Which information should AI menu product comparisons use?
Assess menu assistant comparisons using prices, descriptions and recorded nutrition fields. Build comparisons that keep missing information visible.
A guest deciding between two products may want to see concrete differences more than a long recommendation. Prices, product descriptions and nutrition values verified by the business can support that comparison. Not every field must be filled; explicitly showing that information is unknown provides a more reliable basis for a decision than a table completed with guesses.
Contents
First establish which products are being compared
When a guest says 'Compare these', the products most recently shown in the conversation matter. Bringing an unrelated main course into a comparison after showing two desserts is a context error. Test two approaches: name products explicitly and refer to cards already displayed. Check that the resulting names match the expected products, paying particular attention to different portions with similar names.
If a product is no longer available, a new comparison should account for its current status. An old chat card remaining visible does not mean the product is still served. Orvunai's comparison experience uses recorded information for published, available products. Keeping the menu current is therefore also the foundation for comparisons.
Compare fields that mean the same thing
A price comparison is clear when currency and product quantity have the same scope. If one card describes a single portion and another a serving for two, that should be visible in the recorded description. If the system has no portion information, do not infer the number of people from a product name. Explain what is being compared before reaching a conclusion such as 'better value'.
When nutrition values are recorded, compare the units and values shown for each product with the menu record. An empty protein field for one product and a filled field for another does not prove that the first has less protein. The example below is a presentation outline showing how to keep missing information visible, not actual product data.
| Criterion | Product A | Product B |
|---|---|---|
| Price | Recorded amount | Recorded amount |
| Protein | Recorded value | Information not entered |
| Serving description | Verified text | Verified text |
Do not jump from a measurement to a personal judgment
A lower price does not make a product more suitable for every guest. Likewise, a high or low nutrition value should not become personal health advice by itself. The assistant's comparison can reveal recorded differences; special personal requirements or unknown preparation details may need support from the business's team. The conclusion should preserve the limits of the criterion used.
Taste, how filling a dish is and suitability for sharing between two people should not be presented as automatic conclusions without a clear basis in the business's records. A product looking large in a photograph or having a high price does not verify those judgments. Adding genuinely verified serving information to the menu description is more useful than generating speculative commentary later.
Check what happens after the comparison too
After seeing the table, a guest might ask 'Is there a cheaper alternative?' Check that the new results fit the relevant product group and any budget context. If the guest does not want the compared products shown again, understanding that request correctly matters. Record the initial question, compared products and expected effect of the follow-up message together in the test notes.
Two differently priced products, one product with a missing nutrition field and two similarly named options can form a sufficient starting set for a business check. They show both the normal flow and missing-information cases. When reading the results, assess field sources, product identity and freshness before the table's visual appeal.
Explore the relevant Orvunai product guide to put these steps into practice with your own menu.
Explore AI restaurant menu and Luca AI waiter