The loop after launch.
Your first build is a starting point. Real-world signal turns it into a real product.
The loop, on one page
One real signal. A bounce, a quote, a screenshot from a friend.
One specific edit. Sentence-sized prompts beat paragraph rewrites.
Re-send the link to a fresh handful of people. Watch their first move.
Did it move the needle? Keep it. If not, undo and try the next thing.
How to write a good iterate prompt
Good prompts are specific, small, and measurable. Hit all three and you usually get a result you can keep.
Name the page and the section. Not just "make the site better".
One section, one change. Combining four edits hides which one helped.
You can tell within a day whether it worked. If not, narrow the scope.
- Make the pricing section the second section on the page.
- Replace the hero subhead with one line about deliverability.
- Add a three-question FAQ above the contact form.
- Change the CTA button text from "Learn more" to "Get a demo".
- Make the whole thing nicer.
- Use better colors.
- Add more conversion.
A four-week rhythm that compounds
Replace one image with a real one. Fix the headline to name your specific audience. Test on three phones.
Watch where readers stop scrolling. Move the most-asked answer above that fold.
Add one quote, one number, one logo. The cheapest trust upgrades you can make.
Rewrite the CTA. Test verb-first text. Move the button if it sits below the fold on mobile.
When to stop iterating
Two signals to stop:
- You cannot describe the problem. If you cannot finish "people are bouncing because...", go talk to five users instead.
- Three small edits in a row did not move the metric. The problem is probably the offer, the audience, or the channel, not this page.
Open your latest run and iterate.
One small ask at a time. The compound effect after 8 to 10 prompts is huge.
Sharing + domain