Do styling A/B tests work?

Sometimes — and this is the most-run change type by a wide margin. Styling changes win 15% of the time. That is mid-pack. What is not mid-pack is the volume: styling accounts for more tests in this data than any other category, and it has a median lift of +15.0% when it does land.

Most-tested, least-rewarding per win. Styling is often what gets tested when there is no hypothesis — it is easy to build and easy to justify after the fact.

The numbers

Tests analysed	1,000+
Beat control	15%
No measurable difference	70%
Lost to control	16%
Median lift when it won	+15%

Typical winning range: +8.5% to +31.5% for the middle half of winners. Median traffic per variant was 3016 visitors. Most ran on homepages (596), product pages (296), other pages (142).

What counts as a styling test

Visual treatment shifts — spacing, weight, borders, colour, shadows — with the same content in the same order. Move the content and it is a layout test. Rewrite it and it is a copy test.

Control and variant wireframe for a styling A/B test — Control on the left, variant on the right. Only the changed element is highlighted.

Adjacent categories, and how often they win: layout (19%), hero image (17%), CTA copy (12%).

Styling tests by page type

page type	win rate	tests
homepages	15%	500+
product pages	14%	250+
category pages	16%	100+
lead capture pages	10%	50+
landing pages	11%	50+
checkout pages	11%	47 tests

Only page types with at least 30 styling tests appear here.

Styling tests that won

Moves the news badge into the share buttons container, strips its text nodes, and restyles it with flex layout, smaller max-width, transparent background, and resized image without changing wording
Adjusts top padding of main container, resizes an h1 headline text, and shrinks font size of text inside some buttons
Sets a fixed height of 760px on an element, changing its layout size without altering content
Adjusts hero H1 spacing/size and restyles the trust row, replacing award badge images with a star-rating/review-count badge (visual/layout change, no wording change).
Repositions and resizes a fixed floating element on the page without changing any text

Styling tests that did not

Makes header and nav bar sticky and shrinks/resizes header elements (logo, search input, nav links) on scroll
Hides SKU text, adds a page-load hiding overlay/anti-flicker mechanism, and styles buyer-counter and countdown elements without changing their wording
Hides an existing button (second link's button in the second div) from view
Changes text color to green and adds bottom padding on the targeted element, no wording change

These are individual tests, not rules. Each ran on one site, with one audience, against one page we are not showing you. They are picked to be illustrative rather than sampled at random, and a change that won here can lose on your page for reasons none of this captures. Read them as prompts for what to test, not as findings to copy.

Why the rate looks like this

A styling change is usually the smallest real intervention available: the visitor sees the same things saying the same thing in the same order, only dressed differently. Small interventions produce small effects, which is what the lift figure shows.

How this compares

change type	win rate	tests
Layout	19%	351
Hero image	17%	160
Form	16%	81
CTA colour	16%	32
Styling (this page)	15%	1414
Price framing	13%	117
CTA copy	12%	315
Body copy	11%	337
Split URL	11%	1164
Headline	10%	918
Social proof	9%	196

FAQ

Why do so many people test styling?

It is the cheapest variant to build and needs no research to justify. That is a statement about cost, not about expected return.

Is it worth testing at all?

At 15% it is not worthless. But if the same effort could go into a layout or hero-image test, those carry better odds and larger wins.

Who ran these tests

This is every Mida account that ran a readable test — in-house marketers, founders, product teams, and agencies working on client sites. Nothing here is filtered by who ran the experiment or how experienced they are.

Low win rates are normal in experimentation, including at the top end. Microsoft's experimentation team, reporting on its own platform, found that only about one third of ideas improve the metric they were designed to improve — and that roughly another third actively hurt it. That is a dedicated experimentation organisation with research, prioritisation and review behind every test.

A mixed population like this one runs below that. The gap is roughly what disciplined practice buys you: ideas grounded in research rather than opinion, one variable at a time, and tests built so the result can actually be read.

So treat these as a general base rate, not a target. If you would rather not close that gap the slow way, find a CRO agency in the United States, the United Kingdom, the Netherlands or Australia — or browse every region.

A win is a variant that beat its control on that test's primary goal with a statistically significant result. Tests that never got enough traffic to say anything either way are excluded. The methodology has the full detail, including what these numbers cannot tell you.