Why Most "Test Ideas" Lists Are Useless
Search for "A/B test ideas" and you'll find hundreds of lists. "Test your CTA button color." "Try a different hero image." "Add a countdown timer."
These aren't test ideas — they're changes. A list of changes tells you what to modify but nothing about why it should work, what business outcome you're targeting, or what you'll learn regardless of the result.
This is a different kind of list. Each of the 10 tests below comes with: a hypothesis template, which pages it applies to, a realistic lift range based on practitioner benchmarks, the primary metric to use, and what to test next after a win or a loss. Use these as starting points, not final answers.
The ROI Framework for Prioritizing Tests
Before any test makes it onto your roadmap, it should clear a basic ROI screen:
Estimated impact = (Expected lift % × Monthly conversions × Revenue per conversion) ÷ Implementation cost
If your checkout page converts 2,000 times per month at $150 average order value, and you expect a 5% lift from a test, that's:
0.05 × 2,000 × $150 = $15,000/month in incremental revenue
If the implementation takes 2 engineering days, the ROI is positive within weeks. If it takes 3 weeks of engineering work, the ROI requires the lift to hold for months before breakeven.
This calculation doesn't need to be precise — it needs to be good enough to rank ideas against each other. Run the numbers for each candidate test and prioritize the highest expected impact per unit of implementation cost.
**Pro Tip:** Implementation cost includes more than engineering hours. Factor in QA time, design time, the opportunity cost of engineering bandwidth (what else could they be building?), and the ongoing maintenance cost of the variant once it ships as a winner. A "small" test that creates a long-term tech debt burden may have lower net ROI than it appears.
Test 1: CTA Copy and Button Text
Hypothesis template: "Because [X% of visitors who reach the CTA don't click it, and user research indicates the current copy is ambiguous about what happens after clicking], we believe changing the CTA from [current copy] to [benefit-led, outcome-specific copy] will increase CTA click-through rate for [new/returning visitors], because reducing cognitive uncertainty about the post-click experience lowers the decision cost of initiating the action."
Where to test: Primary conversion CTAs on landing pages, pricing pages, homepage hero sections. Any page where the CTA is the main conversion action.
Realistic lift range: 5-25% relative improvement in CTA CTR. Copy changes on high-intent pages tend to produce the largest effects. Lower-intent pages (blog posts, content pages) see smaller effects.
Primary metric: CTA click-through rate. Secondary: downstream conversion (sign-up, purchase) to ensure the CTR lift isn't driven by unqualified clicks.
After a win: Test CTA placement (above fold vs. repeated throughout page) and CTA visual design (color contrast, size, button vs. link).
After a loss: Test the page section above the CTA. If the value proposition hasn't been established, CTA copy optimization has a ceiling. The issue may be earlier in the page.
**Pro Tip:** The most consistently high-performing CTA copy pattern is specific outcome + low friction signal. "Start Your Free Trial" outperforms "Get Started." "See My Personalized Energy Report" outperforms "Submit." The more specific the outcome and the lower the perceived commitment, the better.
Test 2: Social Proof Placement
Hypothesis template: "Because [user session data shows only X% of visitors scroll to our testimonials section below the fold, yet users who see testimonials convert at 2.3x the rate of those who don't], we believe moving testimonials above the fold — directly below the hero section — will increase sign-up rate for new organic visitors, because social proof from identified customers in similar roles reduces purchase risk perception before the user decides whether to invest attention in the rest of the page."
Where to test: Homepages, landing pages, pricing pages. High-value conversion pages where trust is a likely barrier.
Realistic lift range: 3-15% relative improvement in conversion rate. Higher end typically seen on higher-consideration purchases (B2B SaaS, financial services, high-ticket e-commerce).
Primary metric: Primary page conversion (sign-up, trial start, lead form submission).
After a win: Test testimonial format (written vs. video, with vs. without photo, with vs. without company/role attribution). Attribution specificity ("Sarah M., VP Marketing at a 500-person tech company") typically outperforms generic attribution.
After a loss: Investigate whether the testimonials themselves are the issue. If your testimonials are generic ("Great product!") rather than outcome-specific ("Reduced our energy costs by 22% in 6 months"), the placement change will have limited impact. Test testimonial content before placement.
Test 3: Form Field Reduction
Hypothesis template: "Because our form analytics show [X% field-level abandonment at the phone number / company size / job title field], we believe removing [specific field(s)] from the lead form will increase form completions for [first-time visitors on paid landing pages], because reducing the perceived cost of form completion by eliminating fields that signal high-pressure sales follow-up removes the primary abandonment driver."
Where to test: Lead generation forms, free trial sign-up flows, newsletter subscription forms. Any form with more than 3-4 fields.
Realistic lift range: 10-40% relative improvement in form completion rate. Field reduction tests are among the highest-impact tests in B2B. The lift is correlated with how friction-heavy the removed field is — phone number removal tends to outperform removing optional fields.
Primary metric: Form completion rate. Include a lead quality metric if you have one — removing a qualification field may increase volume but reduce quality.
After a win: Continue removing or simplifying fields iteratively. Test replacing a text input with a dropdown for complex fields (company size as a range selector vs. free text). Test multi-step form design (ask one question per screen for high-field forms).
After a loss: Check whether the removed field was actually driving abandonment (field-level analytics should confirm this before you run the test). If you removed the wrong field, run the test on the correct one.
Test 4: Pricing Page Layout
Hypothesis template: "Because [our pricing page exit rate is X% and session recordings show users spending the majority of time on feature comparison rows without clear resolution of which tier fits their situation], we believe [adding visual hierarchy with a highlighted 'Most Popular' tier and feature-led differentiation language per tier] will increase Business tier selection for first-time visitors comparing plans, because social proof anchoring ('most popular') combined with clearer tier differentiation reduces the decision paralysis of choosing between undifferentiated-seeming options."
Where to test: Pricing pages with 3 or more tiers. SaaS, subscription services, multi-tier products.
Realistic lift range: 5-20% relative improvement in conversion to paid (or tier selection), with meaningful variation in tier mix. High-variance test — pricing changes affect not just conversion rate but revenue per customer.
Primary metric: Conversion to paid from pricing page. Secondary: revenue per new customer (track tier mix, not just conversion volume).
After a win: Test pricing anchoring (order of tiers, whether showing a higher tier first increases middle-tier selection), annual vs. monthly toggle visibility, and free trial availability per tier.
After a loss: Investigate whether the issue is price perception (too expensive relative to alternatives) vs. decision complexity (can't tell which tier is right). These require different solutions. User interviews on the pricing page are often the most efficient diagnostic.
**Pro Tip:** Never run a pricing test without tracking revenue per customer, not just conversion rate. A test that increases conversion by 10% but shifts the tier mix toward lower-priced plans can be revenue-negative. The metric that matters is revenue per new customer, not conversion volume.
Test 5: Hero Headline Value Proposition
Hypothesis template: "Because our homepage bounce rate for organic search visitors is X% and user session data shows an average of [Y seconds] time-on-page for non-converting visits (suggesting visitors don't quickly understand what we offer or why it matters to them), we believe changing the hero headline from [feature/category-led copy] to [specific outcome-led copy for our primary customer segment] will reduce bounce rate and increase homepage-to-sign-up CTR for organic search visitors, because outcome-led copy answers the visitor's implicit question ('what will this do for me?') before requiring them to invest attention in reading further."
Where to test: Homepage hero, top-of-funnel landing pages, paid search landing pages.
Realistic lift range: 5-15% relative improvement in bounce rate; 3-12% lift in downstream conversion. Headlines have high variance — a headline that matches the mental model of a specific segment very well can produce outsized results; a headline change that misses the segment produces no improvement.
Primary metric: Homepage-to-sign-up CTR or trial start rate. Bounce rate as a secondary diagnostic, not the primary outcome.
After a win: Test headline specificity (more specific customer segment, more specific outcome claim), test subheadline that supports the new headline, and test whether the same framing works in paid ads.
After a loss: Segment results by traffic source. Organic search, direct, and paid often have very different mental models when they arrive on a homepage. A headline that fails in aggregate may be winning for one segment and losing for another.
Test 6: Navigation Simplification
Hypothesis template: "Because our navigation currently presents X top-level items and analytics show [Y% of visitors use the navigation at all on key landing pages], we believe reducing navigation options from [X items] to [fewer, more consolidated items] on high-intent conversion pages will increase conversion rate for paid traffic visitors, because reducing the number of available exit paths and decisions reduces cognitive load and navigation-induced distraction for visitors who arrived with a specific intent."
Where to test: Landing pages (especially paid), checkout flows, sign-up pages. Any high-intent page where navigation provides more ways to exit than reasons to stay.
Realistic lift range: 5-20% relative improvement for landing pages with full navigation. Higher impact seen on paid landing pages where visitor intent is high and navigation provides unnecessary distraction.
Primary metric: Primary page conversion rate.
After a win: Test completely removing navigation on your highest-intent paid landing pages. Test minimal footer navigation vs. full footer.
After a loss: The navigation may not be the distraction — investigate whether visitors are leaving because of navigation or because of missing information further down the page.
Test 7: Checkout Progress Indicators
Hypothesis template: "Because our checkout abandonment rate is X% between step 1 and step 2, and user research indicates visitors don't know how many steps remain in the checkout process, we believe adding a clear progress indicator showing step count and current position will reduce abandonment between steps 1-3 for logged-in users, because perceived proximity to completion (the goal gradient effect) increases commitment to finishing a multi-step process once the end is visible."
Where to test: Multi-step checkout flows, multi-step sign-up forms, onboarding flows with 3+ steps.
Realistic lift range: 3-12% relative improvement in funnel completion rate. Higher impact when the funnel has 4+ steps and current step count is invisible to users.
Primary metric: Funnel completion rate (full checkout or sign-up completion, not intermediate step CTR).
After a win: Test progress indicator design (percentage vs. step count vs. named steps), test whether labeling steps reduces abandonment further ("Step 2 of 3: Payment Details").
After a loss: Check whether abandonment is concentrated at a specific step. If 80% of abandonment happens at a single step regardless of progress indicator visibility, the issue is that step's content, not the navigation context.
Test 8: Mobile CTA Stickiness
Hypothesis template: "Because mobile conversion rate on our key landing pages is [X]% vs. [Y]% on desktop, and heatmap data shows mobile users frequently scroll past the primary CTA without clicking, we believe adding a sticky bottom CTA bar on mobile that persists through scroll will increase mobile conversion rate for new visitors, because maintaining CTA visibility throughout the page removes the friction of requiring users to scroll back up to convert after consuming the page content."
Where to test: Any high-value page with significant mobile traffic and a primary CTA in a fixed position. Product pages, landing pages, pricing pages.
Realistic lift range: 8-25% relative improvement in mobile conversion rate. Mobile-specific tests often show higher relative lift because mobile UX is frequently worse to start with.
Primary metric: Mobile conversion rate (isolated to mobile segment).
After a win: Test sticky CTA copy (should it match the primary CTA or be more specific to the mobile context?), test whether desktop also benefits from a persistent CTA.
After a loss: If the sticky CTA is being seen but not clicked, the issue is likely the offer itself or the page's ability to build intent before the CTA appears — not the mechanics of CTA visibility.
**Pro Tip:** Always segment A/B test results by device type when running tests that include CTA or layout changes. Mobile and desktop users often respond to the same change in opposite directions. A "neutral" aggregate result can mask a winning mobile result and a losing desktop result.
Test 9: Trust Signal Prominence
Hypothesis template: "Because our add-to-cart rate is X% but checkout completion rate is only Y% (suggesting users are abandoning due to trust barriers, not intent barriers), we believe moving security badges, customer count, and money-back guarantee copy above the fold on the checkout page — directly adjacent to the payment form — will increase checkout completion for first-time customers, because trust signals at the point of maximum anxiety (payment entry) directly address the risk perception that causes checkout abandonment."
Where to test: Checkout pages, high-consideration purchase pages, any form asking for sensitive information (payment, personal details, company information).
Realistic lift range: 3-10% relative improvement in form submission or checkout completion. Higher end when trust signals are currently absent or buried.
Primary metric: Checkout completion rate or form submission rate.
After a win: Test trust signal specificity (generic "Secure Checkout" vs. "[X] customers served" with a specific number vs. specific certifications relevant to your industry).
After a loss: If trust signals are already prominent and the test is neutral, the abandonment driver may be price, form friction, or shipping cost revelation — not trust. Use exit surveys to diagnose.
Test 10: Urgency and Scarcity Signals
Hypothesis template: "Because our pricing page conversion rate drops by [X%] on days when no promotional pricing is active, and session recordings show users navigating to the pricing page multiple times before converting, we believe adding a time-limited offer indicator ('Offer ends Sunday') to the pricing CTA will increase same-session conversion rate for returning visitors who've viewed the pricing page 2+ times, because loss aversion and the perceived cost of delay motivates completion of a decision that has already been partially made."
Where to test: Pricing pages, checkout pages, pages targeting users in a consideration-to-decision transition. Most effective for returning visitors who haven't yet converted.
Realistic lift range: 5-15% relative improvement in conversion rate for the targeted segment. Critically: urgency signals lose effect if they're permanent or clearly false. Honest, time-limited urgency (a real promotion ending at a real date) outperforms perpetual "only 3 left!" messaging, which users correctly identify as fabricated.
Primary metric: Same-session conversion rate for the targeted segment.
After a win: Test urgency copy specificity (date-based vs. quantity-based vs. feature-based), test delivery mechanism (page banner vs. inline vs. email reminder for non-converting visitors).
After a loss: If urgency signals aren't moving the needle, either the user's decision isn't time-sensitive (they're still in research mode) or the barrier to conversion is something other than procrastination (price, trust, fit). Urgency is effective when the user has already resolved their objections but is delaying — it doesn't work when objections are unresolved.
Common Mistakes
Mistake 1: Picking tests because they seem like good ideas, not because data points to them as the highest-impact opportunity. Use funnel analysis, heatmaps, session recordings, and user interviews to identify where users are dropping off and why. That's where high-impact tests live.
Mistake 2: Running tests on low-traffic pages where the required duration makes them impractical. If a page gets 1,000 visitors/week and you need 48,000 per variation, you're looking at a year-long test. Either increase traffic to the page first or test on a higher-traffic page.
Mistake 3: Optimizing for clicks instead of downstream conversions. A CTA test that increases clicks by 20% but doesn't improve sign-ups (or worsens lead quality) is not a win. Always track the conversion chain, not just the immediate click.
Mistake 4: Not documenting what you learned from each test. Wins tell you what works. Losses tell you what doesn't — and often what to test next. Both have value. Archive every result with the mechanism you were testing and the interpretation.
How to Build the Next 10 Tests
After running these 10, you'll have data on which mechanisms are most effective in your specific product/market context. Use that to build the next round:
- If CTA copy tests consistently win: go deeper on copy frameworks (FOMO vs. clarity vs. authority framing)
- If trust signal tests win: segment by customer type and test trust signals specific to each segment's concerns
- If form reduction tests win: explore progressive disclosure (multi-step forms that collect more fields after initial commitment)
- If urgency tests win: test personalization (urgency signals triggered by specific behavioral conditions, not shown to all users)
The compound effect of a well-run test program is that each test generates multiple hypotheses for the next cycle. Within 12-18 months of consistent testing, your hypothesis library should be generating test ideas faster than your team can run them — which is exactly the position you want to be in.
What to Do Next
- Run the ROI calculation for each of these 10 tests against your specific metrics. Rank by expected lift × monthly conversions × revenue per conversion ÷ implementation cost.
- Pick the top 3 by ROI and write full hypotheses using the template from the hypothesis writing guide.
- Verify traffic is sufficient before building: calculate required sample size, divide by weekly page visitors, confirm test duration is under 8 weeks.
- Launch with Stats Engine enabled, a pre-registered primary metric, and a defined end date.
For a complete experimentation workflow — from hypothesis documentation to Stats Engine configuration to post-test archiving — see the Optimizely Practitioner Toolkit at atticusli.com/guides/optimizely-practitioner-toolkit.