Two navigation systems were fighting for the same thumb on the same mobile screen, and the fix that won wasn't a clever new idea — it was a decade-old, publicly available piece of UX research that nobody had actually tested against this specific site (Exp-052).
TL;DR
- A large energy retailer's mobile site had organically grown a super-nav and a separate quick-links row — two full navigation systems stacked on top of each other, each asking the visitor to parse a different set of options before doing anything else.
- The case for consolidating them wasn't an internal opinion. It was checked against Luke Wroblewski's published research on mobile thumb-zone ergonomics and interaction cost, and against the foundational choice-architecture literature behind Hick's law.
- Being right on paper still isn't the same as being right in this funnel — the change went through a live experiment (Exp-052) before it scaled, rather than shipping on the research's authority alone.
- The consolidated navigation won by a 10% to 20% lift, one of the larger margins in this experiment set, and was scaled sitewide.
- The real takeaway isn't the win. It's that even well-established, externally published mobile navigation UX research is still a hypothesis the moment you apply it to a specific product — and the discipline is testing it anyway.
| Navigation state | What the visitor saw | Choice complexity | Outcome |
|---|---|---|---|
| Before (two systems) | Super-nav + separate quick-links row, stacked | Two full option sets to parse before any action | Baseline |
| After (consolidated) | One dropdown menu | One option set, ranked by priority | 10-20% lift, scaled (Exp-052) |
A mobile nav that had quietly become two products
The site's mobile navigation hadn't been designed badly so much as it had been designed twice, at two different points in time, by two different sets of good intentions. A super-nav handled the broad structural map of the site. A quick-links row, added later, sat directly underneath it as a set of shortcuts. Neither piece was unreasonable on its own. The problem was that they had never been reconciled into a single system, so a visitor scanning the top of a phone screen was actually looking at two separate menus and had to decide, without being told, which one was the "real" way to get where they were going.
This is not a cosmetic problem. Primary navigation sits upstream of every conversion path on a mobile site — every product comparison, every plan detail, every signup flow starts with a visitor deciding where to go next. A navigation change carries a risk profile that has nothing to do with how simple the fix looks on a wireframe, which is exactly why cluttered navigation survives in production long after everyone privately agrees it's a mess. Nobody wants to be the person who broke the primary funnel to make the header look tidier.
An internal UX heuristic review flagged the stacked-navigation pattern. So did an independent competitor screenshot comparison. Two separate reviews landing on the same conclusion is useful signal, but it's still an opinion with corroboration, not evidence. That gap — between "we think this is a problem" and "this is measurably working against how mobile users actually navigate" — is where the decision had to go next.
Why Hick's law, not a redesign brainstorm
The behavioral mechanism at stake here has a name: Hick's law, the finding that the time and cognitive effort required to make a decision increases with the number and complexity of the choices in front of you. Two navigation systems stacked on the same screen doesn't just double the amount of visual clutter — it compounds the decision cost, because a visitor has to resolve two separate choice-architecture problems (which menu, then which item) before they've made any progress toward what they actually came to do.
This is where Luke Wroblewski's published research on mobile interaction design and thumb-zone ergonomics matters. His work on how limited screen real estate and one-handed thumb reach concentrate the cost of every additional interaction is not a niche opinion — it's one of the more widely cited bodies of applied research in mobile UX, precisely because it's grounded in stable physical and cognitive constraints rather than shifting design trends. A visitor holding a phone in one hand has less patience, and less working memory, to hold two competing mental models of "where is this site organized" at once.
Wroblewski's research and Hick's law sit on top of a broader foundation: choice-architecture literature — most associated with Thaler and Sunstein's work on how the structure of a decision environment shapes outcomes without removing anyone's options. It's not a mobile-specific idea. It's a lens for any environment where you're asking someone to choose. Mobile navigation is one of the clearest places to apply that lens, because the entire choice environment is compressed into a few square inches of thumb-reachable screen, with no room to hide the cost of a bad structure behind whitespace or a bigger monitor.
Put together, the diagnosis wasn't "the quick-links row is ugly" or "the super-nav is outdated." It was that the site was running two competing choice architectures for the same underlying decision — where do I go next — and charging every mobile visitor the cognitive tax of reconciling them, for free, before they'd done anything else.
The methodology call: publish-first isn't ship-first
Here is the actual senior-practitioner decision in this story, and it's not the redesign. It's what happened between "the research says this pattern is a problem" and "we changed the navigation." There are two easy, wrong moves available at that fork. One is dismissing the external research in favor of internal instinct — "our users are different, our data will tell a different story" — which throws away decades of applied research for no reason other than not-invented-here. The other, less obvious failure, is the opposite: treating a well-known, credible name as sufficient authority on its own. Wroblewski's research is real and well-regarded; that is exactly why it's tempting to cite it once and ship.
Published research, however well-replicated, is calibrated on someone else's product, someone else's audience, someone else's page density. Mobile navigation UX principles rooted in thumb-zone ergonomics and Hick's law hold up across an unusually wide range of products because they're anchored in stable properties of the human hand and attention span, not in aesthetic preference. But "usually holds" is a probability statement about the field in general. It is not a guarantee about this specific stacked navigation, on this specific site, in front of visitors comparing energy plans on their phones.
That's the actual reason this became an experiment instead of a mandate. Running Exp-052 rather than shipping the consolidated navigation outright meant the case for the change didn't rest on Wroblewski's reputation, or on the internal audit's judgment, or on the competitor screenshot. It rested on what real visitors did once the two systems became one. Citing the right research earns you the hypothesis. It does not earn you the ship — that only comes from watching how actual visitors behave once the two navigation systems become one.
This is the difference between a growth function that produces defensible decisions and one that produces confident opinions. Anyone can point at published UX research and declare a fix obvious. Fewer teams build the discipline of treating that research as a starting hypothesis and letting production data decide whether it transfers.
The result: Exp-052
The variant that consolidated the super-nav and the quick-links row into a single dropdown menu won, landing a 10% to 20% lift — one of the larger margins across this experiment set. The design scaled sitewide.
That is the entire result, stated with the precision the data supports and no more. What matters more than the size of the number is what it confirms: an externally published, well-established UX principle, tested directly rather than assumed, held up under real production conditions instead of staying a theoretical best practice cited in a slide deck.
The finding that actually matters
If the goal here had only been "make the mobile navigation less cluttered," a competent designer could have shipped that redesign on judgment alone, with no experiment required. What makes Exp-052 worth writing up isn't the redesign — it's that a genuinely well-established, externally validated principle was still treated as a hypothesis rather than a fact, and it happened to hold up here as cleanly as the textbook version of the argument.
That "held up" part is doing real work, and it's not something to take for granted. Plenty of published UX guidance doesn't transfer cleanly into a specific production context — a different audience, a denser page, a different level of trust in the brand, a different reason visitors showed up in the first place can all break a principle that reads as universal in a research summary. The discipline was treating a decade of externally validated mobile navigation UX research as a hypothesis worth testing rather than a fact worth assuming — and being willing to find out it didn't transfer here, even though, this time, it did.
That's the pattern worth taking away from Exp-052 more than the specific lift number: the research earns your attention, not your default. Testing it is what turns "this is probably right" into "this is right, here, for these visitors" — and that distinction is exactly what separates an experimentation program from a design opinion with a citation attached.
FAQ
Why test an established UX principle instead of just shipping it?
Because "established" describes how often a principle holds across contexts in general, not whether it holds in this specific one. Hick's law and Wroblewski's thumb-zone research are about as well-supported as mobile UX guidance gets, and the redesign still went through Exp-052 before scaling. The research changes what you expect to see. It doesn't replace watching what actually happens.
How do you know the two navigation systems were actually a problem before running the experiment?
An internal UX heuristic review and an independent competitor screenshot comparison both flagged the stacked-navigation pattern on their own. That's useful corroboration for forming a hypothesis worth testing — it's not the same as evidence that the fix will work, which is exactly why it became an experiment rather than a straight redesign.
What made this margin larger than most other experiments in the portfolio?
Primary navigation touches every session on the site, not a single page in a funnel, so a genuine reduction in choice complexity there compounds across the whole mobile experience rather than a single conversion step. That breadth of exposure is part of why navigation changes carry outsized risk when they're wrong, and outsized upside when the underlying mechanism is correct.
Does this mean any "combine two navigations into one" redesign will win?
No, and treating it that way is exactly the mistake this experiment was designed to avoid. Hick's law and choice-architecture effects are context-dependent — they show up strongly where visitors are actively trying to navigate and weakly where motivation or intent is low. The result here is evidence for this site, this audience, this specific double-navigation pattern. It's a strong prior for a similar setup elsewhere, not a guarantee.
What's the takeaway for a founder or CMO evaluating their own site's navigation?
If your mobile navigation UX has grown feature by feature rather than being designed as one system, there's a reasonable chance you're running two choice architectures at once without realizing it. The fix isn't to trust a redesign brief or a competitor teardown on faith — it's to treat the published research as a strong hypothesis and confirm it against your own visitors before you scale it.
Bottom line
A stacked mobile navigation was flagged by internal review and competitor comparison, then validated against Luke Wroblewski's published research on thumb-zone ergonomics and the choice-architecture literature behind Hick's law, before the consolidated design was tested in production rather than shipped on the research's authority alone. The variant won by 10% to 20%, one of the larger margins in this experiment set, and scaled sitewide. The number is the outcome. The finding worth remembering is that even decade-old, externally validated UX research doesn't get a free pass into your funnel — it gets tested, the same as any other hypothesis.
If your team is carrying navigation, pricing, or onboarding decisions on internal opinion alone — or citing external best practices without ever confirming they hold up in your own product — that gap between assumption and evidence is exactly what an experimentation program is built to close. If you want a second set of eyes on where that gap exists in your funnel, reach out to talk about working together or explore what an experimentation program built around this kind of methodology would look like for your team.
Evidence sources and free next step
In my experience, the most reusable part of this case is the comparison between external guidance and a local production test. Baymard's mobile navigation research provides a public mechanism to compare with the first-party result. Use the evidence-graded A/B testing examples and navigation test plan to judge transferability. Then try GrowthLayer free to document the control, consolidated treatment, primary metric, and guardrails.