Can my agent (Claude or Codex) run the test loop?

Yes. Connect the hosted MCP server with one command, claude mcp add humblytics --transport http https://mcp.humblytics.com/v1, and every A/B testing surface, recommendations, test creation, variant changes, results, resolves to a tool call. It reads everything and changes nothing without you: every write waits on your approval, and applying a winner stays a human decision. The MCP is included on every plan, and a plain REST API is there if you prefer it.

Will the testing script slow down my website?

The script loads asynchronously, so it does not block page rendering, and only the elements under test are held back until their variant is applied. Like any third-party script it adds some load, so check performance on your own pages after installing.

How do you calculate statistical significance?

We use frequentist statistics with automatic sample size calculation. The platform shows real-time confidence intervals and tells you when a variation has reached statistical significance, no manual calculations needed.

What happens when a test reaches significance?

You decide. When a variant hits 95% confidence, the dashboard surfaces it for review and you apply the winner in one click, that's when the variant becomes the default for 100% of traffic. The original control stays live until you click Apply, so there are no surprise deploys.

A/B testing

A/B testing that measures revenue.

Edit any page without code, split your visitors between two versions and keep the one that earns more revenue. Claude or Codex can set up the test for you to approve.

Start free trial Book a demo

14-day trial · Card required to start · Cancel anytime

Edit the page, not the code

Click any headline, button or section and change it in place. Your original stays live until the test runs.

Headline test/pricing

Set the goal, split and stop rule

Pick a goal, split traffic and set a stop rule. Revenue is a goal, not an afterthought.

New experiment/pricing

2 of 3

Goal

Stripe revenuePayments per visitor

Click“Start free trial”

Form submitSignup form

Traffic splitControl 50%Variant B 50%

Stop ruleAt 95% confidence or 30 days

BackStart test

Ask your data a question

Connect Humblytics to Claude or Codex. Inspect a funnel and turn the answer into a test you review first.

Which page earns the most, and what should we test?

2 tools usedPages breakdown · 42 pagesFunnel query · 4 steps, 8,400 sessions

/pricing earns $18,40038% of revenue

68% of visitors leave before the trial CTA. Test a clearer headline and move the CTA above the fold.

/solutions/analytics earns $9,750

20% of revenue, mostly from organic search.

Tests scored in conversions and revenue

Every variant gets a clear winner, a confidence level and the lift it earned. Review the winner, then apply it.

Pricing page · headline testWinner found

Visitors: 2,480
Goal: Stripe revenue
Lift: +59.4%95% confidence

VariantConv.$/visitor

Control3.2%$1.12

Variant B5.1%$1.79

Apply variant BView report

Updated June 2026

What is revenue A/B testing?

Revenue A/B testing scores each variant by the revenue it produced, not clicks or conversion rate.

Humblytics shows each version to a share of your visitors and matches every sale in Stripe to the version the buyer saw. You keep the version that earned the most. You build versions in a visual editor, or Claude or Codex can draft them for you. Nothing goes live until you approve it.

How a split test assigns visitors.

Each visitor is assigned to a version at random, in the split you set, before the page is shown. Move across the figure to change the split.

How split URL testing works

Back lane: original URL (A)Switch blade and split settingEvery visitor lands on the original URL first. Before the page is shown, the switch assigns them at random in the set proportion: kept on the back lane (A) or redirected down the front lane to the variant URL (B). Each lane has its own counter, so a redirected visitor is counted once, on the variant. Over many visitors the trays settle near the split, but any short run can drift, which is one reason a test needs enough traffic.

How Humblytics compares with other testing tools.

VWO, Optimizely and AB Tasty are capable platforms, but they score on clicks and conversion rate and gate pricing behind sales. Humblytics ties every test to Stripe revenue at a flat price.

Humblytics vs VWO

FeatureHumblyticsVWOOptimizelyAB Tasty

Winners scored in Stripe revenue

No-code visual editor

AI assistant drafts tests, scored in Stripe revenue

Stops at 95% confidence

Transparent price from $79/mo

No sales call to start

What customers say.

Results from teams that use Humblytics on their websites.

See all success stories

7×

Revenue in 90 days

“Heatmaps showed 70% of users never scrolled to our booking button. We moved it, conversion jumped, and we shipped 15 tests in 90 days with zero dev tickets.”

Jani T

Co-Founder, MatchDay Health

Story

+63%

Conversion lift

“A benefit-oriented copy test drove a 63% higher conversion rate, and the attribution data floored my Meta ads agency.”

Dom Trovato

Founder, The Host Report

Story

+28%

Conversion lift

“A/B testing across every client site lifted conversion rates by 28%, rolled out in a week with no per-site dev work.”

Andrew

Co-Founder, Dedication Agents

Story

Choose the plan that fits your website.

Analytics, heatmaps, funnels, A/B testing and revenue attribution on every plan. Start with 14 days free.

Business

Scale

$239

Save $480

Per month, billed $2,868 yearly

For growing teams and several brands with more volume.

1M events/month
Unlimited active A/B tests, funnels & heatmaps
5 team members
Everything in Business, at higher volume

Book a demo

14-day trial available

Enterprise

Custom

Get a custom quote

For organizations with custom capacity and support needs.

Unlimited events and feature usage
Unlimited team members and data retention
Custom API and integrations
Priority support and dedicated specialist

Talk to sales

Scale1M events · 5 team members$239/mo EnterpriseCustom capacity and supportCustom

Frequently Asked Questions.

Answers to common questions. You can also ask our AI assistant about your site.

See all questions

Our visual editor lets you create split tests by pointing and clicking on your live page. Change headlines, buttons, images, and layouts, then set traffic allocation and go live. One script handles the testing logic. No tag managers, no code deployments, no engineering sprints.

Plan your test before you start it.

Free, no-signup calculators that run in your browser. Then run the test for real with revenue-scored results.

Sample size calculatorHow much traffic a valid test needs.Open Test duration calculatorHow many days before you can call it.Open Significance calculatorDid the variant actually win?Open Conversion rate calculatorYour rate vs. industry benchmarks.Open AI A/B test ideasGet ranked test hypotheses for any URL.Open Best A/B testing tools (2026)How 12 platforms compare, ranked.Open

Start your first A/B test.

Add the script to your site, connect Stripe and see the revenue from each page. Then test your first change.

Start free trial Book a demo

14-day trial · Card required to start · Cancel anytime

/pricing · headline testWinner found

Control$1.12

Variant B$1.79

+59.4%revenue per visitor · 95% confidence

A/B testing that measures revenue.

Edit the page, not the code

Set the goal, split and stop rule

Ask your data a question

Tests scored in conversions and revenue

What is revenue A/B testing?

How a split test assigns visitors.

How Humblytics compares with other testing tools.

What customers say.

Choose the plan that fits your website.

Business

Scale

Enterprise

Frequently Asked Questions.

How do I run A/B tests without developers?

Can my agent (Claude or Codex) run the test loop?

Will the testing script slow down my website?

How do you calculate statistical significance?

What happens when a test reaches significance?

Ask AI a questionAnswers about your site

Plan your test before you start it.

Start your first A/B test.