Hotjar

I tested Hotjar across the MytheAi production site for a full week using the Observe tier free plan (35 sessions per day), recording heatmaps on the homepage plus /tools plus /best/coding plus /tools/claude, and running one feedback widget on /compare. Intent was to peer-double against Maze (session 144), PostHog (session 159), Mixpanel and Amplitude (session 158): which one earns the analytics slot for a small-team SaaS where session recording is more valuable than event funnels.

What worked was the heatmap depth and the recording trim UX. Heatmaps showed exactly where users tapped on mobile (most clicks on the search bar, then category cards, very few on the role grid below the fold), confirming a homepage layout hypothesis we had only guessed at. Session recordings with rage-click detection caught two real UX bugs the same week: a tooltip on /compare that opened off-screen on iPhone Mini, and a sticky-mobile-bar overlap I thought session 110 had fixed. The free tier is genuinely usable for early-stage validation, no credit card required.

What broke was the 35-sessions-per-day cap (worth knowing for any site over a few hundred daily visitors), and the slow load time of the playback UI when scrubbing through 8-plus minute sessions on slow wifi. The Slack integration sends rage-click notifications but the message format buries the URL behind two clicks. Funnel analysis on the paid tier felt like a worse PostHog or Mixpanel, the strength is recordings not metrics.

Verdict: Hotjar wins on session-recording UX and qualitative UX research depth. PostHog wins for self-hosted plus event analytics plus feature flags as a unified platform. Mixpanel and Amplitude win for cohort-funnel-retention analytics at scale. Maze wins for moderated usability studies. This batch unlocks /compare/hotjar-vs-posthog plus hotjar-vs-mixpanel surface for the analytics decision cluster.