AGConsult
Conversionista
MarketingExperimentsA pricing table wins in our A/B test database when it makes the choice easier, not when it looks better: fewer, clearly packaged plans, one plan visibly recommended, and features laid out so people can compare them.
Server Density's switch from per-server pricing to named plans roughly doubled revenue per visitor, and RuneScape's "Recommended" badge added about 6% to purchases.
The losses came from hiding price details, removing choices people wanted, and showing prices so low they looked suspicious.
This hub covers pricing tables, plan comparison layouts, billing options and free-trial offers on SaaS and lead-gen sites.
Every matching test appears in the grid on this page.
What follows is the analysis: what won, what lost, and how to test your own.
For discount and sale-price tests on ecommerce stores, see pricing and discount tests.
Plan comparison tables: what the tests say about layout
The most consistent pattern is simple: let people see all the plans, and what each one includes, at the same time.
Paltalk's original subscription page didn't fit every plan above the fold, and features were hard to compare.
Wingify Services rebuilt it with all plans above the fold in a feature comparison table, plus testimonials.
Over 53 days on all traffic, purchases rose 6.74%.
Lyyti replaced short plan descriptions with a table of each plan's features, added free trial buttons above and below every plan, and demoted "Request a Quote" to a text link.
Over five months, visits to the trial page rose 93.71% at 96% significance.
Note the metric: visits to the sign-up page, not completed trials.
MarketingSherpa added a members vs non-members comparison matrix to its free trial page, with testimonials beside the chart.
That bundle produced +76% free trial subscriptions. If your product differs from free alternatives in specific ways, a "with us vs without us" matrix is a strong first test.
See comparison chart tests for more on this format.
Keeping prices visible also helped. PureGym made prices sticky at the top while people scrolled its membership comparison, for +2% conversion.
Hivelocity, working with Cro Metrics, moved pricing and filters above the fold on its dedicated server page and reported +187% checkouts, though the case study gives no sample size for what it calls a low-traffic site.
The clearest loss in this group is about hiding detail.
Conversionista redesigned Fortum's electricity price blocks and moved the deeper price details into a modal.
Click-through fell 9%, and only 18% of variant users who opened the details went on, against 75% on the original.
The modal version lost. The next iteration kept the drop-down, clarified the copy and color, and won by 9%.
Same goal, opposite results, and the difference was whether people could check the details without leaving the table.
SaaS pricing table design: packaging, highlighting and billing
The biggest numbers in this hub come from how plans are packaged, not how they're styled.
Packaged plans beat build-your-own pricing
Server Density replaced per-server pricing (about $11 per server plus $2 per website) with named plans starting at $99 a month.
In the write-up on Patrick McKenzie's blog, trial sign-ups fell from 150 to 113, but revenue added rose from $420 to $876, from about $0.19 to $0.41 per visitor.
VWO reported confidence of 93% on revenue, which is below the usual 95% bar, and the number of paid conversions was small.
A separate VWO case study of Server Density's packaged pricing tiers reports +114% revenue ($394 to $833) on 20 vs 15 paid conversions, with free sign-ups down 24.96%.
Both reports point the same way: fewer sign-ups, more revenue.
That is exactly why a pricing table test needs a revenue metric, not a sign-up metric.
Highlight one plan
Marking one plan as the default choice has the cleanest evidence in this hub.
Jagex first added a section showing four key packages on RuneScape's Treasure Hunter page for about +10% purchases.
In a follow-up, marking one of the four as "Recommended" added about 6% more purchases.
Bundled redesigns point the same way. CodaLift LLC gave Planoly's Growth plan a contrasting color and distinct button, rewrote the heading around benefits, and showed annual savings per year instead of per month.
Revenue rose 38% and average order value rose 24%.
Interestingly, the biggest jump in clicks (+74%) went to the top Professional plan, not the highlighted middle one.
VWO Services' ArchiveSocial redesign marked the mid-priced plan "most popular" and simplified the options, for +64.76% leads over seven weeks.
Why does a highlighted plan work? The usual explanations are anchoring and the decoy effect, but the tests here can't tell you which, if either, is at work.
A plain reading is that a recommendation saves people from deciding on their own.
Make limits concrete
Powtoon tested storage labels across three variants. Showing the Business plan with 10 GB, against 2 GB on Pro, beat showing it as "unlimited" and produced +27.9% revenue.
A specific number gave people something to compare.
Powtoon also tested a new monthly vs annual price display that lost, though the case study gives no details.
Billing periods and longer terms
Discounts for committing longer can pay off. Meebox offered 20-40% off for a two-year term and saw +121.56% revenue, +46.24% order value and +51.85% conversions.
Whitepages put three plans side by side with a monthly/annual toggle at the top and got +23% pricing page conversion and +31% monthly memberships in a one-week test with several changes bundled.
Taking away the term choice backfired. MarketingExperiments removed a subscription-length dropdown so visitors were locked into monthly billing, and conversion fell about 40%.
Fewer options is not automatically better; the paradox of choice doesn't mean people want no choice at all.
Pricing page A/B tests: price points, trials and what to measure
Several tests here changed the offer itself, not the page.
They tend to produce the largest effects, and the hardest to interpret.
Price points. In a four-day MarketingExperiments test on PPC traffic, $10 a month brought 156 orders and $1,560 against 94 orders and $1,175 at $12.50 and 74 orders and $1,106 at $14.95, so the lowest price earned 33% more revenue than the next best.
The Boston Globe test pointed the other way: after a $0.99 intro rate, a $3.99 regular weekly rate produced the most total revenue, while $0.99 won on subscriptions.
The winner depends on the metric you pick. See price testing before you run one.
Showing a price at all. SafeSoft Solutions put the product price next to a lead form and reported +100% leads, with no sample size given. iProspect added a "starting at" price with a monthly cost to enquiry-led product pages for +15% leads at 95.26% significance; the high-price variant was switched off early.
PriceCharting found the opposite edge: a lower "Starts at $2" or "$1" label lost to $4, at only 91% significance.
Free trials vs free plans. Acuity Scheduling sent sign-ups into a 14-day trial of its $10/month Professional plan, card required, instead of a free-forever plan, and got +268.14% paid sign-ups.
Going changed "Sign up for free" to "Trial for free," moving people into a premium trial, for +104% trial starts month over month, without a controlled split.
Patrick McKenzie cut Bingo Card Creator's free trial from 15 cards to 8 and paid conversion went from 1.04% to 1.89% at 99% confidence. More on this in free trial conversion.
Flat results happen. MarketingExperiments rebuilt a financial information subscription page with new copy, layout and a shorter order path, and conversion moved from 0.24% to 0.27%, not significant, across roughly 175,000 visitors. A full redesign is no guarantee.
How to run your own pricing table test
- Pick revenue per visitor as the primary metric. Server Density lost sign-ups and gained revenue. A sign-up metric would have killed the winner.
- Change one thing when you can. Most of the large wins here (Lyyti, ArchiveSocial, Paltalk, Whitepages) are bundled redesigns. They tell you a better page won, not which part did.
- Size it before you launch. Pricing pages get less traffic than homepages, and paid conversions are rarer still. Use the sample size calculator; many tests here would not pass a 95% bar.
- Watch downstream behavior. Price and trial changes affect cancellations and lifetime value, which a two-week test can't see. The Boston Globe test is a good reminder.
Plan names: what to call your tiers
None of the tests in our database isolate plan names, so treat this section as guidance, not proven results. Name tests are cheap but rarely published.
The closest evidence is Server Density again. Patrick McKenzie argued that plans named for the buyer remove the need to count usage: "name the plans correctly and they won't even need to count servers", because a company that sees itself as an enterprise picks the Enterprise plan.
The test changed names, packaging and price together, so the names alone can't take credit.
AGConsult's work with Yoast is related. Free vs Premium differences were unclear, so they built pages for different audiences with benefit-led copy explaining when each Premium feature helps.
After a year of testing, revenue was up 118%, a program-level result rather than a single split test.
Practical rules that follow from the tests above:
- Name plans for who they're for (Starter, Team, Enterprise) or what they unlock, so people can pick without reading every row.
- Keep labels on the table plain. OnSwitch's $10 bundle with a "Sale" headline lost revenue; relabeled "Bundle Sale" with a screen explaining the contents, it turned into a gain. Neither report gives figures.
- Pair a name test with a "Recommended" or "Most popular" label test, since that has the stronger evidence.
For the full page around the table, see SaaS pricing page best practices and our SaaS A/B test results. Browse every element or search all tests.