The best pricing page A/B tests in our archive changed how prices were packaged and framed, not just how the page looked.
Meebox got +121.56% revenue by offering 20-40% off a 2-year term, and Server Density got +114% revenue by swapping per-server pricing for packaged tiers.
On app paywalls, adding a 7-day trial to Stardust's yearly plan lifted paywall conversion by 131%.
How we ranked them: we pulled every SaaS, app and lead-gen pricing test from the LiftArchive database, kept the winners with a published number, and ranked clean single-change A/B tests on revenue or revenue per user above click and conversion metrics.
We did not run these tests; every number comes from the company or agency that did, linked on each test page.
| Rank | Test | Change | Result | Who ran it |
|---|---|---|---|---|
| 1 | Meebox 2-year discount | 20-40% off for a 2-year term | +121.56% revenue | Meebox |
| 2 | Server Density packages | Packaged tiers from $99/month | +114% revenue | Server Density |
| 3 | Stardust trial | 7-day trial on yearly plan | +131% paywall conversion | Stardust |
| 4 | Preu AI annual plan | Annual plan added next to monthly | +80% proceeds per user | Preu AI |
| 5 | Stardust badge | Discount badge, weekly-equivalent price | +62.5% proceeds per user | Stardust |
| 6 | PropGPT price | Weekly $6.99 → $9.99 | +47.5% proceeds per user | PropGPT |
| 7 | Planner 5D tiers | Two higher price tiers | +83% ARPU (highest tier) | Planner 5D |
| 8 | Mojo monthly anchor | Annual shown as per-month price | +45% Brazil, +26% Mexico, +8% USA | Mojo |
| 9 | Three price points | $10 vs $12.50 vs $14.95 | +33% revenue at $10 | MarketingExperiments |
| 10 | Powtoon storage | '10 GB' instead of 'unlimited' | +27.9% revenue | VWO |
| 11 | BuyBye price rise | Yearly price 50% higher | -19% conversions; +25% proceeds per user | BuyBye |
| 12 | Rash ID plans | Three plans, best-offer badge | About +20% ARPU | Rash ID |
| 13 | Fits paywall | Toggle vs timeline layout | +19% ARPU | Fits |
| 14 | Hivelocity layout | Prices above the fold | +187% checkouts | Cro Metrics |
| 15 | Veeam link copy | 'Request pricing' link text | +161.66% CTR | Veeam Software |
| 16 | Avid bonus banner | Bonus gift in top banner | +34% transactions | Avid |
| 17 | Cascading offers | Offers that change by view count | +8.3% ARPU | Superwall |
| 18 | Fortum price blocks | Clearer blocks, drop-down kept | +9% click-through | Conversionista |
| 19 | RuneScape badge | 'Recommended' badge | About +6% purchases | Jagex |
| 20 | PureGym sticky prices | Sticky prices on table | +2% conversion | PureGym |
Want design patterns rather than numbers? See our SaaS pricing page examples.
For app subscription screens specifically, there's a separate list of the best paywall tests.
The 20 best pricing page A/B tests
1. Meebox: 20-40% off for a 2-year term
The change: The control pricing page listed monthly cost, space, bandwidth and domains with no discounts.
The variant offered 20-40% discounts to customers who committed to a 2-year term.
The result: +121.56% revenue, with +46.24% AOV and +51.85% conversions.
Meebox ran it as an A/B test in VWO and reported a 98% chance to beat the original.
Why it likely worked: A longer term at a lower effective price gave buyers a reason to commit now, and each committed customer paid for more months up front.
Try it if: you sell hosting, software or any plan where a longer commitment costs you little to serve.
2. Server Density: packaged tiers instead of per-server pricing
The change: Configurable per-server and per-website pricing was replaced with packaged pricing tiers starting at US$99/month.
The result: +114% revenue ($394 → $833), while free sign-ups fell 24.96% (6.92% → 5.19%).
Server Density ran the A/B test in VWO on 1,950 vs 1,925 visitors, with 20 vs 15 paid conversions.
Why it likely worked: Named packages remove the math from the buying decision and set a higher floor price.
Fewer people signed up, but the ones who did paid more.
Try it if: your pricing makes buyers calculate their own bill.
Note the small number of paid conversions before you treat +114% as a forecast.
3. Stardust: 7-day free trial on the yearly plan
The change: Stardust, a cycle and pregnancy tracking app, compared $24.99/year with no trial against the same price with a 7-day free trial (plus five other price variants).
The result: +131% paywall conversion; trials overall brought +78% proceeds per user and +45% paying subscribers. Stardust ran it as a multivariate test.
Why it likely worked: A free trial lowers the risk of a yearly commitment.
Higher-price variants without a trial converted worse with similar proceeds, so the trial did more than the price.
Try it if: your app or SaaS asks for a yearly payment from people who haven't used the product yet.
4. Preu AI: annual plan priced against monthly
The change: The control paywall showed a monthly plan only with a 'Try Free' button.
The variant added an annual plan priced at 60% of the monthly price.
The result: +80% proceeds per user and +22% trial conversion.
Preu AI, a university entrance exam prep app in Chile, ran it as an A/B test.
Why it likely worked: A second plan gives people something to compare against.
The monthly price makes the annual price look like a deal, and some people pick it.
Try it if: you only show one plan today.
5. Stardust: discount badge and weekly-equivalent price
The change: The control showed yearly $24.99 and weekly $2.99 plainly.
The variant added a discount percentage on the yearly plan plus weekly-equivalent pricing.
The result: +50% conversion, +62.5% proceeds per user and +30% yearly share. Stardust ran it as an A/B test.
Why it likely worked: Breaking the yearly price into a weekly figure lets buyers compare plans in the same unit, and the badge says the yearly plan is the better deal.
Try it if: you sell more than one billing period and the long one is hard to compare at a glance.
6. PropGPT: weekly price from $6.99 to $9.99
The change: The weekly price went from $6.99 to $9.99, with the same images and copy.
The result: +47.5% proceeds per user. PropGPT, a sports betting insights app, ran it as an A/B test and described the conversion drop as minimal, without a number.
Why it likely worked: Buyers in this niche cared more about the product than the price, so a 43% higher price mostly turned into revenue.
Try it if: you've never tested a higher price.
Measure revenue per visitor, not conversion rate alone.
7. Planner 5D: two higher price tiers
The change: Planner 5D, an interior design app, tested its control price against two higher price tiers on the same paywall design (A/B/C).
The result: +83% ARPU for the highest tier.
Planner 5D ran it as a multivariate test; the price levels were not published.
Why it likely worked: Testing more than one higher price shows where demand starts to fall off, instead of guessing one number.
Try it if: you want to find a price ceiling.
Run two or three price points at once, not one.
8. Mojo: annual plan shown as a monthly price
The change: Mojo, a video editing app, changed the annual plan from a yearly price to a per-month price, billed annually.
The result: New revenue per paywall impression: +45% Brazil, +26% Mexico, +8% USA.
Mojo ran it as an A/B test; the lift came from more paywall-to-trial conversions and more annual picks.
Why it likely worked: A per-month figure is smaller and easier to compare.
The effect was biggest in markets where a full yearly price is a bigger share of income.
Try it if: you sell annual plans in several countries. Read the result by market.
9. Subscription site: $10 vs $12.50 vs $14.95 a month
The change: Three identical landing pages priced a subscription at $14.95, $12.50 and $10.00 per month, with PPC traffic split evenly three ways.
The result: $10/mo: 156 orders and $1,560 revenue vs $12.50: 94 orders, $1,175 vs $14.95: 74 orders, $1,106. The lowest price brought +33% revenue over the next best.
MarketingExperiments ran the four-day A/B/C test (324 total orders, no significance figures).
Why it likely worked: Here demand was price-sensitive: the lower price more than doubled orders against the top price, so volume beat margin.
Try it if: you're not sure if your price is too high.
Compare it with test #6 and the losing reference publisher test below: the same move can win or lose.
10. Powtoon: '10 GB' instead of 'unlimited' storage
The change: The control showed the Pro plan at 2 GB storage and the Business plan at unlimited.
Three variants were tested; the winner showed Business at 10 GB vs Pro at 2 GB.
The result: +27.9% revenue. VWO's services team ran the A/B test for Powtoon.
Why it likely worked: 'Unlimited' can sound vague or like more than you need.
A concrete number makes the gap between plans easy to judge.
Try it if: your plan table uses 'unlimited' to separate tiers.
11. BuyBye: yearly price 50% higher
The change: BuyBye, a spending-limit and savings app, raised its yearly price by 50% with a trial and the same design.
The result: -19% conversions but +25% proceeds per user. BuyBye ran it as an A/B test.
Why it likely worked: Fewer people bought, but each buyer paid enough more to cover the loss.
This is the clearest example of why conversion rate alone is the wrong metric for price tests.
Try it if: your conversion rate is healthy and you've never raised prices.
12. Rash ID: three plans with a best-offer badge
The change: The control showed two plans (monthly, yearly) with no best-offer label.
The variant showed three plans (monthly, yearly, weekly) with yearly marked as the best offer.
The result: About +20% ARPU and over +50% conversion to paid.
Rash ID, an AI skin condition app, ran it as a 50/50 A/B test.
Why it likely worked: A third, lower-upfront option gives people an entry point, and the badge steers them to the plan you want them to pick.
Try it if: you show two plans and no recommendation.
13. Fits: toggle paywall vs timeline paywall
The change: Fits tested a Blinkist-style trial timeline paywall against a toggle paywall.
The result: +19% ARPU for the toggle paywall.
Fits ran it as an A/B test. A follow-up found the lift came from visual hierarchy (normalized yearly price, 'Most Popular' badge, bigger annual option), not the toggle itself.
Why it likely worked: How the plans are laid out mattered more than the trial mechanic.
Isolating the toggle in a follow-up is what showed this.
Try it if: you've copied a paywall style from another app. Test its parts one at a time.
14. Hivelocity: prices and filters above the fold
The change: Pricing and filter controls were moved above the fold on the Dedicated Servers page, which had shown neither there before.
The result: +187% checkouts. Cro Metrics ran the A/B test; the case study gives no sample size or significance and calls it a low-traffic site.
Why it likely worked: Buyers comparing servers want price and specs first.
Hiding them below the fold made people work to find the basics.
Try it if: your pricing sits below a long intro or behind a click.
15. Veeam: 'Request pricing' instead of 'Request a quote'
The change: A product page link to the sales inquiry form changed from 'Request a quote' to 'Request pricing'.
The result: 0.54% → 1.40% CTR (+161.66%), with reported 99.9%+ confidence.
Veeam Software ran the A/B test in VWO.
It measured clicks to the inquiry page, not submitted leads.
Why it likely worked: 'Request pricing' says what the visitor will get. 'Request a quote' sounds like the start of a sales process.
Try it if: you sell B2B software with hidden pricing and a sales form.
16. Avid: bonus gift in a top banner
The change: The control showed a bonus gift as a bold bullet under each subscription option.
The variant promoted it in a purple top-of-page banner.
The result: +34% transactions. Avid ran the A/B test in VWO; it's unclear whether there was also a no-offer control.
Why it likely worked: An offer buried in bullet lists is easy to miss.
The banner made it the first thing people saw.
Try it if: you already run a bonus or promo that's hard to spot on the pricing page.
17. Subscription app: cascading paywall offers
SuperwallThe change: Version 1 showed one offer for a user's first two paywall views and a second offer after that.
Version 2 showed the standard paywall first, an extended-trial paywall on views 2-3 and a discount paywall from the fourth view.
The result: +8.3% ARPU and about +$14.7K/month proceeds (about $177K/year).
Superwall ran the A/B test for an anonymized customer; offer prices were not given.
Why it likely worked: Full price goes to people ready to pay.
Only people who keep coming back without buying see the cheaper offers, so less margin is given away.
Try it if: you show a discount to everyone on the first view.
18. Fortum: clearer price blocks, drop-down kept
The change: After a losing version that moved price details into a modal, Conversionista iterated on Fortum's electricity price page: clearer copy and color, with the drop-down for price details kept.
The result: +9% click-through rate. Conversionista ran the A/B test; sample size and confidence were not stated.
Why it likely worked: The first version hid details in a modal and lost.
Keeping details in place while cleaning up the blocks fixed it.
Try it if: a pricing test lost and you can see why in the data. Iterate rather than drop the idea.
19. RuneScape: 'Recommended' badge on one of four packages
The change: Jagex took a page that showed four packages equally and marked one as Recommended.
The result: About +6% purchases. Jagex ran the A/B test in VWO for in-game digital item purchases.
Why it likely worked: Four equal options make people stop and compare. A recommendation gives them a default.
Try it if: you show three or more plans with no visual lead.
20. PureGym: sticky prices on the membership table
The change: On the membership comparison step, prices used to scroll out of view.
The variant kept them sticky at the top while users scrolled.
The result: +2% conversion. PureGym ran the A/B test in AB Tasty.
Why it likely worked: Long comparison tables make people scroll back up to check what each plan costs.
Sticky prices keep the main fact in view.
Try it if: your plan comparison is longer than one screen, especially on mobile.
What the winners have in common
They measured revenue, not just sign-ups
Server Density lost 24.96% of free sign-ups and still doubled revenue.
BuyBye lost 19% of conversions and still made +25% proceeds per user.
If these teams had judged on conversion rate, they would have shipped the wrong version.
Our price testing guide covers how to set this up.
Price cuts and price rises both win, depending on the buyer
A $10 monthly price beat $14.95 for one subscription site, while PropGPT, BuyBye and Planner 5D all made more by charging more.
The reference publisher test shows the risk: cutting $69.95 to $50 brought more subscriptions but about 15% less revenue.
Test your own price; don't borrow someone else's answer. More on this in SaaS pricing strategy.
A second plan and a clear default
Preu AI added an annual plan, Rash ID added a third plan with a badge, and RuneScape marked one package as Recommended.
Each gave buyers a comparison and a suggested pick. See pricing plan tests for more.
Prices in the unit people think in
Mojo's per-month annual price and Stardust's weekly-equivalent price both made the long plan easier to compare.
Powtoon's "10 GB" did the same job for features.
Hiding things usually loses
Hivelocity and PureGym won by keeping prices in view.
Fortum lost 9% when price details moved into a modal, and won when they stayed on the page.
Stardust lost when it showed only the yearly plan, and MECLABS saw about 15% fewer subscriptions when plan options moved into a dropdown.
Pricing tests that lost
These are worth reading before you copy a winner.
Each one is a reasonable idea that hurt the main metric.
Conversionista
MarketingExperimentsMarketingExperimentsWhat to test next on your pricing page
- Add a longer commitment option (annual or 2-year) at a lower effective price, as Meebox and Preu AI did.
- Replace calculators or per-unit pricing with named packages if buyers have to work out their own bill.
- Test one or two higher price points against your current price and judge on revenue per visitor.
- Mark one plan as recommended or most popular, and make it the visually biggest option.
- Show long plans in the same unit as short ones (per month or per week).
- Move prices above the fold and keep them visible on long comparison tables.
- On B2B pages with hidden pricing, test the CTA wording before you test publishing prices.
For the full playbook, see pricing page best practices and subscription pricing for apps.
Browse more ranked lists in our research section.
Watch out for
- Small samples. Server Density's +114% came from 20 vs 15 paid conversions, and the $10 test ran four days. Big numbers on small samples often shrink when rerun.
- Clicks are not sales. Veeam's +161.66% and Fortum's +9% measured click-through, not completed purchases or leads.
- Refunds and churn. A price or trial change can lift first payments and raise churn later. Check retention for at least one billing cycle before calling it.
- Fake discounts. A "save 40%" badge only works long term if the saving is real. Made-up anchors erode trust and can break consumer protection rules.
If you want help planning a pricing test on your own site, a specialist can set up the revenue tracking these tests depend on. See our list of pricing CRO experts.