Reordering three prices beat a full pricing-page redesign, and the margin wasn't close. The useful lesson is not that every team should copy the order; it is that a cheap, diagnostic change can deserve priority over an expensive package redesign.
TL;DR
- A retail energy pricing page wasn't underperforming because of its design — it was underperforming because of an unmanaged comparison (Exp-055).
- The fix cost almost nothing to build: show all three price points instead of a subset, badge one specific mid-tier bundled plan, and make that plan surface first when it's available.
- This is a textbook decoy effect pricing setup — asymmetric dominance guiding a three-way comparison toward a specific anchor.
- Result: a 10% to 20% lift, one of the stronger margins in this experiment set. The variant shipped.
- The real lesson isn't the tactic — it's the sequencing: test the cheap, surgical choice-architecture fix before the expensive redesign, not after.
| Control | Variant | |
|---|---|---|
| Price points shown | Subset of available plans | All three price points |
| Mid-tier bundled plan | No visual distinction | Callout badge |
| Display order | Standard / unmanaged | Bundled plan surfaces first |
| Outcome | Baseline | +10–20% lift, shipped |
The instinct in the room is almost always "redesign it"
When a pricing page underperforms, the first idea in most planning meetings is a rebuild: new layout, new visual hierarchy, new copy, sometimes a new component library. It feels proportionate — a page that isn't converting looks like a design problem, and design problems get design solutions. I've sat in enough of those meetings to know the instinct isn't stupid. It's just usually wrong about where the decisive constraint actually is.
The page in Exp-055 was a product chart for a retail energy brand — the page where a household compares multiple price-point plans and picks one. The plans themselves weren't the issue; margins and eligibility rules were fixed upstream. The open question was purely architectural: which options does the page show, in what order, and does anything about the layout signal which one is the sensible default. That's a decoy effect pricing problem, not a design problem — and those two problems get solved with entirely different tools.
We didn't touch the visual system. We changed three things: which price points were visible, whether one option carried a callout, and what order they appeared in. Everything else about the page — typography, layout, imagery — stayed exactly as it was.
Why the decoy effect works on a three-option page
The behavioral mechanism here is asymmetric dominance, better known as the decoy effect: when people choose between options, they rarely evaluate each one against some absolute standard of value. They evaluate it _relative to the other options in front of them_. Introduce or reposition a third option that's clearly dominated by one of the first two, and you don't just add a choice — you change which of the original two looks like the obvious pick.
The finding traces back to Huber, Payne, and Puto's original 1982 research on asymmetric dominance in choice sets, and it's the same mechanism Dan Ariely popularized with his magazine-subscription example — where adding a strictly-worse "decoy" option made a specific target option look far more attractive by comparison, without that target option changing at all. The point in both cases is the same: comparison sets are not neutral backdrops. They're an input to the decision.
Applied to a three-tier pricing chart, this means the choice a household makes is shaped as much by which plans are shown together and in what order as by the plans' actual terms. Showing all three price points instead of a subset gave people the full comparison set the decoy effect needs to work. Badging the mid-tier bundled plan gave the eye a place to land first. Making that plan surface first in display order — when it was available — reinforced the same signal through position, which carries its own weight independent of the badge. None of this changed what any plan cost or included. It changed how the three were read against each other.
Why we tested the cheap fix before the expensive one
This is the part that's actually a methodology lesson rather than a lucky guess. A redesign and a choice-architecture change are not two versions of the same idea — they're different bets with very different costs, and the order you test them in should follow from that, not from which one feels more thorough.
A redesign touches layout, component code, copy, QA, and usually a design review cycle. It's expensive to build, expensive to revert if it underperforms, and it bundles multiple changes into one variant — so even a clean win doesn't tell you _which_ change did the work. A reordering-and-highlighting change touches none of that. It's a display-logic change to an existing chart: same components, same copy, same visual system. It's cheap to build, cheap to revert, and because it isolates one mechanism — the comparison set itself — a win tells you exactly what moved the number.
The senior judgment call wasn't "try the decoy effect instead of a redesign" — it was recognizing that the cheap, isolating experiment belongs first in the sequence whenever it's available, because it either resolves the problem outright or narrows what the expensive experiment needs to test next. Reach for the redesign first and a win leaves you unable to say why it won. Reach for the surgical change first and a loss still tells you the problem lives somewhere other than the comparison set — which is information a redesign alone would never have isolated. Sequencing cheap-and-diagnostic before expensive-and-bundled isn't caution. It's how you avoid spending a redesign's budget to learn something a smaller experiment would have told you for a fraction of the cost.
What happened
The variant won. Over roughly five weeks, the reordering-and-highlighting change produced a 10% to 20% lift (Exp-055) — one of the stronger margins across this experiment set, and directionally consistent enough across the run that we scaled it rather than treating it as a fluke. It shipped as the new default. No redesign followed it, because none was needed: the comparison set, not the page's visual design, had been the constraint the whole time.
That's the outcome that matters most to anyone deciding whether to greenlight an experimentation program: the win came from the cheaper of the two available bets, and it resolved the underlying question without requiring the more expensive one to also be built and tested.
The pattern most teams don't check for
Here's the diagnostic catch, and it's the reason I bring this experiment up when people ask what a mature testing program actually looks for. Most teams treat an underperforming pricing page as a visual-design problem by default — new layout, new hierarchy, sometimes a full rebuild — without first asking whether the _comparison_ being presented is doing any work at all. This result is evidence that the order and relative positioning of existing options often carries more conversion weight than the visual design of the page itself. A page can be beautifully designed and still lose conversion because the three things on it aren't arranged to guide a decision — they're just arranged.
The catch most teams stop short of isn't running the decoy effect pricing experiment. It's asking the question that leads to it: before redesigning, has anyone actually tested whether the existing options, shown differently, already solve this? Most audits never get that far, because "redesign" is the default fix reached for before the comparison set itself has been examined.
FAQ
Isn't this just manipulating people into paying more?
No — nothing about the plans, their prices, or their terms changed. The experiment changed which existing, legitimate options were shown and how they were ordered. Every plan a household could pick was one they were already eligible for at the same terms; the change was in how clearly the comparison was presented, not in what was being sold.
How do you know the lift came from the ordering and not something else on the page?
Because the change isolated exactly one variable: the comparison set (which plans, badged how, in what order). Nothing about layout, copy, or visual design moved between control and variant. When a single-variable change produces a directionally consistent lift across the run, you can attribute it to that variable with real confidence — which is precisely why testing the isolating change before a bundled redesign matters.
Why not just run the redesign and the decoy effect pricing change together?
Bundling them would have meant a win with no way to attribute it — was it the layout or the comparison set that moved the number? Testing them separately, cheapest first, is what let us say with confidence that the comparison set was the lever, and that a redesign wasn't a prerequisite for the win.
Does this generalize past energy pricing pages?
The mechanism generalizes; the specific setup doesn't. Asymmetric dominance shows up anywhere someone is comparing a small set of priced options — SaaS tiers, e-commerce bundles, service packages. Whether reordering or highlighting is the right lever, and which option should anchor the comparison, is a call that has to be tested in each specific context, not assumed from this result.
How long should an experiment like this run before you trust it?
Long enough to see the effect hold up across enough of the natural variation in traffic and behavior that it's not just a good week — in this case, on the order of several weeks. The goal is a result that's directionally accurate and consistent enough to act on, not a single point-in-time reading.
Bottom line
A pricing page that isn't converting is usually diagnosed as a design failure and treated with a redesign. Exp-055 is evidence that the cheaper, more diagnostic question — is the comparison itself doing its job — deserves to be asked and tested first, because when the answer is no, the fix costs a fraction of a redesign and the result tells you precisely what moved the number.
If your pricing page has plateaued and the default plan in the room is "let's redesign it," that's usually the moment to test something smaller first. I design and run experimentation programs for teams that want evidence before they commit budget to a rebuild — get in touch if you want a second opinion on where your next test should actually go.
Evidence sources and free next step
The published choice-overload meta-analysis is useful background, but the production result remains first-party evidence with its own boundaries. Compare the pricing page optimization review and A/B testing examples. Then get started with GrowthLayer free to sequence the cheap diagnostic test before the redesign.