Pick one question your prospects already ask ("Am I ready for my test?"), decide three or four outcomes that each lead to something you sell, and have AI draft six to ten questions that sort people into them. Build it in a quiz tool with scoring, ask for an email just before the result, and send each outcome its own follow-up.
The common mistake is starting with the questions. AI will write twenty quiz questions in ten seconds, and they'll be pleasant and pointless. The value of a quiz funnel comes from the outcomes: if the result doesn't tell someone something useful and point them to a specific next step, you've collected an email address from a person who will never open your next message.
Pick the quiz type that fits what you sell
There are three kinds of lead-capture quiz, and each suits a different business.
| Type | What the person gets | Suits | Example |
|---|---|---|---|
| Scored assessment | A score and what it means | Services where readiness or level matters | "How close are you to passing your driving test?" |
| Outcome or match | A type or category | Businesses with several distinct offers | "Which yoga class suits you?" |
| Recommender | A specific package or product | Clear packages with different price points | "What level of wedding planning help do you need?" |
Scored assessments tend to feel most valuable because people want to know where they stand. Outcome quizzes are the easiest to share. Recommenders convert best when the packages really are different; if they overlap, people end up confused by a result that could have been any of them.
Design the outcomes before any questions
Take an illustrative driving school with four instructors. Its prospects are learners, and parents of learners, who all ask some version of the same thing: how many more lessons will this take? That becomes a scored quiz with four outcomes, each tied to an offer (the score bands here are a first guess; they get corrected once the questions exist):
| Outcome | Score band | What the result says | Next step offered |
|---|---|---|---|
| Building foundations | 0-9 | Core car control still needs work; regular weekly lessons will move fastest | Block of 10 lessons |
| Getting there | 10-15 | Most skills are in place; a few situations need practice | 5-lesson bundle focused on weak spots |
| Nearly test-ready | 16-20 | Ready to find out for sure | A mock test lesson |
| Nervous starter | Any score, if "very anxious" chosen | Nerves are normal and workable; start calm | A 90-minute quiet-roads intro lesson |
Notice the fourth outcome overrides the score. A learner who is technically capable but frightened needs a different first lesson from a confident one, and a quiz that ignores that sends the wrong offer to exactly the people most likely to hesitate.
Draft the questions with AI
With the outcomes fixed, AI is genuinely fast at drafting questions. Give it the outcomes, the scoring range and some constraints:
I run a driving school. Draft a scored quiz: "How close are you to passing
your driving test?"
Outcomes and score bands (0-20):
- Building foundations 0-9, Getting there 10-15, Nearly test-ready 16-20
- Nervous starter: triggered by one anxiety question, regardless of score
Rules:
- 8 multiple-choice questions, max 12 words each, 3-4 answers each.
- Each answer worth 0-3 points; show the points in brackets.
- One question on anxiety, with a "very anxious" answer that triggers
Nervous starter.
- Plain words a 17-year-old and their parent both understand.
- No trick questions, no legal questions, no questions that need a
right answer from the road rules.A typical first draft (illustrative, trimmed to four questions):
1. How many hours of lessons have you had so far?
None (0) / Under 10 (1) / 10-30 (2) / Over 30 (3)
2. How confident are you with hill starts and junctions?
Not at all (0) / A bit (1) / Mostly (2) / Very (3)
3. Can you reverse park without your instructor's help?
Never tried (0) / Sometimes (1) / Usually (2) / Every time (3)
4. Have you driven on busy, fast multi-lane roads?
No (0) / Once or twice (1) / Regularly (3)Three fixes are needed. Question 2 asks about two skills at once; someone confident at junctions but shaky on hills has no honest answer, so split it or pick one. Question 4 jumps from 1 point to 3, which makes it count for more than the others without anyone deciding it should. And the draft never asks the question the school most needs answered, which is whether the learner has a test date booked. That isn't a scored question, but it changes the follow-up completely, so add it as an unscored question near the end.
Make the scoring hold up before launch
A quiz where everyone lands on "Getting there" is broken, and you won't notice until the results come in. Test the scoring first with made-up people. Ask the assistant:
Invent 6 learners with different experience levels (include one who has had
40 hours but is very anxious, and one total beginner who is confident).
For each, choose the answers they'd give to this quiz, total the score,
and say which outcome they land in. Flag any learner who lands somewhere
a driving instructor would disagree with.In an illustrative run, a learner with 25 hours who couldn't yet reverse park scored 14 and landed in "Getting there", which an instructor would agree with. But a learner with 40 hours, strong skills and no experience of busy roads scored 17, "Nearly test-ready", when an instructor would want a couple of lessons on fast roads first. The fix was to make the busy-roads question cap the outcome: anyone answering "No" can't land higher than "Getting there". A quick sum is also worth doing by hand: with eight questions worth up to 3 points, the maximum is 24, not 20, so the bands in the brief needed rescaling to 0-11, 12-18 and 19-24.
Where to put the email gate
The usual pattern is to ask for contact details after the last question and before the result, framed as "where should we send your result?". People who have answered eight questions are invested; asking at the start loses most of them before they begin.
- Ask for the minimum. First name and email. Make the phone number optional and say why you'd use it ("so an instructor can suggest lesson times").
- Get consent properly. In many places marketing email needs clear opt-in. Use an unticked box: "Send me my result and a few tips for my stage of learning. Unsubscribe any time." The result itself can still be shown or sent either way.
- Show the result on screen as well. Emailing the result only can feel like a bait-and-switch; show a short version immediately and send the detail by email.
- Parents are leads too. Add one question: "Who's filling this in?" with "the learner" or "a parent". It changes the tone of the follow-up.
Build it in Tally, Typeform or ScoreApp
All three handle scoring and branching. The differences that matter for a small business are the price, the response limits and how much the tool does for you. List prices in USD, checked September 2026:
| Tool | Cost | Response limit | Worth knowing |
|---|---|---|---|
| Tally | Free; Pro $29 a month or $290 a year | Unlimited submissions within fair use | Conditional logic, calculations, redirect on completion and webhooks are all free; Pro removes branding and adds custom domains, pixel tracking and drop-off insights |
| Typeform | Basic $29 a month ($25 annually); Plus $59 ($50); Business $99 ($83) | 100, 1,000 and 10,000 responses a month | Scoring, calculations and outcome quizzes on paid plans |
| ScoreApp | Free; Starter $39 a month ($30 annually); Business $99 ($75); Pro $149 ($112) | 10, 100, 1,000 and 3,000 responses a month | Built for scorecard quizzes; an AI quiz builder, landing pages and dynamic result pages on every plan |
Do the response sum before choosing. If you expect 300 completed quizzes a month, Typeform Basic's 100-response cap and ScoreApp Starter's 100 are both too small, and Tally's free tier becomes very attractive.
The build itself, in any of them, follows the same steps:
- Enter the questions and give each answer its points (a calculated field in Tally, scoring in Typeform, built in to ScoreApp).
- Add the override rules: the anxiety answer sends people to "Nervous starter"; the busy-roads "No" caps the outcome.
- Use logic to show each score band its own result, on screen or on a separate results page.
- Pass the outcome to your email tool as a tag, through a native integration or a Zapier or Make step, so the right follow-up sequence starts.
- Test every path on a phone. Most quiz traffic from social media arrives on mobile.
If you'd rather host the results on your own site, a simple page per outcome works well; building a landing page with AI that converts covers the layout.
Write result pages people actually read
A result page has four jobs: name the outcome, explain it in terms of the person's answers, give one piece of useful advice they can use today, and offer the next step. Ask AI to draft each one with the answers in front of it:
Write the result page for "Getting there" (score 12-18).
Include: a one-line headline; 2 sentences on what this score usually means;
one practical tip the learner can use on their next drive; the offer
(5-lesson bundle focused on weak spots, $[PRICE]); a button label.
Refer to their weakest answer using the merge field {weakest_skill}.
80-120 words. Encouraging, not salesy. No pass-rate claims.An illustrative draft:
You're getting there.
Most of the basics are in place. What usually separates learners at this
stage from test day is a handful of situations, and for you that's
{weakest_skill}.
Try this on your next drive: talk through what you see out loud. It slows
your thinking down and shows your instructor exactly where to help.
Our 5-lesson bundle focuses on the skills your answers flagged.
[Book my 5 lessons]The one edit here is practical: check that {weakest_skill} has a sensible fallback for people who scored well on everything, or they'll see "for you that's ." with a gap.
Follow-up emails for each outcome
Three short emails per outcome over ten days is plenty. For "Nearly test-ready", for example: day 0, the result and the mock test offer; day 3, the things that tend to trip up confident learners in a mock test, written from the instructors' own experience rather than invented statistics; day 8, the next two available mock test slots. Tag anyone who books so they leave the sequence immediately. The same pattern as a welcome email sequence applies, but each outcome gets its own version, which is where AI drafting saves the most time: four outcomes times three emails is twelve messages, drafted in under an hour and edited in another.
Promote the quiz where the question already comes up
A quiz nobody sees captures nobody. The best places to put it are the moments when prospects are already asking the question it answers:
- Your DM and email replies. When someone asks "how many lessons will I need?", reply with a line and the link: "Honest answer: it depends where you're starting. This 2-minute quiz gives you a proper estimate." Save it as a quick reply so everyone sends the same thing.
- Instagram bio and a pinned post. The link in bio goes straight to the quiz, not the homepage.
- Google Business Profile posts. A monthly post with the quiz link and a button, aimed at people comparing driving schools.
- Printed material. A QR code on the car's rear window sticker or a leaflet at the local college, pointing at the quiz with a tracking parameter so you know it came from print.
- Existing customers' parents. A line in lesson confirmation emails: "Got another learner in the family? Here's where to start."
The call to action matters as much as the placement. "Take our quiz!" tells people nothing. "Find out how many lessons you've got left: 8 questions, 2 minutes" says what they get and what it costs them. AI is useful for drafting ten versions of that line; pick two and run each for a fortnight to see which gets more starts.
The driving school's first month in numbers
Here are illustrative numbers for the four-instructor school, promoting the quiz from its Instagram bio, its Google Business Profile and a banner on the website:
| Step | Number | Rate |
|---|---|---|
| Visitors to the quiz page | 1,200 | - |
| Started the quiz | 480 | 40% of visitors |
| Finished all questions | 330 | 69% of starters |
| Gave an email for the result | 190 | 58% of finishers |
| Booked a lesson within 30 days | 14 | 7% of leads |
Compare that with the school's old contact form, which produced six enquiries a month. The quiz leads also arrive pre-sorted: the office knows which offer to mention before it replies. Setup took about six hours across a week: outcomes and questions (two hours), scoring tests (one hour), building in Tally (two hours) and the twelve follow-up emails (one hour, plus edits). Running cost on Tally's free plan was nil beyond the email tool the school already paid for. Your rates will differ; the point of the table is to show which step to fix when numbers are low.
Quiz ideas for a yoga studio and a wedding planner
A yoga studio's class match
A yoga studio's "Which class suits you?" quiz has five questions about energy levels, flexibility, injuries and what people want from a class (sweat, calm, strength). Outcomes: Beginners' course, Slow flow, Power, Yin. The injury question matters most: anyone who mentions a recent injury gets a result that suggests speaking to a teacher before booking, not a straight sales pitch. The follow-up offer is the intro pass, with the result naming which classes on the timetable to try first.
A wedding planner's help finder
A wedding planner's recommender quiz sorts couples into full planning, partial planning and on-the-day coordination. Six questions: wedding date, whether the venue is booked, guest count, how much time they have each week, how they feel about managing suppliers, and a budget range. The budget question is optional but useful, because it stops couples with a small budget being recommended the most expensive package. The result page invites them to book a consultation, and the planner sees the answers before the call, which shortens it.
Why quiz funnels leak, and how to spot the hole
- A drop-off at one question. Most tools show where people quit. A cliff at question 4 usually means the question is confusing or personal too early; move it or reword it.
- Low opt-in at the gate. If fewer than half of finishers give an email, the promise of the result is weak. Make the result sound specific ("your score and the two skills to work on next") rather than generic.
- Everyone gets the same result. Export the first 50 submissions and count outcomes. If 40 of 50 land in one band, rescale the bands.
- Junk emails. A burst of obviously fake addresses means the result is shown before the email is checked. Send the full result by email, which gives people a reason to type a real address.
- Leads that don't buy. If leads arrive but nobody books, the outcomes aren't tied tightly enough to offers, or the follow-up is too slow. Reply to phone-number leads within a working day.
A quiz is one kind of lead magnet; if your audience would rather have a checklist or a calculator, creating lead magnets with AI compares the options. And if quiz traffic is healthy but people still aren't booking, finding out why website visitors don't buy helps you look further down the funnel.
Further reads
- Best AI Landing Page Builders for Small Businesses (2026) — Where to host the quiz page and results pages.
- What Is Lead Nurturing and How Can AI Automate It? — How to keep quiz leads warm until they're ready to buy.
- AI Lead Generation for Small Businesses: What Actually Works — How quizzes compare with other lead sources for small firms.
- How to Create Customer Personas With AI From Real Data — Base your quiz outcomes on real customer types.
- How to Add an AI Chatbot That Captures Leads on Your Website — The other always-on way to capture website leads.
- What Is AI Lead Scoring and Does a Small Business Need It? — Use quiz answers to decide which leads to call first.
- Can AI Qualify Wedding Leads by Date, Budget and Guest Count? — How AI reads messy wedding enquiries, the five ways dates and budgets get misread, three set-ups by effort, and a back-test on last season's leads.
- How Coaches Use AI to Qualify Discovery Call Bookings — Set fit criteria, ask four questions before the calendar, let AI sort and brief each booking, and send people who aren't ready somewhere useful.
- How to Test a New Service Idea With AI Before You Launch It — A six-week way to test a new service with AI doing the drafting and real customers supplying the evidence, with pass marks set before the results arrive.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: Tally pricing page and Tally Pro help article; Typeform pricing page; ScoreApp pricing page (all checked September 2026).