Study data
Explore the full dataset
This page shows the same dataset we used for the ranking and the analyses, so you can draw your own conclusions instead of just trusting ours. It includes the data quality notice because it's essential to correctly interpret any figure below.
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.
- Friday · Afternoon · DeepSeek (API): 129 of 129 calls with no capturable text
- Saturday · Morning · OpenAI, without web search: 43 of 44 calls with no capturable text
- Sunday · Midday · OpenAI, without web search: 43 of 44 calls with no capturable text
This data cut, in numbers
Queries generated
596
Runs launched
514
Succeeded
451
Failed
61
With capturable text
236 / 451
Mentions extracted
1278
Distinct restaurants
498
Measurement window
2026-09-11 · 3d
Queries by variable
Each query crosses several of these variables at once (see the methodology). These tables show how many queries and mentions exist per individual value.
By cuisine
- Italian and pizzeria55 queries · 118 mentions
- Mexican54 queries · 104 mentions
- Asian53 queries · 117 mentions
- Restaurants in general51 queries · 190 mentions
- Valencian rice dishes49 queries · 148 mentions
- Japanese and sushi48 queries · 108 mentions
- Tapas48 queries · 175 mentions
- Seafood47 queries · 146 mentions
- Fine dining46 queries · 172 mentions
By search intent
- Future booking113 queries · 325 mentions
- Discovery113 queries · 372 mentions
- Immediate booking113 queries · 297 mentions
- Situational query112 queries · 284 mentions
By budget
- Unspecified232 queries · 698 mentions
- High77 queries · 221 mentions
- Medium72 queries · 206 mentions
- Low cost70 queries · 153 mentions
By language
- Spanish293 queries · 819 mentions
- English105 queries · 306 mentions
- Valencian53 queries · 153 mentions
By Dénia zone
- Dénia in general195 queries · 493 mentions
- The beach55 queries · 178 mentions
- Les Rotes54 queries · 182 mentions
- The port54 queries · 181 mentions
- The old town49 queries · 101 mentions
- Les Marines44 queries · 143 mentions
By AI assistant
- DeepSeek (API)306 queries (177 with capturable text) · 1052 mentions
- OpenAI, without web search88 queries (2 with capturable text) · 7 mentions
- OpenAI, with web search57 queries (57 with capturable text) · 219 mentions
By day and time slot
- Saturday · Morning195 queries (152 with capturable text) · 871 mentions
- Friday · Afternoon142 queries (13 with capturable text) · 10 mentions
- Sunday · Midday114 queries (71 with capturable text) · 397 mentions
Definitions
- Query (call)
- A question sent to an AI assistant that completed its run. Includes queries with no capturable response text (see the data quality notice): those count as a query but contribute no mentions.
- Mention
- An identifiable reference to a specific Dénia restaurant within an assistant's response. A single query can produce several mentions.
- Distinct restaurants
- The number of distinct restaurant names after the system's deterministic normalization (the same name always normalizes the same way, but the process isn't infallible).
- Average position
- The average order in which a restaurant appears within the responses that mention it, not a search-engine ranking position.
Dataset download
The file includes every query (with its cuisine, intent, language, budget, zone, time slot, and assistant), every extracted mention, and the aggregates that power this page and the ranking.
It doesn't include cost, tokens used, or each run's internal id: that's operational data, not part of the research dataset.
Compare your own conclusions against the ranking
This dataset is the same one we used to build the visibility ranking and the per-variable analyses. If something doesn't add up, we want to know.