Part 2 of 2. Read Part 1 for the build and the data finding that set this up.
Part 1 ended on a humbling note: our “Help me choose” quiz converted far better than the page average — and almost nobody could find it. Fewer than 2% of visitors ever clicked in. It was a ghost button: a faint outline, on a busy hero, sitting next to a louder call-to-action, with a chat popup covering it on mobile. The tool wasn’t broken. The front door was.
So in the rebuild we did the counterintuitive thing. We resisted the obvious instinct — “make the quiz smarter” — and fixed discovery instead.
Stop improving the feature. Fix the front door.
The data was unambiguous: people who used the quiz loved it. The entire problem was getting them in. So we left the recommendation engine mostly alone and rebuilt the entrance. Three changes did the work.
- Promote the entry to a real primary CTA. The transparent outline became a solid, high-contrast “Help me choose” button — and instead of fighting for attention in a crowded hero, it now appears once the visitor scrolls about a quarter of the way down the page. That timing matters: past the first scan, into genuine evaluation, exactly when a “help me decide” offer is most welcome.
- Trade spec questions for intent paths. The old quiz opened with abstract questions about specs. The new one opens with “What’s your big move?” and a short list of plain-language paths — launch a business, start selling online, scale up, put AI to work, run an agency — each routing to tailored suggestions. You say who you are; it figures out the rest.
- Make the choices recognizable. Real product logos in the options instead of plain text, so each path reads instantly and feels trustworthy rather than generic.
We tagged it like it mattered — because it did
Every new path and state got instrumented and verified on the live page before launch. That’s not bureaucracy; it’s self-defense. A tag that looks right but is wrong silently corrupts the exact adoption numbers you’re relaunching to move — and you won’t notice until you’re staring at a dashboard that doesn’t add up. If you’re going to judge a relaunch by the data, you have to be able to trust the data first.
What moved
Promoting the front door didn’t just lift clicks at the top. It compounded down the funnel. In the weeks after relaunch, week over week:
The quiz didn’t get smarter. More of the right people simply found it — and the gains showed up all the way to revenue, not just at the top of the funnel. That’s the tell of a discovery problem: when you fix the entrance, the wins land everywhere downstream at once.
When adoption is the problem, the highest-leverage fix is almost never the feature — it’s the front door. A great tool nobody can find converts nobody. Treat discovery as a feature in its own right, because it’s usually the one with the most leverage left in it.
In Part 1, the best feature on the page was the one nobody could find. In Part 2, we just opened the door — and let the tool do what it was already good at.
What’s the best thing you’ve shipped that’s quietly underused — and is the problem really the feature, or is it the way in?