Start with the buyer problem a recommendation is meant to solve, then inspect the evidence and proposed action. A missing product dimension, an unanswered purchase question and an opportunity for independent coverage require different work.
Lantern's Opportunities view groups that work into three lanes: Fix your site, Publish content and Reach out.
Choose the lane by the action
Lane | Work it groups | What to inspect first |
|---|---|---|
Fix your site | Changes to pages and product information you control | The affected page, current fact and proposed correction |
Publish content | Content drafts that address a buyer question | The audience, source evidence and missing answer |
Reach out | Drafted messages for relevant independent publishers or contacts | The recipient's relevance and the evidence worth sharing |
The lane organizes the action; it does not guarantee a particular visibility lift or publication outcome.
Open the item before acting
Read the evidence attached to the opportunity. Confirm that it refers to the right brand, product, variant and page, and that the issue is still present. A recommendation based on an old price or discontinued item may need to be dismissed or revised.
Use the action offered for that item. When it generates a draft, review the draft in Content Studio. Drafting, approving and publishing are distinct steps: verify which one the interface is performing instead of assuming that starting the work makes it live.
For outreach, review the message and its claims before sending through the intended process. A prepared pitch does not mean the publisher has agreed to cover the subject.
Prioritize a verifiable fix
A useful first action has a clear problem, reliable supporting evidence, a specific destination and an observable completion check.
For an illustrative bottle-page issue, the missing fact might be base diameter. The completion check is that the correct measurement appears on the visible page and agrees with relevant product data. The later question is whether assistants use it correctly. Those are separate checks.
Prioritize work with a clear buyer benefit and reliable supporting evidence. Consider the importance of the issue, the effort needed to fix it and whether the result can be checked on the public page.
Review the proposed content
Check specifications, prices, availability, sources and buyer constraints. Remove unsupported promises and make examples clearly illustrative. Structured data should describe the visible content, and a product-specific claim should have evidence for that product.
For comparison drafts, use equivalent criteria for each option and state the tradeoffs. For independent coverage, retain the distinction between a factual correction request and an editorial opinion you do not control.
The product-page guide and comparison guide provide concrete review patterns.
Verify publication and keep a record
After the supported publication or application step, open the actual destination. Confirm the right page changed, the intended content is visible and important links or purchase steps still work.
Record the before state, after state, publication time and expected buyer benefit. An approved draft is not evidence that the live page changed. Keep the record with the task so a later result can be interpreted against the actual intervention date.
Measure observed outcomes
Repeat the relevant checks under comparable prompts, assistants and settings. Keep the observation window and sample sizes visible. If a result improves, describe it as an observed change unless the study design supports attributing it to the intervention.
Choose the review horizon to suit the change and available evidence. There is no universal seven-day window that proves a recommendation worked. Use the weekly review guide to check coverage and sample changes before drawing a conclusion.