The first step? Step back.
That’s especially important if you’re close to the system, because when you’ve used it a hundred times, you stop seeing the friction.
A broken flow doesn’t always scream; sometimes it just feels off.
So open your phone. Pretend you’re a new user trying to accomplish one specific task. What do you notice? What’s frustrating? What’s not happening the way you’d expect?
Then say it out loud:
“Hey, this isn’t really working.”
It sounds obvious, but saying it shifts you from passively noticing problems to actively solving them.
From there, I start digging into the data. Does it back up what I’m feeling? Are users dropping off where I got stuck?
A real example: Yakima’s system pages
This happened to me recently when I realized our system pages weren’t flowing the way they should. The process is supposed to be simple:
You enter your vehicle’s make and model, answer a few fit-related questions, choose to build out a system, and then see only the products that are compatible with your car.
But with hundreds of vehicle fit combinations and edge cases, the backend logic had gotten… messy.
So we mapped it out
We went through the full flow as if we were first-time users. Screenshotted every step. Recorded screen actions. Paid attention to how things looked, what we expected to happen, and what actually happened.
We asked:
- What does adding a vehicle look like?
- What happens when I click a product?
- Is the “Add to Cart” button where I expect it to be?
- What parts feel confusing or awkward?
You don’t always need a full redesign. You just need to walk through it with fresh eyes.
Sometimes it’s a broken link. Sometimes it’s a clunky dropdown or too many steps.
Clarity comes when you slow down and at each interaction.
Then, make a plan
Once we understood the flow, it was time to fix it. That meant asking:
- What needs to be improved?
- What can be simplified?
- What should we leave alone?
The UI was refined to make the flow feel smoother and more intuitive, with an emphasis on reducing friction and confusion.
And finally, test it
Test it on your phone. On your computer. On your coworker’s five-year-old iPad. On anything you can find.
Break it. Try weird edge cases. Try adding the wrong car. Try skipping a step. You’re not done until it works in the messiness of real life, not just the perfect version.
After launch, watch the data. If you got it right, you’ll see the difference: lower bounce rates, higher conversions, fewer confused customers.
When the flow is better, you can feel it. And the numbers should prove it.