East Fork — Agentic Readiness Audit
Worked example #2 · Audited 20 October 2026 · Pottery & home goods · 456 published products
Read this first: East Fork is not a client. We audited their public store as a complete dress rehearsal of a client engagement — the audit, the change list, and the approval process below are the real documents, produced exactly as a client receives them. No changes were made to East Fork's store. We publish it so you can judge the work before you ever speak to us.
60/100 "At risk"
Scored on publicly verifiable criteria only: 44 of 73 verifiable points. 27 points could not be verified from public pages and were excluded from the score, never guessed (listed at the end). 7 areas scored + 2 not publicly verifiable = 9 areas · 73 verifiable + 27 not verifiable = 100 points.
| Area checked | Score | Status |
|---|
| Product titles & descriptions | 10/15 | Gaps |
| Structured attributes & specifications | 4/15 | Fail |
| Product identifiers (GTIN / brand / MPN) | Not public | Excluded — see below |
| Variant naming & option clarity | Not public | Excluded — see below |
| Image alt text | 8/10 | Strong |
| Categorisation & taxonomy | 6/8 | Gaps |
| Shipping, returns & warranty information | 7/12 | Gaps |
| FAQ & Knowledge Base coverage | 6/8 | Gaps |
| Feed readiness signals | 3/5 verifiable (of 10) | Gaps |
The 5 issues that matter most
1. The best sellers publish no dimensions or capacity. On The Mug (4.9 stars, 1,731 reviews, badged "Best Seller"), the page text contains no capacity figure at all — the word "oz" does not appear anywhere in it. The Dinner Plate states no diameter; the Everyday Bowl (1,227 reviews) states no capacity. An assistant asked "how much does The Mug hold?" cannot answer from the store — yet retail partners publish these facts for the same products (a Zola listing gives a bowl's capacity per glaze; a Quince listing gives a serving bowl's diameter and capacity). The facts exist. They are not on the store's own pages in readable form.
2. Specifications are narrative-only on pottery — the candle pages prove the store can do better. Across eight pottery products checked, specs appear only as prose ("Made in Asheville from regionally sourced clay. Dishwasher and microwave safe."), never as a structured list. The one exception, a candle, publishes a proper specification list — dimensions, capacities, materials, burn times. The Big Mug states "Comfortably holds 14 oz"; its sibling The Mug states nothing. Assistants answer filter questions ("under 12 oz") from structured data; on the core line, it isn't there.
3. Policy facts are split across pages, and two are stale. The refund policy is complete and specific. But the shipping policy states no processing time and no shipping costs — the dispatch time ("currently shipping out orders within 10–14 business days") appears only on the FAQ page. That FAQ page opens with a free-shipping promotion that ran 14–18 May 2026 — five months out of date — and the shipping policy still carries a COVID-era paragraph about international transit. No warranty terms are published anywhere we could find, for breakable stoneware.
4. The collection structure has sprawled — including internal collections exposed publicly. Roughly 95+ public collections, including ones that appear to be internal: employee-pottery-discount-collection, pos-pottery-8-packs, referral-credit-collection, and one named test-featured-collection-blocks-holiday-2025. Near-duplicates sit side by side (gifts-under-50 alongside under-50). And the Dinnerware collection page exposes no product names or prices in its text layer at all.
5. Alt text is a strength with named exceptions — plus two data anomalies. Product alt text is genuinely good where it exists. But the featured images of several products carry none — including the Donabe Cookbook, Tiny Cup, The Kulhad, Flower Frogs and the Serving Board. On one candle page, four variants sit at two unexplained price pairs ($32/$68 and $28/$60), and the page's price metadata ($28.00) disagrees with its displayed price ($32).
The shopper's question, answered from the store's own data: "How much does The Mug hold — and is it your biggest mug?" An assistant can answer only half. The Big Mug's page states 14 oz, so "your biggest holds 14 oz" works. The Mug's own capacity is stated nowhere in its page text — so the question shoppers will ask about the 1,731-review product cannot be answered from eastfork.com. (Product pages checked 20 October 2026; a text search of The Mug's page returns no capacity figure.)
The change list a client would receive
The audit produced a frozen, 16-line change list — every line a before/after traced to a finding, with nothing applied before written approval. Four of the sixteen lines:
Line 2 — The Mug · specifications
Before
No capacity or dimensions stated anywhere in the page text (verified: no "oz" on the page).
Proposed
Add the mug's capacity and dimensions in specification format. Exact figures supplied by the merchant from its spec sheet — never invented.
Why
The flagship product's most-asked fact is missing; its sibling Big Mug states 14 oz, proving the format.
Line 11 — Collections · public list
Before
Internal collections publicly listed in the sitemap: employee discounts, point-of-sale packs, a referral-credit collection, a holiday test collection.
Proposed
Unpublish all five from the public catalogue (kept available internally, as Shopify allows).
Why
Employee and test collections are not shopper categories; in public catalog data they confuse automated readers.
Line 14 — FAQ page · content
Before
The FAQ opens with a free-shipping promotion that ran 14–18 May 2026 — expired five months ago.
Proposed
Remove the expired promotion block; current shipping terms live on the shipping policy (Line 13).
Why
An expired promotion presented as current makes every shipping answer the store gives unreliable.
Line 16 — The Candle in Tender Earth · variant & price data
Before
Four variants at two unexplained price pairs ($32/$68, $28/$60); price metadata ($28.00) disagrees with the displayed price ($32).
Proposed
Label what the lower-priced pair is (e.g. Seconds, if that is what it is — merchant to confirm), or correct the variant data; align displayed and metadata prices.
Why
Unexplained dual pricing is exactly the contradictory data that gets listings limited and shoppers misinformed.
…plus 12 further lines covering product titles, specification blocks, alt text, duplicate collections, and policy wording.
The approval gate, demonstrated
In this rehearsal, we role-played the merchant and approved the odd-numbered lines only: 1, 3, 5, 7, 9, 11, 13, 15. The resulting fix log contains exactly 8 work orders — 8 approved, 8 applied, 0 unapproved. Partial approval changes the scope, never the discipline: because Line 2 was not approved, The Mug's capacity stays unpublished in the rehearsal outcome — recorded as not done, in writing, exactly as it would be for a client who declined it.
That is the whole control system: no approval, no change; approved lines only; a fix log you can reconcile line by line against your own approval. There is no step where work happens off-list.
What we could not verify publicly
Product identifiers (12 points, excluded). No GTIN/UPC/EAN appears on any public page checked. Whether identifiers are populated in Merchant Center is visible only inside that account. Variant naming (10 points, excluded). Variant counts and pricing are public; the option names and values themselves are not exposed in the public text layer, so we will not score what we cannot read. Feed approval status (5 of 10 points, excluded) and Shopify Knowledge Base population likewise live inside the merchant's accounts. In a client engagement, these are completed from a product export or collaborator access — and the price band is confirmed from the exact SKU count, in writing, before work starts.
Your store, next. The free audit is this document, produced for your catalog: read-only, from your public pages, delivered within 2 business days — yours to keep whether or not you buy the fix.
Request yours.