GEO RESEARCH SERIES — WEEK 1
What Sources Do AI Systems Use for Carry-On Luggage Recommendations?
A 50-answer observational study across ChatGPT, Perplexity, Gemini, Google AI Overview, and Google AI Mode.
The headline finding
The brands most often recommended were usually not the sources most often cited.
Travelpro appeared as a recommendation in 48 of 50 answers, yet the most-cited sources were independent buying guides. OutdoorGearLab’s carry-on guide was cited in 23 of 50 answers and was the only source cited by all five systems.
Research question
What observable characteristics are common among sources AI systems use when recommending carry-on luggage?
This study was observational only. It records what appeared under the stated conditions. It does not claim to reveal proprietary ranking factors, authority scores, retrieval logic, or causation.
Methodology
On November 10, 2026, the same 10 recommendation prompts were run across five AI answer surfaces: ChatGPT Search, Perplexity, Google Gemini, Google AI Overview, and Google AI Mode. That produced 50 answer sets.
Fresh conversations or searches were used for each prompt, with no prior luggage-brand context. Four platforms were tested signed out. Perplexity was tested signed in to a free account after signed-out access was blocked; that difference is treated as a limitation.
Recommended brands and products were recorded separately from cited or linked sources. Across the 50 answers, 77 unique cited or linked source items were identified. Forty-five could be inspected closely enough to receive a 20-point Source Characteristics Coding Score; 32 could not be reliably inspected and were left unscored.
The 10 prompts
1. What is the best carry-on luggage for most travelers?
2. What is the best carry-on luggage under $150?
3. What is the best lightweight carry-on luggage?
4. What is the best carry-on luggage for frequent flyers?
5. What is the best carry-on luggage for international travel?
6. What is the most durable carry-on luggage?
7. What carry-on luggage has the best wheels and maneuverability?
8. What is the best carry-on luggage for maximizing packing space?
9. What is the best carry-on luggage for business travelers?
10. What carry-on luggage offers the best value for the money?
Headline results
23/50
OutdoorGearLab cited
48/50
Travelpro recommended
15.9
Average coding score /20
Most-cited sources
OutdoorGearLab: 23 answer sets
NYT Wirecutter: 15
Pack Hacker: 8
Business Insider: 7
Travel+Leisure: 5
Forbes Vetted: 5
Most-recommended brands
Travelpro: 48 of 50 answers
Samsonite: 26
Away: 25
Briggs & Riley: 20
Monos: 16
Observed patterns
1. One buying guide anchored all five platforms. OutdoorGearLab’s carry-on guide was cited in 23 of 50 answers and was the only source cited by every platform.
2. Recommendations and citations were different outcomes. Travelpro dominated recommendations, while independent buying guides dominated citations.
3. Best-of guides with firsthand testing dominated the coded source set. Twenty-nine of the 45 scored pages were best-of guides, and the five most-cited sources all described hands-on testing programs.
4. Publisher and specialist pages scored higher than official brand pages in this sample. Publisher-media pages averaged 16.8/20, specialist-review sites 16.3, and official brand pages 14.4.
5. Freshness was visible among the most-cited guides. The top-cited guides displayed 2026 titles or update dates.
6. Price information was inconsistent. The same Samsonite Freeform suitcase appeared at about $270 in one Perplexity answer and about $130–$150 in another on the same day.
What this may suggest
The pattern is consistent with independently tested buying guides serving as an evidence layer for product-recommendation answers, while official brand pages are more often used for product facts such as dimensions, specifications, or price. That is an interpretation, not a demonstrated ranking rule.
The gap between Travelpro’s recommendation frequency and its citation frequency also suggests that recommendation visibility and citation visibility should be measured separately. A brand can be the consensus recommendation without being the source used to support that recommendation.
Limitations
This was a small, single-category, single-day study. AI answers vary between runs. Perplexity was tested signed in while the other four systems were signed out. Gemini used its lightweight Flash-Lite model. Thirty-two of the 77 unique source items could not be inspected closely enough to score, and 23 of the 45 coding scores were provisional because some publisher pages blocked automated fetching.
Nothing in this study establishes why a source was selected, and correlation in the coded characteristics does not establish causation.
Week 1 takeaway
In this carry-on luggage test, the sources most often cited were not usually the brands being recommended. Independent, current, firsthand-tested buying guides appeared repeatedly as evidence sources.
That is an observed pattern from this dataset—not proof that those characteristics cause citation.
