Topical Authority Map

Wine & Spirit Academy Bangkok runs a bilingual (EN/TH) WordPress + WooCommerce site selling sommelier certification training, wine and sake classes, tastings, corporate events and wine tours. The English inventory (119 indexable URLs of 229 total; the remainder are Thai /th/ variants, treated as regional mirrors rather than separate graph nodes) already shows deliberate clustering: posts live under /wine/ and /news/, learning hubs under /learn/…, and a WooCommerce catalogue of course products.

The graph reconstruction found 6 primary clusters around the central entity Wine & Spirit Education, with the commercial intent “get professional wine, sake or sommelier education or certification in Bangkok”. The dominant weaknesses are structural rather than topical: a coarse 10-category WordPress taxonomy in which the generic “Wine” category acts as a catch-all, a sparse inferred internal-link layer (77 edges across 121 graph nodes), and several clusters whose supporting content is thin relative to their commercial products (Sake Education, Wine Tours, Corporate & HORECA). No strong cannibalization groups were detected among English pages. Twenty expansions are recommended (Section N), prioritized by bridge value, cluster fit and cannibalization safety, with a four-phase implementation roadmap (Section R).

Source: https://wsa-bangkok.com/sitemaps.xml (Yoast-generated index, 5 child sitemaps, fetched 2026-08-18). 229 URLs discovered; 119 English, ~110 Thai variants under /th/.

robots.txt disallows only WooCommerce internals (wc-logs, transient/upload dirs, ?add-to-cart) and wp-admin — no content paths blocked. Hreflang pairs EN ↔ TH confirmed on the homepage.

Top semantic weaknesses

1. Catch-all “Wine” category. The generic Wine category absorbs posts from every wine-related cluster, blurring the boundary between Sommelier Training, Tasting Classes and Pairing content for crawlers and users alike.
2. Thin sake & tour clusters. Sake Education (6 EN pages) and Wine Tours (1 EN page assigned) are commercial offerings with product pages but little supporting informational depth.
3. Certification comparison gap. WSET / CMS / FSS comparison content exists as a single post; there is no level-by-level or cost-comparison support for the highest-intent certification queries.

Top architectural weaknesses

1. Sparse internal-link layer. Only 77 inferred edges across 121 graph nodes; many posts show no detectable contextual links to their cluster hub or a product page.
2. Flat taxonomy depth. Content lives at depth 3 (/wine/slug/) with no intermediate topic directories, so hierarchy must be carried entirely by categories and links.
3. Doubled crawl surface. Full TH mirrors of every page double the crawl and index surface; hreflang is present, but mirror content must stay synchronized.

Top authority risks

1. Commercial course/product pages receive limited informational support relative to their revenue role.
2. Outer lifestyle/pairing content has weak paths back to Core commercial destinations.
3. One dominant entity (“Wine”) risks absorbing topical credit that should accrue to narrower entities (sommelier certification, sake, HORECA training).

Reconstructed semantic graph: 121 nodes (50 posts, 41 products, 21 entities, 1 homepage, archives & utility pages) and 77 inferred relationships. Posts cluster around the six hubs; entity nodes act as bridges between clusters. Orphan risk is concentrated in utility/about/contact nodes, which is expected and harmless.

Colors: ■ posts · ■ products · ■ pages · ■ category archives · ■ entities · ■ homepage. Drag, zoom and hover for detail. Text summary: the graph is hub-dominated with most post nodes attached to one of six cluster anchors; product nodes are only loosely connected to informational posts.

Current vs. recommended cluster size

Bars compare current page count per cluster with the post-expansion target. Text summary: Sake Education, Wine Tours and Corporate & HORECA carry the largest relative gaps.

Confirmed from the live site (homepage, metadata, schema): Wine & Spirit Academy Bangkok (WSA Bangkok), founded by Anthony Caradec. Offerings: sommelier certification courses, sake classes, wine tastings and masterclasses (incl. French wine), corporate and private events, and wine tours. Sales run through WooCommerce (courses bookable as products). Bilingual EN/TH with hreflang. LocalBusiness schema present.

Search performance data (GSC) was not connected for this property at analysis time; the query framework in Section Q is therefore a structural hypothesis, not measured demand. Not confirmed from the supplied website data: pricing details, class sizes, accreditation partnerships.

21 normalized entities extracted from the English inventory. core entities carry commercial weight; outer entities extend semantic reach.

Strategic relationships only; speculative edges excluded. First 60 rows shown.

Derived from WordPress categories as actually used across the analyzed pages (10 categories; counts show how many analyzed pages carry each):

Note: Wine and Tasting function as near catch-alls, which is the main reason cluster boundaries must currently be inferred from content rather than structure.

Six clusters under the central entity, five Core and one Outer. Existing anchor pages and proposed topics are shown as children; Core clusters map directly to revenue products, Outer content must always link back toward a Core destination.

Click a node to expand/collapse its children. Circle color: ● Core · ● Outer/existing/proposed. Text summary of the full tree follows below.

Page counts per entity and search intent, computed from the analyzed English inventory. Navigational intent is near-zero everywhere — expected for a brand still building direct-search recognition.

Text summary: Bangkok and Wine dominate informational coverage; commercial-investigation coverage is thin for every entity except Bangkok — the layer where comparison and “best of” content should sit.

All 100 analyzed content pages (posts, pages, products) with role, intent, lifecycle and cluster classification. Use the filters to slice the inventory.

Method: title/entity/intent/URL-pattern overlap across English pages. Thai /th/ mirrors are excluded as regional variants (hreflang-managed), not competitors. Watch-list: the multiple “beginner wine class Bangkok 2026” posts flagged in the roadmap should be consolidated or sharply differentiated before new beginner content is published.

⚠ Simplified structural simulation based on inferred links and URL hierarchy — not Google PageRank. Treat as directional guidance only.

Each proposal repairs at least one classified gap, connects to ≥3 existing graph nodes (cluster hub + siblings + a commercial destination), and passed the cannibalization screen. Bridge types vary across comparison, commercial, lifecycle, problem–solution and audience bridges.

The full plan is the Section N table ordered by priority. Priority split:

Writing guidance (applies to every page): brief = 1,200–1,800 words unless comparison depth requires more; open with the reader's decision problem; every page links up to its cluster hub, sideways to ≥2 siblings, and down to ≥1 bookable product with varied, non-exact-match anchor text; include Bangkok context and confirmed facts only (mark anything unverified); mirror to Thai only after the English page stabilizes.

• Hub linkage (add): every post in a cluster gains one contextual link to its cluster hub (/learn/…) near the top third of the content; hubs gain a “courses in this track” block linking to the relevant products.
• Product support (add): each course product page receives ≥3 inbound contextual links from related informational posts (currently sparse: many products show few inferred inlinks).
• Cross-cluster bridges (add): sake ↔ wine pairing, corporate ↔ HORECA standards, tours ↔ training pathways. Varied anchor themes, never repeated exact-match anchors.
• Outer → Core paths (add): pairing/lifestyle posts must end with a path to a class or tasting product.
• Reduce: rely less on the catch-all Wine category as the only structural signal; keep category links but stop using them as the primary navigation for topic discovery.

Link inventory is inferred from URL/category structure and sampled content, not a full crawl; verify counts in a crawler before/after implementation.

Structural query hypotheses per cluster seed (GSC not connected — validate against real query data when access is granted):

• Sitemaps analyzed before any recommendation was generated; indexable/duplicate/utility URLs separated (TH mirrors treated as variants).
• 100 English content pages fully classified (posts, pages, products); category archives and utility URLs inventoried but not content-analyzed.
• Internal-link model is a simplified simulation (Section M) — directional only.
• No Google Search Console / Analytics data connected: coverage and recommendations are structure-derived, not demand-validated.
• No business facts invented; unknowns marked “Not confirmed from the supplied website data”.
• All visualizations render from the same JSON datasets used in the tables (data/*.json); no decorative charts.
• Generated 2026-08-18 from live sitemaps and the WP REST API of wsa-bangkok.com.