FACETS-CKS-2026-10-13
Observed October 13, 2026. Research frozen before article drafting.

One English-language public bottle category at CKS Packaging, a US manufacturer
also present in our earlier B2B company sample. This is new category evidence,
not another independent sample of the US market. The category calls its list
partial. Product counts are listed references, not confirmed stock.

Twenty cases fixed before applying filters: baseline + six material singles +
two shapes + two markets + eight pairs + one triple. Two rounds by the same
researcher. Nineteen reachable cases per round, including baseline, yielded
38 complete product URL sets; each collected count matched the visible total.
Counts and sets matched between rounds. F19 could not be selected: Square was
disabled after Industrial + PET in both rounds. F19 count is unknown/null,
not zero. No empty-result page or corresponding HTTP status was observed.

The list is virtualized. We accumulated deduplicated product URLs while moving
through the visible list with ordinary browser controls. Protocol amendments
explain why End/fast-scroll probes were replaced by incremental viewport steps
and settled screenshots/AX observations. Incomplete and wrong-category probes
remain in the local log and are excluded. We cleared non-Bottles selections
between cases; no hidden endpoint, forced disabled control, guessed parameter,
authentication, comparison login or form submission was used.

All reachable selections kept the same /catalog/bottles URL and exposed no
filter anchor links to distinct routes. Three completed reload tests (F01,
F12, F15) reset to the 213-product baseline. The planned F19 reload was not
performed because the requested state was unavailable. No behavior is inferred
for untested selection orders or other filter combinations.

Category HTML and robots.txt retrieved twice with ordinary GET. Category 200;
robots.txt 404. Initial category HTML had no observed canonical/meta robots or
product anchors; rendered browser contained products. These observations do
not establish Google rendering, selected canonical, indexing or crawl budget.
Third-party response cookies, full HTML and DOM snapshots are not republished.

Sets compared using complete unique product URL lists, not count alone.
Industrial (56) and HDPE + Round (56): overlap 21, exclusive 35 on each side.
PVC (14) and Food + HDPE (14): overlap zero. Material counts total 216 while
their union contains 213 URLs. Equal counts do not prove duplicate pages;
even equal sets would need intent/content review.

No Search Console, search volume, internal search or sales evidence was
available for this catalog. Demand is unverified throughout. Nonempty distinct
sets are candidates for research, not automatic index recommendations.
Food/Industrial are source labels, not verified suitability or certifications.
No rankings, traffic, enquiries, AI citations or causal SEO effect measured.
No settings changed on the studied business. Dataset is a dated case study.

Preservation note: article QA accidentally reran the HTTP collector and overwrote
first-round raw HTTP files. Original sanitized metadata and frozen HTML analysis
were retained, along with second-round raw responses and all browser evidence.
Those supplementary QA requests are excluded from the original study counts.