Your lead magnet is still a PDF. Here's what Canva quietly handed every free account this month.
The interactive calculator your funnel needed used to be a developer invoice, now it's a prompt in a tool you already have.
TLDR: On July 14, Canva pushed Canva Code 2.0 to every account on every plan, free included. You describe a calculator, quiz, or scorecard in plain English, Canva builds the working thing, you edit it like any other design inside the editor where your brand kit already lives, and you publish it to a real URL. The interactive lead magnet, an asset that used to mean a developer invoice or one more monthly subscription, is now a prompt inside a tool your team already logs into. By the end of this you'll have a copy-paste prompt that turns the one question your leads keep asking into a working interactive tool, plus the pre-traffic check that catches the embarrassing failures before your audience finds them.
Three weeks ago I wrote about Canva folding the ad-refresh loop into a button, and the 30-minute test that shows whether a creative retainer still earns its invoice. I filed it under Zapier Optionality, the pattern where software you already pay for quietly absorbs a job you were paying somebody else to do, and I figured Canva was done collecting other people's invoices for the month.
It wasn't. On July 14, Canva made Canva Code 2.0 generally available to everyone, on every plan, free accounts included. You type a prompt, Canva generates a working interactive tool, a quiz that scores, a calculator that computes, a page that reflows on a phone screen, and then you edit the result like any other Canva design and publish it to a free Canva URL or your own domain.
The job getting absorbed this time is the interactive lead magnet, the calculator or scored quiz that used to sit behind a freelance developer, a Typeform-style subscription, or a "we'll scope it next sprint" conversation with an agency. If your lead capture is still a static download, the build cost excuse retired on July 14, and what's left is a decision about what to build.
Here's what shipped, what I'd build with it first, and the check I'd run before putting a dollar of traffic behind it.
The July 14 change: Canva Code 2.0 went to every plan including Free, after a preview run that was gated to the first million users who found an easter egg at Canva Create in April.
The Static Magnet: a PDF download tells you somebody wanted a file, a three-question calculator tells you what they're planning to spend.
One prompt, one working tool: the full build prompt to copy, tuned for a lead magnet that solves one narrow job in under two minutes.
The 10-minute pre-traffic check: what to test on your phone before this thing goes in a bio link or behind ad spend.
Who this is for: operators whose lead capture is a static download and whose booking page could use a warmer front door, which covers most call funnels, local service businesses, and agencies I talk to.
What Canva shipped on July 14
Canva Code launched in its first version last year, and Canva says users built more than six million coded sites with it, which told them the demand ran a lot deeper than developers. Version 2.0 arrived at Canva Create in April as a preview limited to the first million users who cracked an easter egg, and on July 14 it went to the full user base, per Canva's announcement and TechRadar's coverage of the general release.
The mechanics, per 9to5Mac's writeup: start from a plain-language prompt, one of 50-plus templates, or an HTML file imported from anywhere, including pages another AI tool generated. The output lands in the normal Canva editor, so you swap images by dragging from the library, change colors and fonts from the toolbar, click into any section and retype the text, and publish to a free Canva domain or a custom one, with responsive layouts and a mobile preview before it goes live. Canva's own line on it: "Canva Code sits inside the same intuitive editor where you already house your brand kit."
The part I'd flag for anyone who tried the first version and bounced: editing no longer means re-prompting the whole build. Canva spent the spring criticizing AI design tools for being AI-first and design-second, where the smallest tweak meant another prompt and an entirely new result, and the 2.0 pitch is that you fix the button color the way you'd fix it on a flyer. Danny Wu, Canva's Head of AI Products, framed the shift to TechRadar this way: "With the barrier to generating code lowered, the value has shifted to visual control: what you build needs to be distinctive and look like your brand or vision."
Canva also reports generation time down 75%, median prompt-to-publish time down 30%, and active Code users up 25% since the feature moved into the main UI. Worth holding loosely: every one of those is Canva measuring Canva. The Crescitaly team's read matches mine: "Treat speed as production capacity, not business impact." Faster builds mean you can test more ideas, and they prove nothing about whether any given idea converts.
The Static Magnet
Here's the name for the thing this replaces. The Static Magnet is the PDF checklist, the swipe file, the ebook that captures an email and then tells you nothing else about the person who grabbed it, because it asked them to do nothing except download, and an asset that asks for no decisions collects no intent. Every download looks identical in your CRM, the tire kicker and the operator with budget land in the same nurture sequence, and your follow-up is generic because the asset gave you nothing to personalize with.
An interactive asset works the opposite job. A three-question calculator makes the reader type in their monthly ad spend, their average deal value, their close rate, and now the "lead magnet" has handed you qualification data before a single email went out. A scorecard makes them self-assess, and the score tells you who's ready for a call this week versus who needs three months of nurture. Crescitaly's framing of why this works on social traffic is the cleanest I've seen: an interactive asset turns a fast scroll into a small decision, and a small decision reveals real intent.
I sell from behind one of these myself. The AI IQ Test is a scored quiz, and the value over a PDF is the same value I'm pointing you at here: the answers tell me where an operator stands before I've written a single follow-up, so the follow-up gets to be specific instead of generic. That asset was a real build project when it went up. As of this month, the same category of asset is a prompt.
Pick the job before the format
The move: find one question your audience already asks on repeat, and build for that question only. Go mine the places the questions already live: your DMs, your sales call recordings, your inbox, the comments on whatever post performed best this quarter. You're looking for a question with commercial weight behind it, the kind where the answer changes what somebody buys.
The mapping (adapted from Crescitaly's planning model, which mostly matches how I'd sort it):
"How do I even start with X?" on repeat -> guided checklist, outputs a personal action plan
"Should I do X or Y for my case?" -> 3-question decision quiz, outputs a recommendation
"What would this cost / return?" -> simple calculator, outputs a planning range
"Are we ready for X?" -> scorecard, outputs a gap read that earns the audit call
What a correct pick looks like: you can write the result as one sentence before you build anything. "You're leaving roughly $X on the table from no-shows every month" is a buildable result, and so is "Your setup scores 6 out of 10 on AI readiness, here are the two gaps." If the result needs a paragraph, the tool is too broad.
The failure mode: picking a format because it demos well. A gorgeous interactive experience wrapped around a question nobody was asking converts worse than the ugly PDF it replaced, because at least the PDF answered something. Format follows question, every time.
Walk away with: one repeated audience question and a one-sentence result definition, written down before you open Canva.
The build prompt
The move: open Canva, start a Code build, and paste this with your blanks filled. It's structured to force the constraints that make lead magnets convert, few inputs, fast result, one call to action.
Build a [FORMAT: calculator / 3-question quiz / scorecard] for
[WHO: e.g. agency owners who book sales calls from paid traffic].
The one job: answer "[THE QUESTION YOUR LEADS KEEP ASKING]"
in under 2 minutes.
Inputs: no more than [3-5] fields.
[LIST THEM: e.g. monthly ad spend, average deal value, close rate.]
Logic, in plain English:
[e.g. booked calls = ad spend / cost per booked call.
Revenue = booked calls x show rate x close rate x deal value.
Show the gap between their current number and a 75% show rate.]
Result screen: one sentence in this exact shape:
"[THE SENTENCE THE USER WALKS AWAY WITH, number filled in.]"
Below it, one button labeled "[CTA TEXT]" linking to [URL].
Style: [BRAND COLORS + FONT], mobile first, large tap targets,
no intro screen, no email gate before the result.
Do not include: fake precision (round all outputs), disclaimers
longer than one line, or any input field I did not list above.
What a correct output looks like: a working preview you can poke at inside the editor, where the math holds when you test it against numbers you already know, and where the branding swap takes minutes in the toolbar instead of another generation. Under the hood Canva's code generation runs on Anthropic's Claude models, per vibecoding.app's review, and simple bounded logic like this is squarely inside what it does well.
The failure mode: the first generation comes back generic or the logic bends somewhere. The fix is scoped follow-up prompts aimed at one element, "the result rounds to the nearest hundred," "make the second question a slider," rather than regenerating the build. And check the arithmetic by hand with two known scenarios before you trust it, because a calculator that's wrong is worse for your credibility than no calculator.
Walk away with: a working, on-brand interactive tool built from one prompt and a handful of scoped edits.
The 10-minute pre-traffic check
The move: before the link goes in a bio, an email, or an ad, run this check start to finish. Interactive assets have more ways to break than an image, and the failures are public.
THE 10-MINUTE PRE-TRAFFIC CHECK
1. Run it on your phone, then somebody else's phone.
Thumb-only, no pinch-zooming required.
2. Feed it garbage: zeros, blanks, a $999,999,999 budget.
It should fail politely. No crashes, no "NaN."
3. Read the result screen out loud. A planning aid reads
like a planning aid. If it sounds like a guarantee,
soften the copy.
4. Click every button. The CTA lands where it claims,
with your tracking parameters intact on the URL.
5. Time a cold run, open to result. Over 2 minutes,
cut an input field.
6. Confirm where responses land (Canva Sheets can collect
form responses) and that a named human checks it.
7. Ask: does the result state what data is stored and
keep the email field optional? Never collect more
than the follow-up needs.
What a correct output looks like: a stranger on a phone gets from your post to a useful, honest result in under two minutes, and you can see the completion in your response sheet with enough context to write one specific follow-up line.
The failure mode: shipping on desktop-preview confidence. Nearly all of this traffic is mobile, and the version of you that built the thing is the worst possible QA tester because you already know how it's supposed to work. Hand your phone to someone who's never seen it, and watch where their thumb hesitates.
Walk away with: a tested link you can put spend behind without wincing.
Where this goes wrong
The novelty build. The tool exists because the builder was excited, so it answers no recurring question, and it gets impressive engagement from people who were never going to buy. The control is boring and useful: Crescitaly's team suggests keeping the same offer live as a plain page or PDF, and giving the interactive version more budget only if it wins on qualified action.
The twelve-field intake. Somebody on the team wants "just one more question" until the calculator turns into an application form, and completion rate falls off a cliff. A social click is fragile, the result has to arrive before the patience runs out.
The overcertain score. A marketing scorecard that reads like a diagnosis creates trust problems and, in some industries, compliance problems. Round numbers, hedged language, one-line honesty about what the tool can't know.
The orphaned inbox. Responses collect in a sheet nobody owns, and the hottest lead of the month sits unanswered for nine days. The asset is the cheap part, the follow-up system it feeds is where the revenue is.
The honest tradeoffs
This is a widget engine, and the ceiling is real. Everything runs client-side, so there's no user login, no database behind it, and nothing that needs server logic, which rules out anything past a bounded tool. The vibecoding.app review puts it plainly: "You cannot build apps. You cannot export code." Your build lives inside Canva's ecosystem, HTML comes in but doesn't come out, and that matters if you ever want to move the asset somewhere Canva isn't.
AI credits meter the whole thing, shared across Canva's other AI features, and free accounts get the tightest limits, so a heavy build week can hit the ceiling before the month resets. The speed and adoption numbers are Canva's own measurements with no independent audit. And generated logic still deserves a human check every time, one operator-focused review put the risk well: "Generated code may contain weak logic or hidden errors," and "publishing speed can discourage proper review." For a prototype or a lead magnet, none of this is disqualifying. For checkout flows, client portals, or anything holding sensitive data, this is the wrong tool and a real developer is still the answer.
Recap
Canva Code 2.0 went to every plan on July 14, which moved the interactive lead magnet from a build-cost decision to a judgment decision. The judgment part: mine your DMs and sales calls for one repeated question with money behind it, define a one-sentence result, build the smallest version with the prompt above, run the 10-minute check on a phone that isn't yours, and measure it against your existing static download on qualified action, keeping the winner. The Static Magnet era of lead capture didn't end because PDFs stopped working, it ended because the thing that outperforms them stopped costing anything to try.
If you want to feel the difference from the lead's side before you build one, take the AI IQ Test. It's a scored quiz I run in front of my own offers, it takes about three minutes, and you'll see the mechanics from the receiving end, the bounded inputs, the specific result, the one next step. Steal the structure for your own.
Reply with the one question your leads keep asking you, and I'll tell you which format I'd build for it.
Andrew
P.S. If you're building this week: the 30-minute Canva Grow test covers the ad-creative side of the same absorption pattern, Zapier Optionality is the original writeup on software eating its neighbors' invoices, the three-prompt Google Business Profile check does the same commoditization math on listings retainers, and the full archive lives at muddventures.substack.com.