For the first four years I freelanced as a conversion copywriter, I delivered exactly one hero section per landing page. One headline, one subhead, one CTA, take it or leave it, here's my invoice. The client would pick the version that sounded most impressive to them in the moment — which, I eventually and painfully realized, is almost never the version their actual buyer responds to. The buyer and the person signing my check are different people with different brains. My win rate on A/B tests against the old control hovered right around 50%, which is a generous way of saying my entire creative process was statistically indistinguishable from flipping a coin while charging an hourly rate for the privilege.
The fix, when it finally landed, was embarrassingly simple. It wasn't getting better at writing the one right hero, because there is no one right hero — that's the lie at the center of how most copy gets made. The fix was generating three deliberately different heroes and letting actual visitors vote with their clicks. That's the whole workflow. The Landing Page Hero Section — Three Variations From Your Product Brief prompt is just the thing that makes it fast enough to do every single time, instead of only on the projects where I happen to remember my own advice.
Three buyers wearing the same job title
The prompt generates three fixed angles every time: urgency (what are they losing today by not acting?), aspiration (who do they get to become?), and cold logic (what are the numbers and who else trusts you?). The thing nobody warns you about when you start writing copy is that three people with identical job titles will respond to wildly different angles depending on their situation. The solo founder running out of runway feels the urgency in their chest. The growth-stage CMO with a board to answer to wants the proof. The first-time buyer who's never used a tool like yours is shopping for the aspiration, the picture of a calmer life. Writing one hero means quietly picking one of those three humans and writing the other two off.
And here's the part that should make any copywriter nervous: you usually pick based on which of those three people you most resemble. I'm an urgency person, so I wrote urgency heroes for years and assumed everyone else was wired like me. They are not. The prompt strips my bias out of the first draft by simply generating all three whether I like it or not.
A freelancer tool and a prediction I got embarrassingly wrong
Last month I used it for a project-management tool aimed at freelancers. Gemini 3 Pro Preview returned the three angles as a clean Markdown table in a single pass, which is exactly the format I want here because I can drop it straight into a shared doc and review the columns side by side with the client. The urgency hero led with 'Every late project is a referral you'll never get.' The aspiration one: 'Freelancers who hit every deadline win better clients.' The logic one opened with a proof slot I filled with our real internal stat about how often freelancers miss a deadline. I was certain — genuinely, would-have-bet-money certain — that urgency would win for a freelance crowd, because freelancers live and die by cash flow. The logic version won by 22% on trial signups. I was confidently, measurably, publicly wrong, which is precisely the reason you test instead of trusting your gut.
- All three variations generated in under three minutes, formatted as a table I could review side by side instantly
- The model's Recommendation note correctly predicted logic/proof would win for this audience — a better instinct than mine, which stung
- Giving the client three options to react to produced far sharper feedback than handing them one option to defend or reject
- We test two, not three, to keep the statistical math clean — the prompt's recommendation tells us which two to run
- The proof slot is a literal placeholder, so I never let the model invent a number — I fill it with our real data
The bottleneck was never the testing framework
Here's the honest, slightly humbling part: the testing framework was never my problem. A/B testing a hero is trivially easy and has been for a decade. My problem was the blank page. Writing three good, genuinely distinct heroes by hand took long enough — easily an hour or two of real effort each — that I'd inevitably talk myself into shipping one and calling it 'data-informed instinct.' Now the generation is cheap and fast, so testing three has become my default and 'just ship the one I like' is the rare exception I have to justify. My win rate stopped being a coin flip the day my creative stopped being a sample size of one.
Grab the Landing Page Hero Section — Three Variations From Your Product Brief prompt on Prompt Dock and run it on your very next page. Ship the two angles the model recommends, run the test, and then tell me which angle you were dead sure would win going in. I'll bet you a coffee you're wrong, because I usually am, and being wrong on the page is a lot cheaper than being wrong in your head. When I need to fix the existing headline before rebuilding the whole section, I reach for my Value Proposition Rewriter — Kill the Jargon, Keep the Punch prompt first.