These three gaps are prioritized because they affect the information a shopping agent can retrieve. The example makes no claim about actual checkout success.
01
Demo gap · Not detected in sample
An offer has no readable price
A shopping agent cannot retrieve a single price from this example product’s Offer.
Evidence, repair and verification
Evidence
Offer.price omitted
Source: fictional product fixture · Day 1
What and where to change
Map the existing approved catalog price into the public Offer. Match the same product and currency. Do not invent a price.
Location: Product-page Offer generator. Exact file depends on the real stack.
Verify: inspect the same product and mode, then recheck. Any change to product facts, policies or customer-visible text requires merchant approval.
02
Demo gap · Not detected in sample
Stock state is missing from the offer
The example page does not provide a machine-readable availability declaration.
Evidence, repair and verification
Evidence
Offer.availability omitted
Source: fictional product fixture · Day 1
What and where to change
Expose the merchant’s actual stock state with the supported vocabulary. A readable declaration does not prove warehouse accuracy.
Location: Inventory-to-storefront mapping. Exact file depends on the real stack.
Verify: inspect the same product and mode, then recheck. Any change to product facts, policies or customer-visible text requires merchant approval.
03
Demo gap · Not detected in sample
The product cannot explain its distinguishing details
The supplied example lacks the description needed to compare this item.
Evidence, repair and verification
Evidence
Product.description omitted
Source: fictional product fixture · Day 1
What and where to change
Reuse approved product facts. Stop for merchant input if the description or facts are missing.
Location: Product content template. Exact file depends on the real stack.
Verify: inspect the same product and mode, then recheck. Any change to product facts, policies or customer-visible text requires merchant approval.
Live Proof format / Illustrative values
A rate needs a denominator.
These numbers demonstrate the presentation. No live merchant was requested for this report.
Product discovery
Find → retrieve → identify one product
3 / 3 successful
100% in these illustrative attempts
Price + variant retrieval
Failure step: resolve a single offer
2 / 3 successful
67% in these illustrative attempts
Inventory declaration
Failure step: retrieve declared stock state
2 / 3 successful
67% in these illustrative attempts
Cart reachability
No cart interaction, order or payment
Manual verification required
Excluded from the denominator
Policy retrieval
Find → retrieve one policy page
3 / 3 successful
100% in these illustrative attempts
Even a real 3/3 would describe three bounded retrieval attempts, not all agents, independent price accuracy or future sales.
Before / After / Fictional
3 improved. 1 unchanged. 0 regressions.
Same product identity. Same check method. Fresh evidence after the illustrated repair.
Measured signal
Before
After
Result
price
Offer.price omitted
Offer.price mapped to the approved catalog value
Improved in example
availability
Offer.availability omitted
Availability declaration now exposed
Improved in example
description
Product.description omitted
Existing merchant-approved description mapped
Improved in example
Cart execution
Not tested
Not tested
Unchanged safety boundary
The real product calls an improvement only when a fresh, comparable deterministic finding moves from Needs attention to Verified. A checkbox or edited explanation does not verify a repair.
Monitoring history / Fictional
Tell me when something changes.
Meaningful alerts preserve the before and after. No-change runs do not create daily email noise.
Day 1
Baseline captured
Three example fields need attention.
Day 2
Recheck: three improvements
Approved catalog values are now exposed.
Day 5
Availability declaration disappears
Regression detected; cause unknown.
Day 5
An alert is shown in this example
What changed, why it matters, evidence and recheck action. No actual email was sent.
Day 6
The signal recovers
An observed recovery would be recorded separately.
Help after the audit
Turn the next fix into a clear brief.
Implementation reports and human review are being scoped. These buttons record interest only. No payment, booking or delivery commitment is created.
Implementation report
A proposed developer brief with prioritized repairs, evidence and a post-fix recheck plan.
Human review
A proposed review of uncertain findings, possible false positives and the implementation order.
Ongoing monitoring
Watch meaningful changes to public evidence, with clear scope and a way to stop notifications.
Evidence stays attached. Real reports include sources, dates, scope, limitations and safe next steps. This fictional example is excluded from product-outcome measurements and research.