Loading tutorials…
Loading tutorials…
Typeform shipped a series of AI features in 2024-2025 — dynamic content via Smart Reference, conversational forms, and AI-powered response analysis. Most are useful. A few are oversold. This is the honest breakdown.
Who this is forMarketers and product teams curious about whether Typeform AI features actually improve outcomes — or are mostly marketing copy. If you are evaluating whether to upgrade to a Business plan specifically for AI, read this first.
What you'll need
Step 1
Most Typeform AI features (Smart Reference dynamic content, AI Insights, advanced response analysis) are Business-tier only.
Open Settings → Plan. Confirm "Business" or higher.
On Plus, you get a limited AI suite (basic recall, simple personalization). Most advanced AI features require Business.
Pricing for AI tier is sometimes positioned as an add-on rather than included — check current pricing on typeform.com.
Step 2
Smart Reference pulls answers from earlier questions into later question text. "Hi {{First name}}, what is your biggest challenge with {{Selected tool}}?"
In any question, type @ to open the Smart Reference picker.
Pick the source question (e.g., "First name" or "Which tool do you use?").
The reference appears as a chip in the question text. At runtime, it is replaced with the respondent's answer.
Smart Reference lifts perceived personalization significantly. Use sparingly — every question with a reference feels personalized; every question with one feels gimmicky.
Best uses: greeting screens ("Welcome {{First name}}"), context-specific follow-ups ("You said {{Tool}} is your main blocker — tell me more.")
Step 3
AI Insights summarizes themes across responses to open-ended questions. Useful for customer-research surveys at scale.
In Typeform → Responses → AI Insights, select a long-text or short-text question with at least 50 responses.
AI Insights generates a summary of dominant themes and sentiment.
Output is best-treated as a starting point for manual review, not a final analysis. AI summaries often miss nuanced or low-frequency themes.
Best uses: NPS open-text 'why,' customer-feedback survey 'what should we build next,' onboarding survey 'what almost stopped you from signing up.'
Limit: works poorly on technical or domain-specific responses. Manual coding still beats AI for niche markets.
Step 4
Typeform offers "conversational" AI-driven forms. In practice, completion rates drop on these vs. traditional forms. Skip unless very specific use case.
Conversational AI forms use an LLM to dynamically generate follow-up questions based on respondent answers.
In testing, completion rates are typically 15-25% LOWER on conversational AI forms vs. traditional logic-jump forms.
Reason: respondents prefer the predictability of a multi-choice question over an open-ended AI conversation.
Skip unless: you are running unstructured customer-discovery interviews where flexibility matters more than completion rate.
Step 5
AI feature behavior in preview differs from production. Always run a 50-respondent pilot before deploying AI features to a main funnel.
Pick a low-stakes form (internal feedback, newsletter signup).
Enable the AI feature you want to test.
Run 50 real respondents through it.
Compare completion rate, time-to-complete, and downstream conversion against a non-AI baseline.
If AI feature lifts metrics: roll out to main funnel. If neutral or negative: disable.
Step 6
AI features get enabled then forgotten. Audit quarterly — disable features that are not moving conversion or response quality.
Every quarter, open each form with AI features enabled.
Pull completion rate, time-to-complete, and lead-quality metrics for the period.
Compare to the pre-AI baseline.
Disable any AI feature that is not lifting outcomes. Re-enable only if a specific hypothesis justifies it.
Common mistakes
Enabling AI features without measuring impact
What goes wrong: You assume AI features are working because they are 'new.' Conversion is flat or down, but you do not notice because you did not baseline. Quarterly review reveals zero lift.
How to avoid: Always baseline completion + conversion before enabling AI features. Compare 30 days post-enable. Disable if no lift.
Using Smart Reference everywhere
What goes wrong: Every question references the respondent's earlier answer. Feels gimmicky after 3 questions in. Completion drops on the longer forms.
How to avoid: Use Smart Reference sparingly — 1-2 references per form. Reserve for high-impact moments (greeting, key follow-up).
Trusting AI Insights as final analysis
What goes wrong: AI summary misses important themes. Product decisions are made on incomplete signal. Wasted product-research cycle.
How to avoid: Treat AI Insights as a first-pass review. Manually code 20-30 responses to validate themes before acting on them.
Switching to Conversational AI forms for "modern UX"
What goes wrong: Completion rate drops 15-25%. Lead volume drops proportionally. The 'modern' feel is not worth the lost leads.
How to avoid: Stick with traditional logic-jump forms for production funnels. Reserve conversational AI for low-stakes customer-discovery only.
Recap
Done — what's next
How to build a Typeform survey that actually gets completed
Read the next tutorial
Hand it off
Most teams enable Typeform AI features and never look back. Specialists know which features actually lift conversion and which are marketing copy. A 1-hour specialist audit will save you 10+ hours of fiddling with features that do not move the needle.
See specialist rates
Most are, but Typeform sometimes positions advanced AI features as add-ons or higher-tier-only. Check the current pricing page before upgrading specifically for AI.
Yes. The referenced field's answer carries forward regardless of which branch the respondent took, as long as the referenced question was answered.
It is good at surface-level theme identification (top 3-5 themes), poor at nuanced or low-frequency signal. Treat as a first pass; validate manually before acting on it.
Probably not for production funnels. Completion rates are typically 15-25% lower than traditional logic-jump forms. Reserve for low-stakes customer-discovery work.
Typeform
Typeform's one-question-at-a-time UX is what makes it convert — and what makes it brittle. This walks through how to structure questions, pick the right field types, and ship a survey that hits 60%+ completion rate.
Typeform
Quizzes are the highest-converting form pattern in B2B marketing. Done right, a quiz captures 3-5x the leads of a plain form. Done wrong, it is a glorified survey with worse completion. This walks through the right pattern.
Typeform
Logic jumps are what turn Typeform from a survey into a smart funnel. Done right, they route every respondent to a personalized path. Done wrong, they create dead ends and respondents stuck on the wrong branch.
Typeform
DIY Typeform is a great idea — until it isn't. This is the honest framework for when hiring beats DIY, plus what a specialist actually does that you may not be doing.