From Pogo-Sticking to Confident Navigation

Problem

Analytics told a clear story: over the past year, the site had seen a 9% YoY decrease in visitors and a 7% YoY increase in bounce rate. But the data that stood out most was the pattern of pogo-sticking — users repeatedly bouncing between mid-funnel pages without progressing toward a product or conversion.

This wasn't random drop-off. Users were landing on pages, hitting a dead end, backing out, and trying again. Something in the journey was actively working against them.

My Role

I led this project end-to-end as the sole designer — from initial discovery and research through concept, testing, stakeholder presentation, and handoff to engineering.

Discovery

Starting with the data

I began by mapping the end-to-end user flow from the homepage through the site's conversion-ready pages. What I found was that several above-the-fold CTAs were pulling users sideways — redirecting them away from natural progression in the buying journey rather than moving them forward.

A/B Tests

To validate this, we ran an A/B test on the page most commonly disrupting the buying journey. Removing the distracting CTAs increased click-through rate by 1.43% — a small but meaningful signal that confirmed users were being pulled off course by competing calls to action.

But the data pointed to a deeper issue. The CTA problem was a symptom. The navigation itself was the root cause.

Listening to users

We conducted user interviews to understand what was actually happening from the user's perspective. What we heard was consistent: users struggled to orient themselves within the product catalog. The existing navigation didn't reflect how they thought about products — and finding something as fundamental as a hardware module required too many steps.

Two moments from the sessions stuck with me:

"What you've done here is perfect because you've organized it both ways — by technology and by hardware."

"I can get to the modules in 1 click. I call that a huge improvement."

Both quotes pointed to the same insight: users needed the navigation to match their mental model, not Silicon Labs' internal taxonomy. They think in terms of product type and technology category — and the navigation needed to support both paths simultaneously.

Framing the Problem

With research complete, I defined the core design challenge as:

How might we restructure the navigation so users can find relevant products through their own mental model — whether they're starting from a technology or a hardware category — in as few steps as possible?

The Solution

The redesigned navigation introduced a dual-organization system: products could now be browsed by technology (wireless, IoT, etc.) or by hardware (modules, SoCs, kits), giving users two valid entry points rather than forcing a single path.

The addition of a dedicated Hardware category was a key decision that came directly from the research. It didn't exist in the original navigation — users had to dig for it. Surfacing it at the top level removed a significant point of friction.

Key changes included:

• Restructured top-level nav to surface both technology and hardware as parallel entry points

• Reduced click depth to reach product modules from multiple steps to 1 click

• Removed redundant or misdirecting CTAs from mid-funnel pages

Validation

Before moving to development, I ran a preference-based A/B test on the redesigned navigation against the current experience.

90% of users preferred the new navigation.

Stakeholder response reinforced the direction. After presenting the findings and proposed solution, the team aligned quickly — the research made the case clearly enough that the path forward wasn't in question.

What’s Next

The redesigned navigation is currently in development. Post-launch, the metrics to watch are bounce rate recovery, reduction in pogo-sticking on mid-funnel pages, and click-through rate on product pages.

What I'd Do Differently

With more time, I would have pushed for a card sorting exercise earlier in the process — before the navigation concepts were formed — to more rigorously validate how users naturally group products. The interview insights were strong, but a structured card sort would have given us quantitative backing for the taxonomy decisions, not just preference data on the final designs.