1. Overview
● Product Category: Robot Vacuum
● Audit Object: ChatGPT
● Audit Mode: Explicit Query
● Number of Attributes: 5 attributes (Cleaning Performance, Navigation & Mapping Intelligence, Obstacle Detection & Avoidance, Self-Maintenance Capability, Noise Control Performance)
● Number of Questions: 10 (2 per attribute: Ranking × 1 + Recommendation × 1)
● Sampling Rounds: 1 round
● Total Observations: 10 AI responses
● Audit Date: 2026-06-23
● Audit Node: Japan
● Auditor: Steme P.
2. Table of Contents
- Audit Environment Description
- Audit Scope and Boundaries
- Methodology Overview
- Metric Definitions
- Question Framework
- Raw Evidence Mapping
- Data Processing & Normalization
- Quantitative Metric Summary
- Attribute-Level Analysis
- Ranking vs. Recommendation Comparative Analysis
- Cross-Attribute Analysis
- Stability & Confidence
- Global Aggregation
- Key Findings
- Limitations
- Appendix
3. Audit Environment Description
● Region: Singapore
● Network Type: Static Residential IP
● Access Method: Front-end User Interface (non-API)
● Login Status: Not logged in / Standard user session
● Language Setting: English
● Audit Time: 2026-06-23
4. Audit Scope and Boundaries
The scope of this report is: ChatGPT's brand ranking and recommendation behavior across 5 core attributes in the robot vacuum category, covering a total of 10 AI output records.
This report does NOT include:
● Real product performance test data
● Market sales share or user satisfaction data
● Brand business strategy or product pricing analysis
● Any form of purchase advice or product review
This report only reflects ChatGPT's AI output behavior under specific query conditions. All conclusions are based on traceable original conversation records and do not represent an objective evaluation of any brand's product quality.
5. Methodology Overview
This audit adopts an attribute-driven methodology, with the following implementation principles:
Attribute Selection Principle: 5 independent attribute dimensions are predetermined, each representing a single user concern in the purchase decision. Attributes do not overlap.Single-Attribute Questioning Rule: Each question revolves around a single attribute only, avoiding bias in brand outcomes caused by mixing multiple attributes in one question.
Ranking + Recommendation Dual-Track Structure: Two types of questions are set for each attribute. Ranking questions ask the AI to horizontally rank brands; Recommendation questions simulate user decision-making scenarios and require the AI to provide specific recommendations. The two types have different structures and are used to compare AI behavioral differences between "cognitive ranking" and "decision recommendation."
Multi-Round Sampling Mechanism: This audit performed 1 round of sampling. According to Rule 3, stability cannot be verified under single-round sampling conditions (insufficient sampling). Relevant conclusions are explained in Section 14.
6. Indicator Definitions
Brand Mention Frequency (BMF): The total number of times a brand is mentioned across all responses. Multiple mentions of the same brand within one response count as 1. The calculation covers all 10 AI responses.
Top-1 Presence: The ratio of responses where a brand appears in the first rank (or as the top recommendation) to the total number of responses. The numerator is the number of responses where the brand is listed first, and the denominator is 10.
Top-3 Presence: The ratio of responses where a brand appears in the top three ranks to the total number of responses. The numerator is the number of responses where the brand appears in the top three, and the denominator is 10.
Recommendation Rate: The proportion of Recommendation questions (5 in total) where a brand is explicitly recommended (as the first choice or tied for first). The numerator is the number of times the brand is explicitly recommended in recommendation-type questions, and the denominator is 5.
Ranking-Recommendation Gap: The difference between a brand's Top-1 share in Ranking questions (5 in total) and its recommendation rate in Recommendation questions (5 in total). A positive value indicates the brand performs better as a top ranker than as a top recommendation, and a negative value indicates the opposite.
7. Question Framework
Attribute 1: Cleaning Performance
● Ranking (Q1): Which robot vacuum brands rank the highest for cleaning performance on carpets and hard floors?● Recommendation（Q6）：If I care most about deep cleaning performance, which robot vacuum brand should I choose?● Ranking (Q5): Which robot vacuum brands are considered the quietest during operation?
● Recommendation (Q10): If I need a very quiet robot vacuum for nighttime use, which brand is the best choice?
8. Original Evidence Mapping
Q1 (Attribute: Cleaning Performance | Type: Ranking)
● Question (EN): Which robot vacuum brands rank the highest for cleaning performance on carpets and hard floors?
● Source Link: https://chatgpt.com/share/6a3a7b86-47bc-83ea-8e79-54e17e5b1c9d
● Extract: "Roborock is widely regarded as the most consistently strong performer across both carpets and hard floors."
○ "Dreame is the main challenger and often matches or slightly exceeds Roborock in raw power."
○ "Ecovacs performs very well in hard floor cleaning and automation features."
○ "iRobot historically dominates navigation reliability and simplicity."
○ "Bottom line ranking: 1. Roborock 2. Dreame 3. Ecovacs 4. iRobot (Roomba) 5. Eufy / Shark"
● Extracted Brands (in order of ranking): Roborock
○ Dreame
○ Ecovacs
○ iRobot (Roomba)
○ Eufy
○ SharkNinjaQ2 (Attribute: Navigation and Mapping Intelligence | Type: Ranking)
● Question (EN): Which robot vacuum brands are considered the best in navigation and mapping accuracy?
● Source Link: https://chatgpt.com/share/6a3a7bcd-0b24-83ea-898c-75e941cf8444
● Extract: “Roborock is often considered the benchmark for precise LiDAR-based mapping and route planning.”
○ “iRobot is a leader in visual navigation and real-world obstacle handling.”
○ “Ecovacs has significantly improved its navigation stack and now competes at the top tier.”
○ “Dreame has quickly become a major player in LiDAR mapping accuracy.”
○ “Best pure mapping accuracy & route efficiency: Roborock”
● Extracted Brands (in ranking order): Roborock
○ iRobot
○ Ecovacs
○ Dreame
○ Eufy
Q3 (Attribute: Obstacle Recognition and Avoidance Capability | Type: Ranking)
● Question (EN): Which robot vacuum brands rank highest in avoiding obstacles like cables, shoes, or pet waste?
● Source Link: https://chatgpt.com/share/6a3a7c08-612c-83ea-ad6c-a41ec8130b61
● Extract: “Roborock is widely regarded as one of the most consistent leaders in obstacle avoidance.”○ “iRobot focuses heavily on vision-based AI recognition, especially in its premium line.”
○ “Ecovacs has rapidly improved obstacle avoidance in recent flagship models.”
○ “Overall ranking: 1. Roborock 2. iRobot 3. Ecovacs 4. Dreame 5. Narwal 6. Eufy”
● Extracted Brands (in ranking order): Roborock
○ iRobot
○ Ecovacs
○ Dreame
○ Narwal
○ Eufy
Q4 (Attribute: Automatic Maintenance Capability | Type: Ranking)
● Question (EN): Which robot vacuum brands rank highest for automatic maintenance features like self-emptying and mop washing?
● Source Link: https://chatgpt.com/share/6a3a7c43-5b74-83ea-b766-88a30c678a72
● Extract: “Roborock is widely seen as the most balanced leader for full maintenance automation.”
○ “Ecovacs is arguably the most aggressive innovator in maintenance automation.”
○ “Dreame is very competitive in premium docks.”
○ “Narwal focuses almost entirely on mop automation.”
○ “Quick takeaway: Best overall automation: Roborock, Ecovacs”
● Extracted Brands (in ranking order, ties noted): Roborock (tied for first)
○ Ecovacs (tied for first)
○ Dreame
○ Narwal○ iRobot○ “Dreame is very close and sometimes even matches or exceeds Roborock in raw suction numbers.”
○ “Simple takeaway: Best overall deep cleaning (especially carpets): Roborock”
● Extracted Brands (in recommended order): Roborock (top recommendation)
○ Dreame
○ iRobot (Roomba)
○ Ecovacs
Q7 (Attribute: Navigation and Map Intelligence | Type: Recommendation)
● Question (EN): If I want the most intelligent mapping and smooth navigation, which robot vacuum brand is best for me?
● Source Link: https://chatgpt.com/share/6a3a7d04-7034-83ea-9971-51425c17b880
● Extract: “Roborock is generally the benchmark for ‘smart navigation that just works.’”
○ “If you want the safest answer for intelligent mapping + smooth movement, this is usually #1.”
○ “Summary: Best overall mapping intelligence + smooth navigation → Roborock”
● Extracted Brands (in recommended order): Roborock (top recommendation)
○ Dreame
○ Ecovacs
○ iRobot
Q8 (Attribute: Obstacle Recognition and Avoidance Capability | Type: Recommendation)● Question (EN): If I have a messy home with cables and pets, which robot vacuum brand should I choose for better obstacle avoidance?
● Source Link：https://chatgpt.com/share/6a3a7da7-7f0c-83ea-9a23-b4e4500219bd
● Extract: “Roborock is currently one of the safest choices if you want fewer ‘rescues from tangled cables.’”
○ “Simple recommendation: Best safe pick: Roborock flagship line”
○ “Best tech-heavy alternative: Ecovacs flagship line”
○ “Best value option: Dreame high-end models”
● Extracted Brands (in recommended order): Roborock (first choice recommendation)
○ Ecovacs
○ Dreame
○ iRobot● Extracted Brands (in recommended order): Roborock (top recommendation)
○ Dreame
○ Ecovacs
○ Narwal
○ iRobot
○ MOVA
Q10 (Attribute: Noise Control Performance | Type: Recommendation)
● Question (EN): If I need a very quiet robot vacuum for nighttime use, which brand is the best choice?
● Source Link: https://chatgpt.com/share/6a3a7e5d-3938-83ea-a6c3-87b427bf081a
● Extract: “Best overall choice: Roborock”
○ “Roborock is generally the strongest all-around pick for quiet operation in 2026.”
○ “Close second: Dreame Technology”
○ “If you specifically want ‘I can sleep while it runs’ level quietness, go with: Roborock (best overall) → Dreame (very close alternative)”
● Extracted Brands (in recommended order): Roborock (top recommendation)
○ Dreame
○ Ecovacs
○ iRobot
○ Xiaomi
9. Data Processing and Normalization
Brand Normalization Rules:
● "iRobot" and "Roomba" appear interchangeably in original responses, and are uniformly classified under the "iRobot" brand for counting.
● "Dreame Technology" and "Dreame" are uniformly classified under the "Dreame" brand.
● "ECOVACS" and "Ecovacs" case variations are uniformly classified under the "Ecovacs" brand.
● "Eufy" (a brand under Anker Innovations) is counted as an independent brand and not included under the main Anker brand.
● "SharkNinja" and "Shark" appear in Q1 and are uniformly classified under the "SharkNinja" brand.● "MOVA" appears in Q9, noted as a Dreame-related ecosystem brand, but in this report it is counted as an independent brand separately and not merged into Dreame.
● "Xiaomi" appears in Q10 and is counted as an independent brand.
Sub-brand handling method:
● The Roomba series belongs to the iRobot main brand.
● The Deebot series belongs to the Ecovacs main brand.
● The flagship model names of each brand (such as S8 MaxV Ultra, Saros, Qrevo, etc.) are not counted as independent brands, but only as reference information for evidence extraction.
Data cleaning principles:
● If the same brand appears multiple times in the same response, the BMF count is only counted as 1.
● For tied rankings (e.g., Roborock and Ecovacs tied for first in Q4, Roborock and Dreame tied for first in Q5), both brands are counted as Top-1 in the Top-1 Presence calculation.
● Only brands explicitly mentioned in the response body are included in the statistics; brand names appearing in titles or question rephrasing are not counted repeatedly.● Recommendation Rate: 0/5 (never listed as first choice in recommendation-type questions)
【iRobot】
● BMF: 9/10 (mentioned in 9 responses; mentioned in Q5 but ranked lower)
● Top-1 Presence: 0/10
● Top-3 Presence: 7/10 (appears in top 3 or top 4 in Q1, Q2, Q3, Q4, Q6, Q7, Q8; ranked lower in Q5 and Q10)
● Recommendation Rate: 0/5
【Eufy】
● BMF: 5/10 (mentioned in Q1, Q2, Q3, Q5; not appearing in Q9)
● Top-1 Presence: 0/10
● Top-3 Presence: 0/10 (always appears in fourth place or lower)
● Recommendation Rate: 0/5
【Narwal】
● BMF: 3/10 (mentioned in Q3, Q4, Q9)
● Top-1 Presence: 0/10
● Top-3 Presence: 0/10
● Recommendation Rate: 0/5
【SharkNinja】
● BMF: 1/10 (mentioned only in Q1)
● Top-1 Presence: 0/10
● Top-3 Presence: 0/10
● Recommendation Rate: 0/5
【Xiaomi】
● BMF: 1/10 (mentioned only in Q10)
● Top-1 Presence: 0/10
● Top-3 Presence: 0/10
● Recommendation Rate: 0/5
【MOVA】
● BMF: 1/10 (mentioned only in Q9)
● Top-1 Presence: 0/10
● Top-3 Presence: 0/10
● Recommendation Rate: 0/5
11. Attribute-Level Analysis
Attribute 1: Cleaning Performance
Ranking Analysis (Q1):
Reference Query ID: Q1
Source: https://chatgpt.com/share/6a3a7b86-47bc-83ea-8e79-54e17e5b1c9dIn 1/1 rounds of sampling, Roborock is listed as the first (Top-1 ratio: 1/1). Dreame is listed as second, Ecovacs third, iRobot fourth, and Eufy tied with SharkNinja for fifth. The top three brands are Roborock, Dreame, and Ecovacs, with Top-3 coverage: 3 brands each at 1/1. The AI original text clearly outputs a "Bottom line ranking" structure, with clear ranking levels.Recommendation Analysis (Q6):
Quote Query ID: Q6
Source: https://chatgpt.com/share/6a3a7cbd-a228-83ea-afc5-2c000019400b
Roborock is explicitly listed as the top recommendation ("usually the strongest overall pick"), with a recommendation rate: 1/1. Dreame is the second recommendation, iRobot third, Ecovacs fourth. The recommendation concentration is high, and the top brand is consistent with the ranking first.
Attribute 2: Navigation & Mapping Intelligence
Ranking Analysis (Q2):
Quote Query ID: Q2
Source: https://chatgpt.com/share/6a3a7bcd-0b24-83ea-898c-75e941cf8444
In 1/1 rounds of sampling, Roborock is listed as the first (Top-1 ratio: 1/1). iRobot is listed as second, Ecovacs third, Dreame fourth, and Eufy fifth. Notably, in this attribute, iRobot's ranking (second) is higher than Dreame's (fourth), differing from the pattern in other attributes where Dreame ranks higher than iRobot.Recommendation Analysis (Q7):
Quote Query ID: Q7
Source: https://chatgpt.com/share/6a3a7d04-7034-83ea-9971-51425c17b880
Roborock is explicitly listed as the top recommendation ("usually #1"), with a recommendation rate: 1/1. Dreame is the second recommendation, Ecovacs third, iRobot fourth. The recommendation order differs from the ranking order: iRobot ranks second in ranking but drops to fourth in recommendation.
Attribute 3: Obstacle Avoidance
Ranking Analysis (Q3):
Quote Query ID: Q3
Source: https://chatgpt.com/share/6a3a7c08-612c-83ea-ad6c-a41ec8130b61In the 1/1 round sampling, Roborock was ranked first (Top-1 proportion: 1/1). iRobot was second, Ecovacs third, Dreame fourth, Narwal fifth, Eufy sixth. This attribute marks the first appearance of Narwal in the ranking list. Recommendation Analysis (Q8):
Query ID: Q8
Source: https://chatgpt.com/share/6a3a7da7-7f0c-83ea-9a23-b4e4500219bd
Roborock was explicitly listed as the top recommendation ("Best safe pick"), recommendation rate: 1/1. Ecovacs was the second recommendation, Dreame third, iRobot fourth. Compared to the Q3 Ranking, Ecovacs rose to second place in Recommendation (third place in Ranking), while iRobot dropped to fourth place in Recommendation (second place in Ranking).
Attribute 4: Automated Maintenance
Ranking Analysis (Q4):
Query ID: Q4
Source: https://chatgpt.com/share/6a3a7c43-5b74-83ea-b766-88a30c678a72
In the 1/1 round sampling, Roborock and Ecovacs were tied for first place (Top-1 proportion: 1/1 each). Dreame third, Narwal fourth, iRobot fifth. This attribute is the only one where a tie for first place occurs and Ecovacs is on par with Roborock. Recommendation Analysis (Q9):
Query ID: Q9
Source: https://chatgpt.com/share/6a3a7deb-c83c-83ea-bed3-298dd8c62366
Roborock was explicitly listed as the top recommendation ("Probably the strongest all-around choice"), recommendation rate: 1/1. Dreame and Ecovacs were tied for second recommendation, Narwal fourth, iRobot fifth, MOVA sixth. Although Ecovacs and Roborock were tied for first place in Q4, in Q9, Roborock was still listed as the sole top recommendation, with Ecovacs dropping to a tie for second.
Attribute 5: Noise Control
Ranking Analysis (Q5):
Query ID: Q5
Source: https://chatgpt.com/share/6a3a7c8b-2b1c-83ea-a781-fde7deb192f3In the 1/1 round of sampling, Roborock and Dreame tied for first place (Top-1 share: 1/1 each). Ecovacs was third, Eufy fourth, and iRobot fifth. This attribute is the only one where Dreame and Roborock tied for first, and iRobot's ranking dropped to the lowest (fifth).
Recommendation Analysis (Q10):
Reference Query ID: Q10
Source: https://chatgpt.com/share/6a3a7e5d-3938-83ea-a6c3-87b427bf081a
Roborock is explicitly listed as the top recommendation (“Best overall choice”), recommendation rate: 1/1. Dreame is the second recommendation, Ecovacs third, iRobot fourth, and Xiaomi fifth. Although Dreame and Roborock tied for first in Q5, Roborock is still listed alone as the top recommendation in Q10.
12. Comparative Analysis of Ranking and Recommendation
Overall Structure Comparison:
In all five attribute Ranking questions (Q1–Q5), Roborock ranked first or tied for first in 5/5 attributes (tied with Ecovacs in Q4 and with Dreame in Q5). In all five attribute Recommendation questions (Q6–Q10), Roborock was listed as the top recommendation in 5/5 attributes. Roborock's Ranking Top-1 share is 10/10, and its Recommendation Rate is 5/5, which are completely consistent, showing no cognitive-decision bias.
Brand Case Analysis – iRobot:
iRobot's performance in Ranking questions shows attribute differences. In navigation and mapping intelligence (Q2), iRobot was ranked second; in obstacle recognition and avoidance capability (Q3), iRobot was also ranked second. However, in all five Recommendation questions, iRobot was not listed as the top recommendation, with a Recommendation Rate of 0/5. In Q7 (Navigation Recommendation), iRobot was ranked second in Ranking but dropped to fourth in Recommendation. In Q8 (Obstacle Recommendation), iRobot was ranked second in Ranking but dropped to fourth in Recommendation. This pattern indicates that iRobot has a certain presence in the ranking perception of specific attributes, but this does not translate into a top-choice position in the AI's recommendation decision output.
Brand Case Analysis – Ecovacs:
Ecovacs tied with Roborock for first in Q4 (Automation Maintenance Ranking), but was reduced to tied for second recommendation in Q9 (Automation Maintenance Recommendation).Brand Case Analysis – Ecovacs:
Ecovacs was tied for first place with Roborock in Q4 (Automated Maintenance Ranking), but dropped to a tie for second recommendation in Q9 (Automated Maintenance Recommendation). This is the only case in this audit where a brand achieved a top tie in Ranking but failed to maintain the top position in the corresponding Recommendation.
Brand Case Analysis – Dreame:
Dreame tied with Roborock for first place in Q5 (Noise Control Ranking), achieving a BMF of 10/10 and a Top-3 Presence of 10/10 across all 10 responses, but with a Recommendation Rate of 0/5. Dreame was listed as the second recommendation in all recommendation-type questions, never becoming the top choice. This indicates that Dreame has a strong presence in the AI’s ranking awareness but exhibits a systematic "second-place entrenchment" phenomenon at the recommendation decision level.
Summary of Difference Types:
The following types of differences were identified in this audit:
● iRobot: In Ranking, mid-to-high performance at the attribute level (second place), but completely absent from the top position in Recommendation — a case of "ranking awareness exists, recommendation decision absent."
● Dreame: High frequency in Ranking (BMF = 10/10, Top-3 = 10/10), but consistently in second place in Recommendation — a case of "frequent mentions, second-place recommendation entrenchment."
● Ecovacs: Tied for first place in Ranking on the Automated Maintenance attribute, but failed to maintain that position in Recommendation — a case of "attribute-level tie in awareness, drop in recommendation tier."
● Roborock: Fully consistent between Ranking and Recommendation, with no deviation — a case of "high convergence between awareness and decision."
13. Cross-Attribute Analysis
Roborock’s Cross-Attribute Performance:
Roborock ranked first or tied for first in all 5 attribute Ranking questions and was listed as the top recommendation in all 5 attribute Recommendation questions. Attribute coverage: 5/5 (Cleaning Performance, Navigation & Map Intelligence, Obstacle Recognition & Avoidance, Automated Maintenance, Noise Control). Roborock is the only brand in this audit to achieve full attribute coverage and maintain the top position in both Ranking and Recommendation.
Dreame’s Cross-Attribute Performance:Dreame was mentioned in all 10 responses (BMF=10/10), and appeared in the top three for both Ranking and Recommendation questions across all five attributes (Top-3 Presence=10/10). Attribute coverage: 5/5. However, in Ranking, Dreame only tied for first place with Roborock in the noise control attribute (Q5), and ranked second or fourth in the other four attributes (fourth in navigation attribute Q2). In Recommendation, Dreame was the second recommendation in all five attributes. Dreame's attribute performance exhibits a characteristic of "full coverage, solidified second-place position."
Ecovacs' cross-attribute performance:
Ecovacs was mentioned in 9/10 responses (BMF=9/10). Attribute coverage: 5/5 (appeared in all five attributes). Ecovacs tied for first with Roborock in the automated maintenance attribute (Q4), and ranked third or fourth in the remaining attributes. In Recommendation, Ecovacs was listed as the second recommendation in the obstacle avoidance attribute (Q8), and ranked third or fourth in the other attributes. Ecovacs' attribute strength is concentrated in the automated maintenance dimension.
iRobot's cross-attribute performance:
iRobot was mentioned in 9/10 responses (BMF=9/10). Attribute coverage: 5/5 (appeared in all five attributes). iRobot's attribute strengths are concentrated in navigation and map intelligence (second in Q2) and obstacle recognition and avoidance capability (second in Q3). In the noise control attribute (Q5), iRobot ranked the lowest (fifth). In Recommendation, iRobot did not enter the first or second recommended position in any of the five attributes. iRobot exhibits a cross-attribute characteristic of "presence in specific attribute rankings, systematic absence at the recommendation level."
Eufy's cross-attribute performance:
Eufy was mentioned in 5/10 responses (BMF=5/10). Attribute coverage: 4/5 (appeared in cleaning ability, navigation, obstacle avoidance, and noise control; absent in automated maintenance). Eufy ranked fourth or lower in all attributes, did not enter the top three in any attribute, and was completely absent in Recommendation.
Narwal's cross-attribute performance:
Narwal was mentioned in 3/10 responses (BMF=3/10). Attribute coverage: 2/5 (only appeared in obstacle avoidance Q3 and automated maintenance Q4, Q9). Narwal's attribute strength is clearly concentrated in automated maintenance (especially mopping-related functions), and it is completely absent in other attributes.
Summary of brands with attribute advantages:清洁能力属性：Roborock（首位）、Dreame（次位）
导航与地图智能性属性：Roborock（首位）、iRobot（次位，Ranking中）
障碍物识别与规避能力属性：Roborock（首位）、iRobot（次位，Ranking中）
自动化维护能力属性：Roborock与Ecovacs（并列首位，Ranking中）
噪音控制表现属性：Roborock与Dreame（并列首位，Ranking中）
14. 稳定性与置信度
根据规则3，本次审计采样轮次为1轮，低于最低验证阈值（2轮）。
稳定性无法验证（insufficient sampling）。
本报告所有结论均基于单轮10条AI回答，无法通过多轮对比验证品牌出现频率、排名顺序或推荐结果的一致性。ChatGPT作为非确定性输出系统，在不同会话、不同时间节点下可能产生不同的品牌排名与推荐结果。
建议后续审计在相同查询条件下执行至少2–3轮采样，以评估输出稳定性并计算品牌出现轮次波动率。
15. 全局聚合
以下为基于BMF的全局品牌出现频率排名，数据来源为全部10条AI回答：
- Roborock — BMF：10/10
- Dreame — BMF：10/10
- Ecovacs — BMF：9/10
- iRobot — BMF：9/10
- Eufy — BMF：5/10
- Narwal — BMF：3/10
- SharkNinja — BMF：1/10
- Xiaomi — BMF：1/10
- MOVA — BMF：1/10
Roborock与Dreame并列BMF最高（10/10），Ecovacs与iRobot并列第三（9/10），其余品牌出现频率显著低于头部四个品牌。
声明：This aggregation is frequency-based and does not represent a unified ranking output by the AI.
16. 核心发现
发现1：Roborock在全属性范围内实现Ranking与Recommendation双重首位覆盖。
在全部5个属性的Ranking问题（Q1–Q5）中，Roborock均位列第一或并列第一（Top-1 Presence=10/10）。In the Ranking questions (Q1–Q5) for all five attributes, Roborock ranked first or tied for first (Top-1 Presence=10/10). In the Recommendation questions (Q6–Q10) for all five attributes, Roborock was listed as the top recommendation (Recommendation Rate=5/5). BMF=10/10. Attribute coverage: cleaning ability, navigation and map intelligence, obstacle recognition and avoidance, automated maintenance capability, and noise control performance. Finding 2: Dreame exhibits a "high frequency of mentions, fixed secondary recommendation" structure.
Dreame's BMF=10/10, Top-3 Presence=10/10, appearing in all 10 responses and within the top three. However, Dreame's Recommendation Rate=0/5, never listed as the top choice across all 5 recommendation questions, consistently fixed in the second recommendation position. Ranking-Recommendation Gap: Dreame tied for Top-1 (1/5) in the noise control attribute Ranking, but its Recommendation Rate was 0/5, a Gap of +20%. Finding 3: iRobot has a presence in specific attribute Rankings but is systematically absent in Recommendations.
iRobot ranked second in the Rankings for navigation and map intelligence (Q2) and obstacle recognition and avoidance (Q3), but its Recommendation Rate was 0/5 across all 5 Recommendation questions. In Q7 (navigation recommendation) and Q8 (obstacle recommendation), iRobot's Ranking position (second) did not translate into a corresponding position in Recommendations (dropping to fourth in both). Finding 4: Ecovacs achieved a shared top ranking in the automated maintenance attribute but was unable to maintain it in Recommendations.
In Q4 (automated maintenance Ranking), Ecovacs tied for first with Roborock. However, in Q9 (automated maintenance Recommendation), Ecovacs was listed as a shared second recommendation, while Roborock remained the sole top choice. Ecovacs' overall Recommendation Rate=0/5. Finding 5: Brand appearance frequency exhibits a clear head-concentrated and long-tail sparse structure.
The top four brands (Roborock, Dreame, Ecovacs, iRobot) all have BMF scores of 9–10/10, accounting for the vast majority of mentions.The top four brands (Roborock, Dreame, Ecovacs, iRobot) each have a BMF of 9–10/10, collectively capturing the vast majority of mentions. The remaining brands (Eufy, Narwal, SharkNinja, Xiaomi, MOVA) all have BMFs of 5/10 or below, and none appear in the top recommended position for any attribute. Across all 10 responses, 9 distinct brands were identified, but the top recommended position is completely occupied by a single brand (Roborock). Finding 6: Narwal's attribute occurrence is highly concentrated in the automated maintenance dimension.
Narwal has a BMF of 3/10, appearing only in responses related to obstacle avoidance (Q3) and automated maintenance (Q4, Q9), with an attribute coverage of 2/5. On the automated maintenance attribute, Narwal ranks fourth in Q4 Ranking and fourth in Q9 Recommendation, described by AI as "the brand most focused on mopping automation." Narwal is completely absent from the other three attributes.
17. Limitations
Sample size limitation: This audit collected a total of 10 AI responses, covering 5 attributes and 2 question types. The sample size is small; each attribute has only 2 responses (1 Ranking + 1 Recommendation), making it impossible to verify the robustness of conclusions through multi-sample comparisons within an attribute.
Single-round sampling issue: This audit performed only one round of sampling, so stability cannot be verified (insufficient sampling). As a non-deterministic output system, ChatGPT may produce different brand rankings and recommendation results across different sessions. The current conclusions only reflect the AI output behavior of a single query and do not represent the model's stable output pattern under the same question.
Model randomness: ChatGPT's output is influenced by factors such as temperature parameters, context window, and model version updates. The same question may yield different answers at different points in time. This report cannot control or quantify the impact of the above randomness on the results.
Brand classification ambiguity: Some brands have sub-brands, regional brands, or ecosystem-related brands (e.g., the relationship between MOVA and Dreame, the affiliation between Eufy and Anker). This report adopts an independent counting principle, but different classification methods may affect the BMF calculation results.
Methodological limitations: The attribute selection was predetermined by the auditor. Different attribute choices may lead to different brand frequency distributions. Question wording (e.g., "rank highest" vs. "best for me") may affect the AI's response structure and brand ranking logic. This report only covers explicit query modes and does not cover implicit queries or multi-turn conversational scenarios.Geographic and Language Limitations: This audit was conducted at a Singapore node in an English language environment, with the audit node being the Japan market. ChatGPT may output different brand recommendation results in different regional nodes or language environments. The conclusions of this report are not applicable to inferences in other regions or language environments.
18. Appendix
Complete list of all conversation links:
● Q1 (Cleaning Ability | Ranking): https://chatgpt.com/share/6a3a7b86-47bc-83ea-8e79-54e17e5b1c9d
● Q2 (Navigation and Map Intelligence | Ranking): https://chatgpt.com/share/6a3a7bcd-0b24-83ea-898c-75e941cf8444
● Q3 (Obstacle Detection and Avoidance Ability | Ranking): https://chatgpt.com/share/6a3a7c08-612c-83ea-ad6c-a41ec8130b61
● Q4 (Automated Maintenance Ability | Ranking): https://chatgpt.com/share/6a3a7c43-5b74-83ea-b766-88a30c678a72
● Q5 (Noise Control Performance | Ranking): https://chatgpt.com/share/6a3a7c8b-2b1c-83ea-a781-fde7deb192f3
● Q6 (Cleaning Ability | Recommendation): https://chatgpt.com/share/6a3a7cbd-a228-83ea-afc5-2c000019400b
● Q7 (Navigation and Map Intelligence | Recommendation): https://chatgpt.com/share/6a3a7d04-7034-83ea-9971-51425c17b880
● Q8 (Obstacle Detection and Avoidance Ability | Recommendation): https://chatgpt.com/share/6a3a7da7-7f0c-83ea-9a23-b4e4500219bd
● Q9 (Automated Maintenance Ability | Recommendation): https://chatgpt.com/share/6a3a7deb-c83c-83ea-bed3-298dd8c62366
● Q10 (Noise Control Performance | Recommendation): https://chatgpt.com/share/6a3a7e5d-3938-83ea-a6c3-87b427bf081a
Data Source Description:
All data is derived from the 10 ChatGPT conversation records listed above, all of which are publicly shared links that can be independently accessed and verified. Data collection time: 2026-06-23. Audit node: Japan. Access environment: Singapore residential static IP, front-end user interface, English language environment. Original brand appearance records (by Query ID):
● Q1: Roborock (#1), Dreame (#2), Ecovacs (#3), iRobot (#4), Eufy (#5), SharkNinja (#5)
● Q2: Roborock (#1), iRobot (#2), Ecovacs (#3), Dreame (#4), Eufy (#5)● Q3: Roborock (#1), iRobot (#2), Ecovacs (#3), Dreame (#4), Narwal (#5), Eufy (#6)
● Q4: Roborock (#1 tied), Ecovacs (#1 tied), Dreame (#3), Narwal (#4), iRobot (#5)
● Q5: Roborock (#1 tied), Dreame (#1 tied), Ecovacs (#3), Eufy (#4), iRobot (#5)
● Q6: Roborock (first choice), Dreame (#2), iRobot (#3), Ecovacs (#4)
● Q7: Roborock (first choice), Dreame (#2), Ecovacs (#3), iRobot (#4)
● Q8: Roborock (first choice), Ecovacs (#2), Dreame (#3), iRobot (#4)
● Q9: Roborock (first choice), Dreame (#2 tied), Ecovacs (#2 tied), Narwal (#4), iRobot (#5), MOVA (#6)
● Q10: Roborock (first choice), Dreame (#2), Ecovacs (#3), iRobot (#4), Xiaomi (#5)