Loading tutorials…
Loading tutorials…
Surveys are the only way to ask users *why* they did what they did. Recordings show the what; surveys close the loop. Most teams either skip surveys entirely or run badly-targeted ones that produce noise. This is the setup that produces signal.
Who this is forMarketers or product teams with Hotjar installed who want qualitative "why" data from real users. Useful especially for landing-page bounce analysis, post-purchase intent capture, and exit-intent friction surveys.
What you'll need
Step 1
Hotjar offers On-site surveys (popup or slide-in), External link surveys (standalone URL), and the Feedback widget (always-on reaction bar). Each fits a different question.
On-site survey (slide-in): triggered by behavior on a page. Best for: exit-intent ("what stopped you?"), mid-funnel checks ("anything confusing on this page?"), post-purchase ("what almost stopped you from buying?").
On-site survey (popup): more intrusive. Best for: high-priority targeted questions to a small audience (e.g., users who spent 3+ minutes on pricing without converting). Use sparingly — popups annoy.
External link survey: standalone URL you share via email, post-purchase email, NPS campaigns, or social. Best for: longer surveys, customer interview screening, post-support follow-ups.
Feedback widget: always-on small button bottom-right of the page with emoji reactions. Best for: passive ongoing pulse-check across the site. Lower response rate but lower friction.
Decision rule: start with one on-site exit-intent slide-in on your top landing page. Add the Feedback widget site-wide. That's 90% of the value with 10% of the maintenance.
Step 2
Surveys → New survey. Keep it short — 1-3 questions. The first question must be specific.
Open Surveys → click New survey (top-right). Pick On-site survey for the standard case.
Name it clearly: "Pricing page exit-intent — May 2026". This appears nowhere user-facing but matters for your filter library.
Pick a template (Hotjar offers Exit Intent, Post-Purchase, NPS, Customer Effort Score, etc.) — these are starting points, not finals. Edit aggressively.
First question rules: specific, not vague. "What almost stopped you from buying today?" beats "How was your experience?" Open-ended text beats multiple choice for diagnostic surveys.
Question 2 (optional): contextual follow-up. "Anything else you wanted to know but couldn't find?" Save 5-point scales for NPS-style usage, not diagnostic.
Question 3 (optional): qualifier or email capture. "Can we follow up? (email)" — converts a survey response into a customer interview lead.
Hard rule: 3 questions max. Every additional question halves response rate. Below 30 seconds of effort is the threshold.
Step 3
Target by URL, behavior, device, and timing. Surveying everyone produces vague answers; surveying the right 5% produces signal.
In the survey editor → Targeting tab.
URL targeting: limit to the specific page(s) where the question is relevant. Exit-intent on /pricing? URL contains /pricing. Post-purchase? URL equals /thank-you.
Behavior trigger: pick from "Exit-intent" (desktop only — fires on mouse leave toward browser chrome), "Time on page" (e.g., after 30s), "Scroll percentage" (e.g., after 60% scroll), "On click" (when a specific element is clicked), "On exit page" (mobile-friendly alternative to exit-intent).
For exit-intent on desktop + scroll-50% on mobile: Hotjar lets you set device-specific triggers in advanced settings. Use this — exit-intent doesn't exist on mobile (no mouse to leave).
Device targeting: most diagnostic surveys should run on desktop AND mobile. Skip targeting if you don't have a specific reason to exclude.
Traffic source: limit to specific UTM sources if the question is paid-traffic-specific ("Why didn't our Google Ads landing page convert you?"). Configured via Hotjar User Attributes or URL parameters.
Frequency: limit to "Once per user" — otherwise the same user gets repeatedly nagged and response quality drops.
Step 4
The Feedback widget is a low-friction, always-visible reaction button. Useful as a passive pulse on overall sentiment.
Open Feedback → New widget.
Pick widget type: Emoji (1-5 emoji scale) or Rating (1-5 star scale). Emoji performs better — more inviting, less formal.
Position: bottom-right is standard. Bottom-left if you have a chat widget at bottom-right.
Targeting: enable on all pages except checkout (don't distract users mid-conversion).
Customize the prompt: "How's your experience?" works. "Tell us about your visit" prompts longer responses but lower fill rate.
Add the optional comment box (recommended): after the emoji rating, users can write a short text. This is where the actionable feedback lives — the rating alone is just sentiment.
Save and publish. The widget appears on the next page load. Review responses weekly under Feedback → Widget responses.
Step 5
Hotjar's preview mode lets you trigger the survey on your live site without it being live for real users. Test before publishing.
In the survey editor → top right Preview → Copy preview link.
Open the preview link in a new browser tab. Navigate to a URL that matches your targeting. Trigger the behavior (e.g., move mouse to the top of the window for exit-intent).
The survey should appear. Click through it. Verify: (a) the questions render correctly, (b) the styling matches your brand, (c) the submit button works, (d) the thank-you message appears.
If exit-intent doesn't fire on your test, the survey targeting is too narrow or the trigger isn't configured correctly. Re-check Targeting tab.
Once preview works, click Publish (top right). The survey is now live for real users matching the targeting rules. Hotjar shows a confirmation modal — confirm to go live.
Step 6
Surveys → click your survey → Responses tab. Skim every response, tag by theme. Use tags to build a backlog of CRO experiments.
Calendar a 15-min weekly review for surveys. Friday afternoon is the standard.
Open Surveys → click your survey → Responses tab. Sort by Most recent.
Skim each response. Tag by theme: "pricing-concern", "trust-concern", "feature-missing", "tech-issue", "competitor-comparison". Use Hotjar's tag UI inside each response.
After 4 weeks, open Responses → group by tag. The top 3 tags by count are your top 3 CRO experiment hypotheses.
Example: 12 responses tagged "trust-concern" = run a trust-signal experiment (testimonials moved above fold, security badges, etc.). 8 responses tagged "pricing-concern" = run a pricing presentation experiment.
Survey responses are gold for ad copy too — direct quotes from real users beat any copywriter's best guess. Pipe them into your Google Ads / Meta ad copy library.
Step 7
Surveys go stale. The same question gets the same answers after enough volume. Rotate questions every 4-8 weeks to surface new patterns.
After 100-200 responses to a question, additional responses produce diminishing new insights.
Rotate: archive the survey (don't delete — keep the responses), draft a new question targeting the same audience.
Example rotation: Month 1 — "What almost stopped you from buying?" → Month 2 — "What's the one thing you wanted to know before buying?" → Month 3 — "If pricing were 20% lower, would that have changed your decision? Why?"
Each rotation surfaces a new angle on the same underlying friction. After 3-4 rotations, you have a comprehensive picture of why users don't convert.
Maintain a "question library" doc with the questions you've run + the patterns each surfaced. New rotations are informed by past ones.
Common mistakes
Asking vague generic questions
What goes wrong: "How was your visit?" produces 4.2/5 average rating with no actionable content. The team concludes "users like the site" while conversion sits at 1.3%. The survey infrastructure is in place but produces no useful signal. Wasted ~$30-100/mo on Hotjar Plus and zero CRO progress for 6+ months.
How to avoid: Ask specific diagnostic questions. "What almost stopped you from buying?" or "What's the one thing you wanted to know but couldn't find?" beat vague rating questions. Open-ended text > multiple choice for diagnostic surveys.
Showing surveys to 100% of visitors
What goes wrong: Untargeted surveys get 0.3-0.8% response rate with mostly noise (users who responded to be polite). High-intent users — the ones with actionable feedback — are diluted in the response pool. The team makes decisions on the noise, not the signal. On a $50K/mo ad-spend account, this typically costs $3,000-8,000/quarter in CRO experiments aimed at the wrong patterns.
How to avoid: Target high-intent segments: users who spent 2+ minutes on pricing, users who triggered exit-intent on checkout, users who added to cart but didn't purchase. Response rates jump to 5-10% with much higher signal-to-noise.
Running 5 surveys at once
What goes wrong: Multiple active surveys compete for attention. Users get survey-fatigued and ignore all of them. Response rates collapse across every survey. The team turns surveys off entirely and concludes 'Hotjar surveys don't work' — when it was the volume that didn't work. Cost: 6-12 months of the survey infrastructure on Hotjar Plus+ ($192-1,200/year) producing no signal.
How to avoid: Run 1-2 surveys at most. Add the always-on Feedback widget for passive pulse. Rotate the main survey every 4-8 weeks instead of stacking surveys.
Never reviewing responses
What goes wrong: Survey gets 200 responses over 3 months. No one opens the Responses tab. Hotjar bill continues at $40-200/mo. The most valuable qualitative data you have is gathering dust while the team makes CRO decisions on guesswork.
How to avoid: Calendar a 15-min weekly review. Read every response. Tag by theme. Build a backlog. If you genuinely can't commit to weekly review, turn surveys off — you're paying for capacity you don't use.
Sharing surveys on checkout pages
What goes wrong: Distracting users mid-conversion. A survey popping on checkout step 2 drops cart-conversion by 5-15%. On a $50K/mo ecom store, that's $2,500-7,500/mo in lost revenue from a survey that was supposed to *improve* conversion.
How to avoid: Exclude /checkout/* and any conversion-critical URLs from survey targeting. Survey post-purchase (/thank-you) or pre-checkout (cart abandonment), never during.
Asking for an email at the start of the survey
What goes wrong: Response rate drops by 40-60% when an email field is question 1. Users who would have answered question 1 anonymously bail at the email ask. You collect fewer responses AND fewer emails. For a B2B SaaS where each lead is worth $200-1,000 LTV, that's $2,000-10,000/quarter of leads not captured because the survey structure scared users off.
How to avoid: Put email capture at the END of the survey, as an optional question. "Can we follow up? (optional)" after they've already invested 30 seconds. 15-25% will leave email at that point — much higher than upfront.
Recap
Done — what's next
How to set up Hotjar Session Recordings the right way
Read the next tutorial
Hand it off
Most DIY surveys ask vague questions to broad audiences and produce vague data. A specialist drafts questions that actually predict revenue ("what almost stopped you?" produces 10x the signal of "how was your experience?") and runs the response review as a recurring practice. Match with a vetted conversion specialist from $14-16/hr — typically $300-700/mo for ongoing survey + recording + CRO work.
See specialist rates
0.5-1% for an untargeted homepage survey. 3-5% for a behavior-targeted survey (exit-intent on pricing). 8-12% for a post-purchase survey on the thank-you page. Below 1% on a targeted survey usually means the question is too vague or the targeting is wrong.
Not natively in Hotjar. The workaround: run survey A for 4 weeks, archive, run survey B for 4 weeks, compare response volume and theme distribution. Don't run both simultaneously — they compete for attention and dilute both.
Open the survey → Responses tab → top-right Export → CSV. Includes all responses with timestamps, page URL, device, and any tagged themes. Useful for pulling into Notion, Linear, or whatever CRO backlog tool you use.
Yes. Push hj("trigger", "survey_id") from your site code to fire a survey on demand. Useful for app-specific events (e.g., "user completed onboarding step 3 → trigger feedback survey"). Set the survey trigger to "Display on JS Trigger" in Targeting.
Yes, as a template. Surveys → New survey → NPS template. Standard 0-10 scale + open-ended follow-up. Useful for tracking customer satisfaction over time — but limited as a diagnostic tool. Most teams find specific friction questions more actionable than NPS.
Yes — Hotjar links a survey response to the session recording when the user is currently being recorded. Open the survey response and there's a "View recording" link. This is gold: read the user's answer, then watch what they did before answering. Use it.
Hotjar
Hotjar's recordings are the most powerful feature in the tool — and the most-wasted. The difference is filter discipline. This is the setup that turns 1,000 recordings/week into 5 useful insights, not 1,000 hours of "someday I'll watch these."
Hotjar
GA4 tells you 60% of users abandoned at checkout step 2. Hotjar Funnels tells you which 60% — and lets you click straight into 5 recordings of the people who dropped off. That's the workflow.
Hotjar
Out of the box, Hotjar tracks page views, clicks, and scroll. Anything beyond that — button clicks on a SPA, app-level milestones, custom funnel steps — needs custom events. This is the install path that doesn't break.
Hotjar
DIY Hotjar is a great idea — until it isn't. This is the honest framework: when the cost of unwatched recordings and unanalyzed surveys exceeds the cost of hiring help, and how to tell which side you're on.