# Hotjar vs PostHog: which do AI models recommend for session replay, October 2026

GTM AI Recommendation Index, October 2026 Edition, Session replay and heatmaps. Five of fourteen models named Hotjar first on the direct prompt; two named PostHog. Page: https://gtm-ai-index.com/gtm-data/session-replay-and-heatmaps/hotjar-vs-posthog/

| | First-choice share | Rank | Negative rate | Labels | Models naming it |
|---|---|---|---|---|---|
| Hotjar | 16% | #2 of 10 | 23% | 62 | 14 of 14 |
| PostHog | 7% | #4 of 10 | 5% | 37 | 14 of 14 |

## The direct prompt, model by model

- Grok 4.1 Fast: hotjar first (first choices: Hotjar) (alternatives: LogRocket, Microsoft Clarity, Mouseflow)
- Mistral Small: hotjar first (first choices: Hotjar, Lucky Orange) (alternatives: Fullstory, PostHog, UXCam)
- Qwen 3.7 Flash: hotjar first (first choices: Hotjar) (alternatives: LogRocket, Matomo)
- Kimi K2: hotjar first (first choices: Hotjar) (alternatives: LogRocket, Mouseflow, Smartlook)
- GLM 4.7 FlashX: hotjar first (first choices: Hotjar) (alternatives: Crazy Egg, Fullstory, Microsoft Clarity, Mouseflow)
- DeepSeek V4 Flash: posthog first (first choices: PostHog) (alternatives: Hotjar, LogRocket, Microsoft Clarity, Mouseflow)
- GPT-6 Luna: posthog first (first choices: PostHog) (alternatives: Fullstory, Hotjar)
- Claude Haiku 4.5: neither first, one named (first choices: Mouseflow) (alternatives: Hotjar)
- Gemini 3.5 Flash: neither first, one named (first choices: Pendo) (alternatives: LogRocket, Mouseflow, PostHog)
- Perplexity Sonar: neither first, one named (first choices: Fullstory) (alternatives: Hotjar, LogRocket)
- MiniMax M2.5: neither first, one named (first choices: Mixpanel) (alternatives: Fullstory, PostHog, UXCam)
- Muse Glimmer 30B: neither first, one named (first choices: Fullstory) (alternatives: Contentsquare, Heap, Hotjar, Microsoft Clarity, PostHog, VWO)
- GPT-5.4 mini: neither named (first choices: Contentsquare, Mouseflow) (alternatives: LogRocket)
- Llama 4 Maverick: neither named (first choices: Smartlook) (alternatives: Adobe Analytics, Fullstory)

## What the models said about Hotjar

- "Hotjar was merged into Contentsquare on 1 July 2025; ... By default Hotjar starts session recording as soon as the script loads, which means it can capture visitor activity before consent is given." (Muse Glimmer 30B, negative prompt, hard negative)
- "Tools to Avoid or Use with Extreme Caution... Hotjar: Most popular heatmap tool, but criticized for capturing more PII than expected" (Grok 4.1 Fast, negative prompt, hard negative)
- "Collects more personal data than most teams realize; has faced significant legal challenges; default settings may not adequately mask PII" (Kimi K2, negative prompt, hard negative)
- "Pick heatmap-first tools like Hotjar or Clarity if you mainly want quick UX insights and broad behavior patterns." (Perplexity Sonar, comparative prompt, first choice)
- "Best for: Beginners and small-to-mid-sized teams doing user experience research ... Strength: Fastest setup, easiest to use" (DeepSeek V4 Flash, comparative prompt, first choice)
- "Very popular for UX research and product teams... Strong on heatmaps, recordings/session replay, surveys, and feedback" (GPT-5.4 mini, comparative prompt, first choice)

## What the models said about PostHog

- "PostHog does offer data scrubbing features, but I was recently surprised to see their announcement about integrating AI to analyze session replays on behalf of users. This raises a red flag" (Muse Glimmer 30B, negative prompt, soft negative)
- "If you enable network capture, scrutinize it separately from replay masking: requests and responses can contain sensitive headers or bodies" (GPT-6 Luna, negative prompt, soft negative)
- "I'd generally recommend PostHog if you want the best balance of visitor behavior analytics, product analytics, and session replay in one place" (GPT-5.4 mini, paraphrase prompt, first choice)

Share is the count of first choices across the direct, paraphrase, budget and scale prompts over all fourteen models, for a mid-market B2B company; rank is within the category. Comparisons are drawn for the top eight products in each category. Published under CC BY 4.0; the output is the models' output, and nothing here is a recommendation by the index.