when no venue matches.
Natural Herb Library also launched its formula system while robots.txt still told general crawlers to stay out of the public data folder. The block was removed and the portfolio audit passed. The release process did not change, so this was another mistake found again.
AINowGuide published a six-page sprint that worked technically but still borrowed Krisada.com's job instead of owning its own public build record. The review failed it and an identity reset is still waiting. A page can pass every build check and still fail the reason it exists.
What we are watching next
Whether BellyUp! removes crawlable empty filter states and what seven complete post-fix days show by crawler.
Whether the ten-city creator pages earn corrections, claims or story submissions from the people they feature.
Whether qualified review changes any Natural Herb formula or pattern record before the first product pairing goes live.
Whether the AISymantix case study earns data-file use, Google clicks or repeatable AI citation results at the next reading.
Whether the same 205 websites stay in the measurement long enough for a clean comparison and whether the new BellyUp! websites begin reporting Search Console data.
Whether AINowGuide receives its own site identity before another technically complete sprint is published.
Explore the work
- BellyUp! Birmingham: tenth city hub
- BellyUp! Jacksonville: Sports & Game Day
- BellyUp! Atlanta: Bars
- BellyUp! Jacksonville: Local Creators
- Natural Herb Library: classic formulas
- Natural Herb Library: traditional Chinese medicine patterns
- Natural Herb Library: receding gums condition guide
- Natural Herb Library: current crawler rules
- AISymantix: AI Search Visibility Using SEO Datasets
- AISymantix: public case-study data
- AIHealthcareNow: healthcare AI company directory
- Midland: Power 7 model catalog
- LeverageBuilder: working asset assessment
- DataSetSEO: October 3 filter recurrence record
- DataSetSEO: public portfolio directory
Dates, scope, and corrections
This edition covers Sunday, October 11 through Saturday, October 17. The candidate review started after the previous cutoff of 1653 and stopped at event 1915, the highest available ID at capture time. All 262 candidate records were reviewed with their structured notes and verification.
The embedded snapshot follows the dated coverage window. It contains 234 portfolio events across 62 primary domains. Twenty-eight later records dated October 10 remain outside the embedded count.
Related approvals, deployment entries and follow-up repairs were grouped into one story. The event count is not a count of unique projects or successful launches.
Search totals come from Google Search Console page data for September 17 through October 14. All 28 dates contain observations. The capture found zero duplicate page-days between parent and child websites in the measured group. A date with rows does not certify every website import.
This measurement includes 205 active public websites. The October 10 issue included 127. This edition does not present the two rolling totals as a like-for-like performance change. The exact current domain list is preserved in the embedded snapshot.
Crawler totals come from the separately validated server-request cache for the same 28 dates under algorithm version 2. They count successful GET requests to content routes and keep search, AI search, training and user retrieval separate. They do not prove indexing, citations, human readership or a ranking effect.
The crawler-filter accountability section uses retained raw requests for October 14 and live tests of representative empty filter combinations. Crawler frequency is not treated as human demand. Server logs and Search Console remain separate instruments.
The three lead stories were selected after the positive-progress and recurrence reviews. They describe useful systems a person can now use or inspect. Repeated failures stay in a separate accountability section.
The lead graphic uses verified work receipts from the current event window. Its counts describe shipped pages, websites and source windows. They are not claims about traffic effects.
Public source links are checked over HTTPS before publication. The issue JSON, rendered article, archive, feed, transcript, canonical, metadata, Article and Dataset schema, sitemap, federation catalog, mobile layout, image files and protected infrastructure are verified again after deployment.
: First publication recorded the event cutoff, portfolio website group, Search Console metrics, crawler totals and open accountability items.
: Added a responsive baseline graph that separates the Google Search Console totals from the server-request mix and shows why the 10.7 million request count is not readership.
: Added a checked Kodi audio readback and collapsed transcript. The frozen metrics and publication date did not change.
: Replaced cohort language with websites throughout the baseline and made the one-week gain of 78 websites the graph's visible takeaway.