Of 177 measured SaaS tests, 157 won, 15 lost and 5 were flat, and the most useful pattern is simple: the changes that won most often removed a step, a field or a distraction between the visitor and the trial, demo or sign-up.
The results come from SaaS companies, agencies and testing tools, including VWO case studies, MarketingExperiments research and a long run of KlientBoost landing page programmes.
Treat the win rate with care: published case studies lean towards winners, so the 20 losing and flat tests are worth reading as closely as the big lifts.
This write-up walks through where SaaS teams test, the largest measured wins, what the losses teach, and which tests to run first.
Which page types have the most SaaS tests
Landing pages lead with 47 tests, followed by the homepage (27), pricing page (22), other pages (22), signup page (14), free trial (10), checkout (9), onboarding (7), demo request (6), email (6), site-wide (4) and ads (3).
- Landing pages have one job: turn paid or search traffic into a trial, demo or lead with as little friction as possible.
- Homepages must say what the product does and route each visitor to one clear next step.
- Pricing pages have to make one plan the obvious choice and remove doubt about cost.
- Signup pages and free trial pages need to feel short and safe.
- Onboarding has to get a new user to value before the trial ends.
- Demo request, checkout and email tests are fewer, but several show large swings.
Biggest measured wins
Pricing changes produced some of the largest revenue numbers.
Server Density moved from per-server pricing to packaged tiers and saw +114% revenue ($394 → $833), even though free sign-ups fell 24.96%.
Meebox offered 20-40% off a 2-year term and recorded +121.56% revenue.
Trial structure matters as much as page design.
Acuity Scheduling swapped a free-forever plan for a 14-day Pro trial and got +268.14% paid signups.
GetResponse added a Free Trial button next to Buy Now for +158.6% free accounts with no drop in paid accounts.
Onboarding tests show that activation can move a lot.
AURUM's floating onboarding checklist gave +187% activation rate, and a VWO case study of its own 'Get Started' checklist reported +127% free trial to paid conversions.
Cutting steps paid off at the top of the funnel too.
A MarketingExperiments test for Britannica removed one registration page for +117% free trial conversion and +130% revenue, and VMware moved try and buy CTAs above the fold for +956% clickthrough.
Tests that lost or were flat
AssessmentDay is the clearest warning against stacking wins.
Removing the FAQ section won +62% sales and removing screenshots won +56% sales, but removing both together lost about 3% of sales. Two winners do not always add up.
Some friction is useful. A MarketingExperiments test that removed the subscription-length choice cut conversion by about 40%, and a three-step free trial sign-up on the RegOnline homepage lost 24.5% of completed leads.
Flat results teach caution about numbers that look too good.
A product page value proposition test showed +445.6%, but a rerun with a cloned control found a 30% gap between identical pages, so the lift was not trusted.
A condensed freemium product page showed +31.8% Try Free clicks but stayed inconclusive.
Britannica also found that red or orange article links instead of blue cut engagement by about 10%. Familiar conventions are often worth keeping.
Patterns that repeat
Where to start
Start where the visitor decides to commit. Add customer logos, testimonials and a no-credit-card note to your signup or trial page, then test a cleaner, single-CTA landing page for paid traffic.
Next, test pricing packaging and a highlighted plan: CodaLift LLC's Planoly pricing page with a highlighted middle plan produced +38% revenue.
After that, add an onboarding checklist and measure activation and trial-to-paid, not just sign-ups.
Watch the downstream metric. Hubstaff's homepage redesign lifted visitor-to-trial by 49% while trial-to-paid fell 7%.
Tools to run these tests
You need an A/B testing tool that can split traffic on your marketing site, a way to change in-app onboarding, and analytics that track the trial or demo through to paid.
Several of the tests here come from VWO case studies.
For low-traffic SaaS sites, test bigger changes on the pages with the most traffic first, and rerun surprising results before you trust them.
Questions
What is the most effective A/B test for a SaaS website?
There is no single winner, but in these results removing friction before the trial or demo won most often: fewer fields, fewer steps, one clear CTA and proof next to the form.
Pricing packaging and onboarding checklists produced some of the largest revenue and activation lifts.
Do SaaS A/B tests usually win?
Of 177 measured SaaS tests here, 157 won, 15 lost and 5 were flat.
Published case studies lean towards winners, so expect a lower win rate on your own site and read the losing tests closely.
Should SaaS teams measure sign-ups or paid conversions?
Measure through to paid where you can. Hubstaff's homepage redesign lifted visitor-to-trial by 49% while trial-to-paid fell 7%, and Server Density lost free sign-ups but more than doubled revenue.
KlientBoost
MarketingExperiments
Mida
AGConsult
Conversionista
Apexure