SaaS A/B tests we've caught

Control shows a 3-tier plan lineup (Acrobat Standard, Pro, Studio for teams) while the variant adds a fourth, cheaper 'Acrobat Express for teams' plan ($11.99→$11.08/mo) ahead of Standard, changing the plan count, order, and comparison table columns.

The interactive 'Periodic Table of AI Pricing' widget defaults to a different category tab and icon set across visits: some visits (1,2,3,6,7) show 'Project Management' selected with the Design/Productivity icon grid, while others (4,5) show 'Support Platform' selected with a different highlighted icon set and no logo captions below the grid.

Two distinct funnel variants: control uses a centered hero with three feature cards (Research agent, AI automations, Call intelligence), a 3-tier pricing table, and a testimonial grid; variant uses a left-aligned hero with a single visual panel, an expandable feature-list section ('Automate to your advantage'), a 4-tier pricing table (adds Enterprise/Custom plan), different section headlines ('From zero to IPO' vs 'Start today for free'), and a Twitter-style testimonial layout.

Control shows a single-hero page with left-aligned copy, a two-column feature explainer, a four-tier pricing table (Free/Plus/Pro/Enterprise), and a testimonial grid before FAQs. Variant shows a centered hero immediately followed by a social-proof tweet cluster, a three-tier pricing table (Free/Plus/Pro, no Enterprise), a three-card 'AI intelligence' feature layout, and a shorter FAQ set with different questions.

The two funnels reorder page sections (logo bar vs pricing table swapped in position) and show a different hero product screenshot/demo widget (a search-style 'Find experienced marketing agencies' UI vs a workflow-style 'Automatically qualify leads' AI card).

The two variants present the same content blocks (AI feature cards, logos, pricing, testimonials, FAQ) in a different order and with different section headlines: variant A shows feature cards directly under the hero and labels the testimonial section 'Loved by thousands of companies' with card-style quotes and 'Frequently asked questions', while variant B leads with a tweet-style testimonial grid right under the hero, moves the feature cards below pricing under the heading 'AI intelligence at your fingertips', and retitles the FAQ section 'Your questions, answered'.

Visit 1 (Safari) shows a different hero headline — 'When your product knows what users need, you don't have to build product tours' — while visits 2 (Chrome) and 3 (Firefox) show 'Launch in-app nudges with in-house quality and no-code speed!' with a different subheadline. Since Safari is the only outlier and Chrome/Firefox agree, this is consistent with the Safari visit receiving a different Statsig experiment bucket (each visit has a distinct stable_id), not a browser-rendering difference.

Visit 3 (Firefox) shows meaningfully different page content: different bullet point copy ('Getting the most ROI out of Box', 'Use cases for your specific industry', 'The right pricing plan for your business', 'A demo so you can see Box in action' vs. the Unlimited storage/Seamless collaboration/Financial Services/Enterprise-grade security copy in visits 1 and 2), different customer logo set (Broadcom, Morgan Stanley, U.S. Air Force, Airbnb, JLL vs. AmTrust, Fidelity, Principal, Regions, FundWell, Pan American Life), and an additional form field ('How did you hear about us?') not present in visits 1 or 2. However, all three visits use different browser engines (Safari, Chrome, Firefox) and the mbox cookie diverges, making it ambiguous whether this is an A/B test or cross-browser rendering with different Adobe Target session buckets.

The badge/pill near the top of the page differs: visits 1 (Safari) and 2 (Chrome) show '4.4 STARS ON G2', while visit 3 (Firefox) shows '25% AVERAGE LIFT IN USER RETENTION'. Since only the Firefox visit diverges and the two Blink-engine visits (Safari/Chrome) agree, this is most likely a rotating/carousel stat element hitting a different position during capture rather than a true A/B test.