A/B Testing Guide: 15 Experiments Every Website Should Run

The testing hierarchy is the most important concept in CRO, and it's almost universally ignored. Teams spend months A/B testing button colours — a small-rock test with a ceiling impact of maybe 1–2% conversion lift — while the headline (a big-rock test that can move conversion rate by 20–40%) remains untouched for years. Prioritise tests by potential impact, not by ease of implementation. Change the message before you change the shade of the button.

15 High-Impact A/B Tests

1. Hero headline. The single highest-impact test on most websites. Test three fundamentally different value proposition framings: outcome-focused ("Get 3× More Qualified Leads"), problem-focused ("Stop Losing Leads to Competitors Who Move Faster"), and process-focused ("The Agency That Builds Revenue Systems, Not Just Campaigns"). These represent genuinely different claims, not variations in phrasing — and they will produce significantly different results with different audience segments.
2. CTA button copy. "Get Started" is the most underperforming CTA in existence — it's generic, commitment-heavy, and communicates nothing about what happens next. Test specificity: "Book Your Free Audit," "Start My Free Trial," "See My Growth Roadmap." The more the CTA describes what the user gets (not what they do), the better it typically performs.
3. Form length. Research consistently shows that every additional form field reduces conversion rate by approximately 11% on average. If you're asking for company size, phone number, and "how did you hear about us" alongside name and email, you're paying a heavy conversion tax. Test removing all non-essential fields. If you need qualification data, collect it after conversion — in the confirmation email or the first call.
4. Social proof placement. Most sites put testimonials at the bottom of the page — below the fold, below the CTA, where the already-decided convert and the doubters never scroll. Test moving your strongest social proof above the fold, adjacent to the primary CTA. For B2B, a recognizable client logo strip immediately below the hero headline can produce 15–30% conversion lifts.
5. Pricing page presentation. Test the annual/monthly toggle default (annual default often increases plan value), the "most popular" or "recommended" badge placement (middle tier usually works best), and price anchoring (showing the highest tier first to make the middle tier feel reasonable). The order and framing of pricing options significantly impacts which plan visitors choose.
6. Hero image or video. Stock photography of diverse teams in glass offices performs worse than product screenshots, demo videos, or dashboard previews in virtually every B2B test ever run. Authentic, specific visuals of your actual product outperform aspirational lifestyle imagery. Test a 30-second explainer video against your static hero — video heroes can increase time-on-page significantly and, when well-produced, improve conversion on high-intent traffic.
7. Trust badges and their order. Money-back guarantees, security logos (SSL, SOC 2, GDPR), client logos, and industry certifications all reduce perceived risk — but their impact varies by audience and stage. Test badge presence versus absence, and test the order (money-back guarantee first typically outperforms leading with security certifications for consumer audiences).
8. Above-fold CTA configuration. Test a single primary CTA against a primary CTA plus a secondary ghost button ("See How It Works"). More CTAs can increase overall conversion by accommodating visitors who aren't yet ready for the primary action — or they can create decision paralysis and reduce conversion on all actions. The result depends heavily on your audience's average purchase consideration period.
9. Benefit-focused versus feature-focused copy. "Automated A/B testing across 12 channels" is a feature. "Stop guessing which ads work — know in 48 hours" is a benefit. Test whether your audience responds more to technical capability descriptions or outcome-focused language. Early-stage buyers often respond to benefits; technical buyers who are mid-evaluation often respond to features. Your highest-converting copy talks about what they get, not what you do.
10. Testimonial format. Text testimonials with name and company outperform anonymous quotes. Photo + text outperforms text alone. Video testimonials outperform photo + text for consideration-stage decisions. Statistical case study results ("Reduced CAC by 43% in 90 days") outperform narrative testimonials for data-driven buyers. Test formats against each other — don't assume video always wins.
11. Navigation simplification. Navigation is a competitor to your CTA — every nav link gives visitors an exit route from your conversion path. Test reducing your navigation to 3 or fewer items, or removing navigation entirely from dedicated landing pages. On focused landing pages (paid ad destinations), removing navigation consistently improves conversion rate by directing visitor attention to the single desired action.
12. Mobile-specific layout optimisation. Mobile and desktop visitors behave differently, have different cognitive contexts, and often respond to different layouts. Don't assume your desktop winner transfers to mobile — run mobile and desktop as separate tests. On mobile, thumb-zone design (bottom-anchored CTAs), click-to-call buttons, and condensed above-fold content typically outperform direct ports of desktop layouts.
13. Page load speed. Even a 0.5-second improvement in load time produces measurable conversion improvement. Use PageSpeed Insights to identify your top speed issues: unoptimised images (use WebP and lazy-loading), render-blocking JavaScript (defer non-critical scripts), and missing CDN configuration. Speed improvements compound — they also improve SEO, ad Quality Score, and user experience across the entire site.
14. Exit intent offer. Test an exit intent popup against no popup for visitors who show exit signals (cursor moving toward browser chrome on desktop). Test the offer type: discount versus high-value content download versus free consultation. The content offer often performs as well as the discount without eroding margin. Frequency capping (show once per visitor, never again if dismissed) prevents the popup from degrading experience for returning visitors.
15. Chat widget presence and trigger timing. Live chat and chatbot widgets can either help or hurt conversion depending on how they're deployed. Test widget presence versus absence. Test proactive triggers (message appears after 30 seconds on pricing page) versus passive availability. Over-aggressive chat triggers can feel intrusive and increase exit rates; well-timed triggers on high-intent pages can capture prospects who would otherwise leave without converting.

How to Run Statistically Valid Tests

The most common A/B testing mistake isn't running the wrong tests — it's calling tests too early. A test that shows a 15% lift after 200 visitors may show no lift at all after 2,000 visitors. Statistical significance at 95% confidence with 80% statistical power requires a minimum sample size that most teams underestimate. Use a sample size calculator (Optimizely has a good free one) before launching any test — enter your current conversion rate, the minimum detectable effect you care about, and the calculator returns the number of visitors per variant you need before you can trust the result.

Run every test for a minimum of two full business cycle weeks, regardless of traffic volume. Day-of-week effects are real — your Monday visitors behave differently from your Friday visitors. A test run only Monday to Wednesday will produce misleading results that don't hold when applied to full-week traffic. For low-traffic sites, this means accepting longer test windows — 4–6 weeks is not unusual for sites with under 5,000 monthly visitors on the tested page.

Test one variable per experiment. Multivariate testing requires dramatically higher traffic to reach significance and makes it impossible to know which change drove the result. The only exception is radical redesign tests (A vs B where B is a completely different page) — these are valid when you want to test a hypothesis about overall page architecture, with follow-up single-variable tests to isolate the winning elements.

Prioritising Your Test Roadmap

Use the PIE framework to score and prioritise tests: Potential impact (how much could this test move conversion rate?), Importance (how much traffic does this page receive? how business-critical is it?), and Ease (how difficult is this test to implement?). Score each dimension 1–10 and average the scores. Tests with high PIE scores get prioritised in the roadmap — not tests that are easy or interesting to your team.

Focus your testing budget on pages with existing conversion intent: product pages, pricing pages, and dedicated landing pages will almost always produce higher-value test results than blog posts or about pages. A 2% lift on a page that converts 5% of 10,000 monthly visitors is worth 100 additional conversions per month. The same lift on a page that converts 0.5% of 500 monthly visitors is worth 1 additional conversion per month. Test where the math matters.