Social Proof A/B Tests: What 170 Experiments on Shopify Stores Show

A social proof A/B test checks whether showing that other people bought, used or endorsed a product changes how many visitors buy. Across 170 social proof experiments DTC Pages ran on Shopify stores between 2021 and 2026, 37.1% won and 19.4% lost at a 90% threshold: 1.91 wins for every loss, against 1.61 across all 1,379 tests we have run.

Social proof wins almost two tests for every one it loses#

Social proof is one of the key elements to improve conversions in any store. People like to see people. It comes up with almost every client we work with, and most of them underestimate how much it does for the store. Every brand knows reviews and testimonials matter, but they often overlook other elements that also build trust: press mentions, customer counts, expert endorsements or live purchase activity.

Of the 1,379 experiments in our full record, 170 were about social proof. 63 of them won and only 33 lost, 1.91 wins for every loss, against 1.61 across everything we test. Social proof loses less often than the rest of what we test, which makes it one of the safest bets a store can make.

The rest of this post covers which types of social proof work, where on the site they work best, and what the losing tests got wrong.

How we counted this#

We went through every experiment with a recorded verdict that we ran between 2021 and 2026, one by one. We consider a test a social proof test if the variant showed visitors evidence that other people bought, used or endorsed the product: reviews and ratings, written and video testimonials, UGC, press mentions, customer counts, expert endorsements and live purchase activity. Trust badges, guarantees and payment icons are not included, since they are claims made by the store, not by other customers. A test only counts if social proof was the main change. When it was one part of a bigger redesign, we left it out, because there is no way to know which part moved the result.

A test is considered a winner when the variant reached a 90% probability of beating the control on conversion rate, average order value, revenue per visitor or profit per visitor, overall, on one device, or on a metric declared for that hypothesis. Losses work the same way: the control has to reach a 90% probability of beating the variant. Tests where we removed social proof are covered separately below and are not part of these numbers. You can go through the full method in our A/B test win rate benchmark.

Which types of social proof work#

Reviews are the most tested type and one of the most reliable. Across 89 review tests, the wins outnumbered the losses two to one. The strongest single cut is reviews inside the checkout: 7 wins and 1 loss in 16 tests. A shopper who has reached the checkout is not looking for persuasion, only for a reason not to leave, and a short review gives them one. We covered how to set that up in our Shopify checkout guide.

UGC and video, including shoppable video widgets, are the opposite story. They win about as often as reviews, 40% of the time, but lose in almost one test out of three, the highest loss rate of any type with enough volume. Video takes up space and adds weight to the page, and the attention it asks for often costs more than the proof is worth.

Press was the surprise. 44% of 25 press tests won and only 2 lost. Several of the winners were homepage sections, from a Shark Tank feature to an "as seen on" strip, where a first-time visitor has no other reason yet to trust the brand.

Smaller groups, with fewer than 10 tests each, are too small to call, but two of them are worth mentioning: review summaries lost both times we tested them, and activity popups won 2 of 6.

Where on the site it works#

Nearly half of our social proof tests ran on the product page. Of those 83, 34 won and 14 lost, about two and a half wins for every loss. The product page is where most buying decisions are made, and a review answers the question the shopper has there: is this product right for me?

The homepage stands out. Across all our experiments, it is one of the weakest places to test, with a 31.8% win rate. With social proof, it won 41.2% of 34 tests and lost only 5. A first-time visitor on the homepage has no reason yet to trust the brand, and proof from someone else fills that gap.

Landing pages are the exception. Only 3 of 15 social proof tests won there, and 4 lost. Our reading is that most landing page visitors have not picked a product yet, so proof about a product arrives before they care about it.

When it does not work, the format is why#

Most of the losses share a pattern, and it has more to do with where and how the proof showed up than with the type of proof.

On one product page, we added a "See review summary" button next to the review count, near the top of the page. Conversion rate fell 13.4% and revenue per visitor 21.5%. Shoppers clicked through, read the summary and lost the momentum they had built toward buying. Both review summary tests we have run lost.

A pet health brand we manage is the number one best seller in its Amazon category. We put that badge, its five-star review count and three review platform logos in a strip right below the hero of a landing page built for cold Facebook traffic. Conversion fell 16.7%. On mobile, where 92% of visitors were, the strip pushed the section that hooked people below the fold, and a row of logos shown before any interest reads like the brand insisting. The same brand raised conversion 13.56% by putting trust signals inside the buy box, at the moment of deciding.

As we usually say, the fact that these didn't work for these stores doesn't mean you shouldn't try them. They are what we observed with specific clients and scenarios, and they don't have to apply to every store. That's why we test.

Same proof, better format#

One supplements brand gave us the cleanest comparison we have. On mobile, a "someone just bought this" popup covered part of the screen, and removing it raised conversion 1.76% at 95% confidence (we told that story in our win rate benchmark). Months later we brought the same purchase activity back, this time as an inline stream inside the product page, and conversion rose 4.99% at 91%. The information was the same both times. The popup interrupted the shopper, while the inline version sat where they were already reading.

Taking social proof out is worth testing too. Of the four tests where we removed it, three won and none lost: the popup above, a review box on a collection page, and the reviews in the checkout of another supplements brand. That last one is the same placement that won 7 of 16 tests elsewhere, so a placement that works for most stores can still be the wrong one for yours.

Our best social proof tests#

Most of these sit where the shopper is already deciding, on the product page or in the checkout, and most of them made the proof more concrete. Stars alone became an exact rating with a review count. A 96% stat became reviews from doctors. On a footwear product page, the brand's own description of the shoe was swapped for a single customer review, and mobile conversion rose 5.95%.

One result goes against the UGC numbers above. Three customer videos on the product page of a water purifier brand lifted conversion 15.19% and revenue per visitor 19.3%, consistent across devices. The product is expensive and hard to judge from a photo, and post-purchase surveys had asked for proof that it works. The videos showed it working, which a photo could not do.

Practical takeaways#

• If you can't afford a loss, start with social proof. It lost in 19.4% of our tests, against 23.6% across everything we test.
• Start with reviews on the product page and a short review in the checkout. They are the largest and most reliable groups in our data.
• Treat UGC and video as the riskier bet, unless the product needs to be seen working. They lost in almost one test out of three, so read mobile and desktop separately when you test them.
• Check the widgets you already have before adding new ones. Three of the four tests where we removed social proof won.
• Put the proof where the decision happens. A "#1 Best Seller" strip below the hero lost 16.7% on a landing page, while the same brand won 13.56% with trust signals inside the buy box.
• Run these tests with the reviews app you already have. Apps like Okendo cover reviews, UGC and on-site widgets, so most of the tests in this post change where and how the proof shows up rather than requiring new features.

The limits of this data#

• Every verdict uses a 90% Bayesian probability threshold, so some wins and losses are close calls.
• Verdicts were recorded by hand over five years and several testing tools.
• A win can come from the overall result, one device or a metric set for that test, so some winners moved on mobile and stayed flat on desktop.
• Lift sizes were not recorded the same way across tests, so the percentages in this post are examples, not averages.
• Any group with fewer than 10 tests is directional, and we don't report it as a percentage.

Frequently Asked Questions#

Does social proof increase conversion rate?#

More often than it hurts it. Across 170 social proof A/B tests DTC Pages ran on Shopify stores between 2021 and 2026, 37.1% won at a 90% threshold, 19.4% lost and 43.5% were inconclusive. That is almost two wins for every loss, against 1.61 across all the tests we run.

Where should reviews go on a product page?#

Close to the buy box, and visible without a click. Across 83 social proof tests on product pages, 34 won and 14 lost. Moving reviews above the fold raised conversion 6.97% at 98% confidence, while adding a "See review summary" button next to the review count cut it 13.4%.

Should you show reviews in the checkout?#

It is one of the safest tests you can run. Reviews inside the checkout won 7 of 16 tests we ran and lost only 1. It does not work for every store, though: at one supplements brand, removing reviews from the checkout won.

Does UGC convert better than written reviews?#

Not in our data. UGC and video tests won 40% of the time, about the same as written reviews at 39.3%, but they lost 32% of the time against 19.1% for reviews, across 25 and 89 tests. Video did best where the product was hard to judge from a photo: three customer videos on a water purifier product page lifted conversion 15.19%.

Can social proof hurt conversion?#

Yes. 33 of our 170 social proof tests lost (19.4%). Most of them added a step, like a review summary button, added weight, like a video carousel, or showed up before the visitor cared, like a logo strip right below the hero.

Are "someone just bought" popups worth it?#

It depends on the format. At one supplements brand, removing a "someone just bought" popup on mobile raised conversion 1.76%, and showing the same purchase activity inline on the product page later raised it 4.99%. Across 6 tests that added activity popups, 2 won and 1 lost, too few to call it a benchmark.

What types of social proof should you test first?#

Reviews on the product page and in the checkout. Across 89 review tests, they lost only 19.1% of the time, with 2.06 wins for every loss. On the homepage, press mentions are a strong option: 44% of 25 press tests won and only 2 lost.

Can you hide or filter out negative reviews?#

No, and we do not test it. Since October 2024, the FTC's Consumer Review Rule bars US businesses from presenting the reviews they show as all or most of their reviews when negative ones were left out because of their rating. What we test is where and how honest reviews appear, and that is where the gains in our data come from.

How much traffic do you need to test social proof?#

Enough to reach a clear verdict. We apply a floor of 10,000 visitors per variant and run every test for at least 14 days, so it covers two full weekly cycles. That floor is a minimum, not a stopping rule. Our statistical significance calculator shows how long a test needs at your traffic before you start.

Next Steps#

If you want to know which of these tests fits your store, that is one of the first things we look at in an audit: what proof you already show, where it sits and what it costs the page. If you want a baseline first, compare your store against our ecommerce conversion rate benchmarks. You can see how we run experiments on our A/B testing service page, or book a strategy call and we will go through your product page together.