Analysis by intent
What changes by what the person needs
The experiment's four intents (immediate booking, future booking, discovery, and situational query) were split almost exactly evenly, 112-113 queries each. What changes is how much the assistant answers and who it mentions.
Summary
Discovery is the intent with the most mentions (372) and the most distinct restaurants cited (198), consistent with being the most open-ended question ("what do you recommend?" versus "I want to book tonight"). Immediate booking and situational queries concentrate responses more heavily on Restaurante Mena and El Pegolí, which top both lists.
Data quality notice
Of the 451 calls in this cut, 236 returned capturable text. The gap isn't spread randomly: it's concentrated in three specific day, time-slot, and assistant combinations (detail below) totaling 217 calls, of which only 2 returned text. Those calls completed their run and used tokens, but the text couldn't be recovered due to a capture-pipeline gap, not because the assistant recommended nothing. Every figure on this page counts them toward the total queries, but they can't contribute mentions.
Queries and mentions by intent
The four intents have an almost identical query volume by design: mention differences reflect how the assistant answers each type of question, not an imbalance in how often each was asked.
Future booking
113 queries · 325 mentions · 203 distinct restaurants
Most mentioned
- Quique Dacosta (16)
- Restaurante Mena (14)
- La Seu (10)
Discovery
113 queries · 372 mentions · 198 distinct restaurants
Most mentioned
- Restaurante Mena (18)
- Quique Dacosta (17)
- Casa Federico (17)
Immediate booking
113 queries · 297 mentions · 171 distinct restaurants
Most mentioned
- Restaurante Mena (18)
- El Pegolí (14)
- La Seu (13)
Situational query
112 queries · 284 mentions · 145 distinct restaurants
Most mentioned
- Restaurante Mena (27)
- El Pegolí (20)
- Casa Federico (17)
What we can't claim from this
- Discovery generating more mentions is a pattern in this cut with these specific assistants, not a general rule about how "AI" answers open-ended questions.
- We don't measure whether an immediate-booking recommendation ended in an actual booking: only which restaurants the assistant cites in its response text.
- Nearly half of this cut's queries have no capturable response text (see the data quality notice): all four intents are affected similarly, but the absolute mention counts are a floor, not the real total.
Frequently asked questions
Before drawing conclusions by intent
Why does "discovery" generate so many more mentions than the other intents?
It's the most open-ended of the four questions ("what do you recommend?" versus "I want to book tonight"), which gives the assistant more room to cite several restaurants instead of just one. It's a pattern in this cut with these specific assistants, not a general rule about how "AI" answers open-ended questions.
If I ask for an immediate booking, does the assistant recommend the restaurant with the most real availability?
We don't know. This experiment doesn't measure whether an immediate-booking recommendation ended in an actual reservation: it only records which restaurants the assistant cites in its response text, not their availability or whether the customer actually booked.
Keep exploring the dataset
This is one of six dimensions we analyze. The rest of the variables, and the full dataset to explore on your own, are on the data page.