Example analysis
Good advice explains why a product fits and when another is better. Here is how to make those criteria verifiable on product and comparison pages.
In this example analysis, Northstar represents an outdoor shop whose team answers many questions during sales advice.
The website should explain the same decision: what is the product suitable for, what are its limits, and which alternative fits?
These are the first points to check in this buying path.
Staff ask about use, size, and features. The product page lists specifications without explaining their benefits or limits.
“Water-resistant” and “waterproof” appear without explanation or evidence. That difference matters for a question about rain.
“Seasonally available” gives neither a restock date nor an available alternative that meets the same needs.
The buying task
Which daypack suits an easy walk in the rain, and which alternative is available?
Each step gives the team a concrete task and an output it can check.
With the team, I collect common questions and the criteria behind a recommendation. Reasons against a product matter as well as its strengths.
For your team
Output: buyer questions with uses, requirements, and relevant comparison criteria.
Short paragraphs explain suitability and limits. Claims about weather protection or materials refer to the available product evidence.
For your team
Output: drafts for “Suitable for”, “Consider before buying”, and their supporting sources.
A comparison page uses the same criteria for every model. Stock and confirmed delivery dates help buyers choose an alternative.
For your team
Output: a comparison that clearly shows differences, missing facts, and available options.
The same questions run before and after the change. I record mentions, reasons, sources, and factual errors across several runs.
For your team
Output: an answer comparison showing what the AI uses correctly and what remains missing.
The same task, before and after the change. These criteria guide the review.
Criteria for the next test
Record the answer and source for each point. Anything unresolved becomes the next finding.
The recommendation refers to the task requirements and relevant product evidence.
Limits of weather protection remain clear; missing evidence is not invented.
The alternative meets the key criteria and is available according to the shop.
Mentions, sources, and errors are recorded for each run and can be compared.
You receive clear criteria for advice and comparison, plus questions for the next measurement. Better content makes recommendations verifiable; measured answers show whether AI uses it.
Example analysis
Example analysis
Which part of your shop should I check?
Check my shop