AI News AI资讯 1d ago Updated 1d ago 更新于 1天前 45

A third of ChatGPT ads appear in irrelevant conversations 三分之一的ChatGPT广告出现在不相关的对话中

OpenAI's ChatGPT ad platform showed a 33% irrelevance rate, with one-third of over 11,000 analyzed ads having no connection to the user's conversation 68% of ads appeared in conversations where users showed zero purchase intent, raising questions about the effectiveness of context-based targeting Sector performance varied dramatically: marketing/B2B services (47% unrelated) and software/SaaS (45% unrelated) performed worst, while data brokers (50% direct match) led in relevance OpenAI's "context 第三方平台Searchable分析2026年7-8月超1.1万条ChatGPT广告,发现33%与对话内容完全无关,仅27%为直接匹配 68%的广告出现在用户无任何购买意向的对话中,广告主大量投放预算消耗在低转化场景 不同行业投放效果差异显著:营销/B2B服务(47%不相关)和软件/SaaS(45%不相关)表现最差,数据经纪商(50%直接匹配)和旅游(43%)最佳 OpenAI以用户"跳过率"作为相关性代理指标,但第三方数据显示不相关率半年内未见改善 广告主需明确:ChatGPT内广告位不能替代成为AI回答本身,错误匹配的广告极易被忽略

65
Hot 热度
68
Quality 质量
60
Impact 影响力

Analysis 深度分析

TL;DR

  • OpenAI's ChatGPT ad platform showed a 33% irrelevance rate, with one-third of over 11,000 analyzed ads having no connection to the user's conversation
  • 68% of ads appeared in conversations where users showed zero purchase intent, raising questions about the effectiveness of context-based targeting
  • Sector performance varied dramatically: marketing/B2B services (47% unrelated) and software/SaaS (45% unrelated) performed worst, while data brokers (50% direct match) led in relevance
  • OpenAI's "context hints" targeting model differs fundamentally from keyword-based paid search, matching ads against live conversational context rather than explicit buyer signals
  • Despite OpenAI's claim that ad dismissals dropped by half since launch, the irrelevance rate remained stagnant at ~33% over the analysis period

Why It Matters

This analysis reveals a critical tension in AI-native advertising: the very conversational nature that makes ChatGPT ads feel integrated also makes precise targeting extremely difficult. For AI practitioners and advertisers, the findings suggest that context-based ad matching in LLM-powered environments requires fundamentally different strategies than traditional keyword-driven search advertising, and that high impression volume does not necessarily translate to meaningful engagement.

Technical Details

  • Data scope: Searchable analyzed 11,000+ ads served in real ChatGPT conversations between July 4 and August 4, 2026, pairing each ad with its conversation and grading relevancy across three bands: direct match, contextual match, and unrelated
  • Targeting mechanism: Unlike paid search keyword selection, ChatGPT's platform uses "context hints" where advertisers describe desired conversations, situations, and topics; OpenAI's AI engine then matches these against live conversations in real time
  • Relevancy breakdown: 27% direct match (advertising the specific product asked about), 40% contextual match (connected to earlier thread topics but not the current question), and 33% unrelated (no connection to the conversation)
  • Sector-level performance: Data brokers/background checks achieved 50% direct match; travel at 43%; automotive at 42%; marketing/B2B services at 47% unrelated; software/SaaS at 45% unrelated; insurance at only 23% unrelated but just 11% direct match (two-thirds fell in contextual band)
  • Purchase intent signal: 68% of ads appeared in conversations with no buyer, booker, hirer, or comparator intent detected at any point in the thread

Industry Insight

  • Context-based targeting is immature: The 33% irrelevance rate and the fact that no sector achieved direct relevancy above 50% suggest that AI-driven conversational ad matching is still in an early, imprecise stage—advertisers should temper expectations about placement quality versus traditional search
  • Intent detection remains a hard problem: The gap between OpenAI's dismissal-rate metric and independent relevancy analysis indicates that user behavior signals (like scrolling past) may not accurately reflect ad relevance, urging the industry to develop better measurement frameworks
  • Organic recommendations may outperform paid placements: The article notes that AI-generated recommendations within responses convert higher than organic channels, suggesting that brands should prioritize earning placement in AI-generated answers rather than relying solely on paid ad slots in conversational interfaces

TL;DR

  • 第三方平台Searchable分析2026年7-8月超1.1万条ChatGPT广告,发现33%与对话内容完全无关,仅27%为直接匹配
  • 68%的广告出现在用户无任何购买意向的对话中,广告主大量投放预算消耗在低转化场景
  • 不同行业投放效果差异显著:营销/B2B服务(47%不相关)和软件/SaaS(45%不相关)表现最差,数据经纪商(50%直接匹配)和旅游(43%)最佳
  • OpenAI以用户"跳过率"作为相关性代理指标,但第三方数据显示不相关率半年内未见改善
  • 广告主需明确:ChatGPT内广告位不能替代成为AI回答本身,错误匹配的广告极易被忽略

为什么值得看

本文揭示了AI原生广告平台的核心矛盾——声称"已了解用户需求"却存在严重匹配缺陷,对广告主、平台方和AI行业均有重要参考价值。第三方独立数据与OpenAI官方说法形成对比,为评估AI广告效果提供了客观基准。

技术解析

  • 数据来源:Searchable平台抓取2026年7月4日至8月4日期间ChatGPT内实际投放的11,000+条广告,每条广告与对应对话配对分析
  • 相关性分级:采用三档评估体系——直接匹配(27%)、上下文匹配(40%)、完全无关(33%)
  • 投放机制:与付费搜索的关键词竞价不同,ChatGPT广告采用"上下文提示"模式,广告主描述目标场景/话题,OpenAI AI引擎实时匹配对话内容
  • 行业细分数据:按行业统计不相关率,营销/B2B(47%)、软件/SaaS(45%)最高;数据经纪商(50%直接匹配)、旅游(43%)、汽车(42%)最佳
  • 购买意向分析:68%广告出现在用户无任何购买/预订/比较意图的对话中

行业启示

  • AI广告匹配精度仍是行业痛点:声称"懂用户"的AI广告平台实际匹配准确率不足30%,平台方需优化上下文理解与意图识别能力
  • 广告主应调整投放策略:避免将AI助手广告位等同于传统搜索广告,优先布局高转化场景(如旅游、汽车),同时重视"成为AI回答"而非仅购买广告位
  • 第三方评估机制将成行业标配:OpenAI内部指标(跳过率)与独立分析结果存在偏差,行业需要更多透明、可验证的AI广告效果评估标准

Disclaimer: The above content is generated by AI and is for reference only. 免责声明:以上内容由 AI 生成,仅供参考。

GPT GPT LLM 大模型 Product Launch 产品发布 Ethics 伦理 Policy 政策