Knowing what to A/B test on your eCommerce store is the difference between rapid revenue growth and wasted months testing things that barely move the needle. The unfortunate truth is that most eCommerce teams waste their first several tests on low-impact changes, button colors, font sizes, and hero image tweaks, while ignoring high-impact opportunities in the checkout flow, product pages, and pricing presentation.
This guide gives you a structured, prioritized framework for choosing what to test first. We will use the ICE scoring framework to objectively rank test ideas by their potential impact, and provide hypothesis templates so you can start running meaningful tests immediately.
The ICE Scoring Framework
ICE stands for Impact, Confidence, and Ease. Each test idea is scored from 1 to 10 on each dimension, and the average becomes the ICE score. Tests with higher scores get prioritized.
| Dimension | What It Measures | How to Score (1-10) |
|---|---|---|
| Impact | How much will this move the conversion rate or revenue if it wins? | 10 = transformative (30%+ lift), 5 = moderate (10-15% lift), 1 = marginal (<2% lift) |
| Confidence | How sure are you this will produce a positive result? | 10 = strong data/research supports it, 5 = industry best practice, 1 = pure guess |
| Ease | How easy is this to implement and test? | 10 = one-hour change, 5 = needs developer for a day, 1 = major rebuild |
ICE Score = (Impact + Confidence + Ease) / 3
For example, a test to simplify the checkout page by removing three unnecessary fields might score: Impact 8, Confidence 7, Ease 6 = ICE Score of 7.0. That is a strong candidate for your testing roadmap.
High-Impact Tests (ICE Score 7+)
These tests target the areas of your funnel where friction has the largest effect on revenue. Start here.
1. Checkout Flow Optimization
Expected Impact: 10-35% conversion lift
Why it matters: The checkout is where purchase intent is highest and abandonment is most costly. Roughly 17 percent of shoppers abandon due to a complicated checkout process.
Test ideas:
- Single-page checkout vs. multi-step checkout
- Remove optional fields (company name, phone number, address line 2)
- Guest checkout as default vs. account creation required
- Add express payment buttons (Apple Pay, Google Pay, Shop Pay) above the fold
- Progress indicator showing checkout steps
- Add trust badges near the payment form
- Show order summary throughout the checkout process
Hypothesis template: “By simplifying the checkout from [current number] fields to [reduced number] fields, we will reduce checkout abandonment by [X]% because user research shows that [data point supporting the change].”
2. Call-to-Action (CTA) Optimization
Expected Impact: 5-25% conversion lift
Why it matters: The add-to-cart and buy-now buttons are the most critical interaction points on your product pages. Their design, copy, placement, and prominence directly affect whether visitors take the next step.
Test ideas:
- CTA copy: “Add to Cart” vs. “Buy Now” vs. “Get Yours” vs. “Add to Bag”
- CTA button size and color contrast (ensure it is the most visually dominant element)
- Sticky add-to-cart bar on mobile that follows the user as they scroll
- Adding microcopy under the CTA (“Free shipping” or “30-day returns” or “In stock, ships today”)
- CTA placement: above the fold vs. after product details
Hypothesis template: “By changing the CTA text from ‘[current text]’ to ‘[new text]’ and adding ‘[microcopy]’ below, we will increase add-to-cart rate by [X]% because it addresses the customer concern of [specific concern].”
3. Pricing and Offer Display
Expected Impact: 10-40% conversion lift
Why it matters: How you present pricing, discounts, and shipping costs has a massive psychological impact. The number-one reason for cart abandonment is unexpected costs at checkout.
Test ideas:
- Show savings as dollar amount vs. percentage (“Save $24” vs. “Save 20%”)
- Free shipping threshold prominently displayed vs. hidden
- Compare-at pricing (crossed out original price) vs. simple pricing
- Payment installment messaging (“4 payments of $12.50”) vs. full price only
- Urgency elements: stock levels, countdown timers for offers (use honestly)
- Bundle pricing vs. individual product pricing
Hypothesis template: “By displaying the free shipping threshold of $[amount] in the cart alongside a progress bar, we will increase average order value by [X]% because customers will add items to qualify for free shipping.”
4. Product Page Layout
Expected Impact: 8-25% conversion lift
Why it matters: The product page is where the buying decision happens. Its layout determines whether customers get the information they need to feel confident purchasing.
Test ideas:
- Image gallery layout: carousel vs. vertical scroll vs. grid
- Product description format: long-form vs. tabbed vs. accordion
- Review placement: above the fold with star rating vs. below the fold
- Social proof elements: “X people viewing this” or “Y sold this week”
- Video on product page vs. images only
- Size guide or comparison tool placement
- Related products section placement and algorithm
Hypothesis template: “By moving customer reviews from below the fold to directly under the product title as a star rating with review count, we will increase add-to-cart rate by [X]% because visible social proof reduces purchase anxiety.”
Medium-Impact Tests (ICE Score 4-6.9)
These tests can produce meaningful lifts but typically have a smaller effect than the high-impact category. Move to these after exhausting the high-impact opportunities.
5. Navigation and Site Search
Expected Impact: 5-15% conversion lift
Test ideas:
- Mega menu vs. simple dropdown navigation
- Search bar prominence: always visible vs. icon-triggered
- Search autocomplete with product images and prices
- Category page layouts: number of products per row, filter placement
- Breadcrumb navigation presence and styling
6. Filtering and Sorting
Expected Impact: 5-12% conversion lift
Test ideas:
- Sidebar filters vs. horizontal filter bar
- Default sort order: bestsellers vs. newest vs. price low to high
- Filter by: color swatches vs. checkboxes
- Show product count per filter option
- Sticky filters on mobile vs. top-of-page only
7. Homepage Hero Section
Expected Impact: 3-10% conversion lift
Test ideas:
- Single hero with strong CTA vs. rotating carousel (carousels almost always lose)
- Product-focused hero vs. lifestyle/brand hero
- Value proposition headline variations
- Hero with video background vs. static image
- Featured products on homepage vs. category links
8. Cart Page Optimization
Expected Impact: 5-15% conversion lift
Test ideas:
- Cart page vs. slide-out cart drawer
- Cross-sell recommendations in cart
- Free shipping progress bar
- Remove the “Continue Shopping” button (reduces cart exits)
- Show payment method icons in cart
- Persistent coupon code field visibility
Lower-Impact Tests (ICE Score Below 4)
These tests rarely produce meaningful conversion lifts on their own. Avoid them until you have exhausted higher-impact opportunities unless you have very high traffic.
9. Color and Visual Tweaks
Expected Impact: 0-3% conversion lift
Test ideas:
- Button color changes (red vs. green vs. blue)
- Background color variations
- Icon style changes
Why these are low priority: Color tests rarely produce statistically significant results. When they do, the effect is usually tiny. Your time is better spent on structural and content changes.
10. Typography and Micro-Copy
Expected Impact: 0-2% conversion lift
Test ideas:
- Font family changes
- Font size adjustments
- Minor wording changes in non-CTA text
A word of caution: testing low-impact changes when you have limited traffic is a recipe for inconclusive results and wasted time. See our guide on A/B testing for small eCommerce stores for traffic requirements and practical alternatives.
11. Footer and Ancillary Pages
Expected Impact: 0-2% conversion lift
Test ideas:
- Footer link structure
- About Us page layout
- FAQ page format
Writing Strong Test Hypotheses
Every A/B test should start with a well-structured hypothesis. A good hypothesis has four components:
- Observation: What did you notice in your data or research?
- Change: What specific change will you make?
- Expected outcome: What metric do you expect to improve, and by how much?
- Rationale: Why do you believe this change will work?
Template:
“Based on [observation/data point], we believe that [specific change] will [expected outcome/metric improvement] because [rationale].”
Strong hypothesis example:
“Based on session recordings showing that 40 percent of mobile users never scroll past the product images, we believe that adding a sticky add-to-cart bar on mobile will increase mobile add-to-cart rate by 15 percent because it keeps the purchase action accessible regardless of scroll position.”
Weak hypothesis example:
“We think making the button green will increase conversions because green means go.”
The difference is clear: strong hypotheses are rooted in data and customer behavior. Weak hypotheses are based on assumptions and generalizations.
Building Your Testing Roadmap
Here is how to build a structured testing roadmap using the ICE framework:
- Audit your store – Use an eCommerce CRO audit checklist to identify all potential issues and opportunities
- List all test ideas – Write down every potential test, no matter how small
- Score each idea – Apply ICE scoring (Impact, Confidence, Ease) to every idea
- Rank by ICE score – Sort from highest to lowest
- Validate with data – Check your analytics for the top-ranked ideas. Does the data support the hypothesis?
- Run one to two tests at a time – Depending on traffic, run your highest-scoring test first
- Document results – Record every test with its hypothesis, variation details, duration, sample size, and outcome
- Iterate – Winners become the new baseline. Losers provide learnings for the next test
Sample ICE Scoring Table
| Test Idea | Impact | Confidence | Ease | ICE Score |
|---|---|---|---|---|
| Simplify checkout to 3 fields | 9 | 8 | 5 | 7.3 |
| Sticky mobile add-to-cart bar | 7 | 7 | 8 | 7.3 |
| Free shipping progress bar in cart | 7 | 8 | 9 | 8.0 |
| Add trust badges to checkout | 6 | 6 | 9 | 7.0 |
| Change button from blue to green | 2 | 2 | 10 | 4.7 |
| Redesign product page layout | 8 | 6 | 3 | 5.7 |
In this example, the free shipping progress bar has the highest ICE score and would be tested first, despite not being the highest-impact change. Its high ease and confidence scores make it the best starting point.
Stop Guessing, Start Testing Strategically
Knowing what to A/B test on your eCommerce store is half the battle. The other half is executing tests properly, interpreting results correctly, and building a compounding optimization program.
At MDigital, we build and manage testing programs for eCommerce stores that want to grow through data-driven optimization. Our CRO specialists identify your highest-impact testing opportunities, write the hypotheses, design the variations, and analyze the results so you can focus on running your business.
Schedule a free CRO consultation to discover the top testing opportunities hiding in your store right now.
