A/B Test Log Spreadsheet: Track Every Experiment That Matters

An A/B test log is a centralized spreadsheet where you document every experiment your team runs, from initial hypothesis through results and the business decision that followed. Without one, teams repeat failed tests, forget what worked, and can’t prove the cumulative impact of their testing program to leadership.

A/B test log: A structured record of every split test or multivariate experiment, capturing the hypothesis, test design, sample size, statistical results, business impact, and the key learning extracted from each test.

VWO’s 2024 State of Experimentation report found that companies running 10+ tests per month are 2.7x more likely to report significant revenue growth than those running fewer than 3. But running tests isn’t enough. VWO also found that 62% of CRO teams have no centralized documentation of past experiments. That means they’re re-testing ideas, missing patterns across tests, and unable to build on previous wins. This template fixes that. It gives you a backlog to prioritize ideas, a live tracker for running experiments, a results archive with revenue impact calculations, and a learnings database that turns individual test results into institutional knowledge.

• Test Backlog: Every test idea starts here. Write your hypothesis in “If we [change], then [metric] will [improve/decrease] because [reason]” format. Score each idea on Impact (1-10), Confidence (1-10), and Ease (1-10). The ICE score calculates automatically, and rows sort by priority. Dropdown menus categorize tests by type: copy, layout, CTA, pricing, form, navigation, or media.
• Active Tests: When you move a test from backlog to live, it goes here. Record the start date, planned end date, sample size target (use the built-in calculator row that estimates required sample size based on your baseline conversion rate and minimum detectable effect), and a plain-English description of what the control and variant look like. Status dropdown options: Setting Up, Running, Minimum Sample Reached, Analyzing, Complete.
• Results Log: After each test concludes, record the winner (Control, Variant A, Variant B, or Inconclusive), percentage uplift, statistical confidence level, and business impact. The revenue impact column uses a formula: (monthly visitors to tested page) x (conversion rate lift) x (average order value or lead value). This annualizes automatically. Over 12 months of testing, this column shows cumulative revenue impact.
• Learnings Database: The most underrated tab. After each test, distill the result into a one-sentence learning: “Short-form testimonials (under 20 words) outperform long testimonials on pricing pages by 18%.” Tag it by category (social proof, CTAs, forms, pricing, copy). Six months later, when someone proposes a new test, search this tab first. The McGaw.io testing framework (2024) recommends maintaining a learnings database as the single most valuable output of any experimentation program.

1. Dump every test idea into Tab 1. Don’t filter at this stage. Landing page headline changes, checkout flow modifications, email subject lines, pricing page layout, CTA button color and copy. Get them all in. Then score each one with ICE and let the math sort your priorities.
2. Move the top-scored idea to Tab 2 when you’re ready to run it. Fill in the control and variant descriptions in plain language. Set your sample size target using the built-in formula (you’ll need your current conversion rate and the minimum uplift you’d consider meaningful). Set start and end dates.
3. Don’t stop tests early. This is the most common mistake. Mark the status as “Running” and don’t touch it until you hit your sample size target. Peeking at results mid-test and calling winners early inflates your false-positive rate from 5% to as high as 30%, according to Evan Miller’s analysis of sequential testing errors.
4. Record results in Tab 3 within 24 hours of test completion. Include the winner, uplift, confidence level, and your team’s decision (Implement, Iterate, or Archive). Calculate the monthly revenue impact using the formula provided. This is the number that justifies your testing program to leadership.
5. Extract the learning into Tab 4. One sentence. Specific. Actionable. “Reducing form fields from 7 to 4 increased lead form completion by 22% on service pages.” Not “shorter forms are better.” The specific detail is what makes the learning reusable.