Marketing Digital Presence

Run conversion experiments to turn more visitors into leads

How do you turn more of your website visitors into leads?

In an A/B test, different visitors see two versions of a page. Your conversion rate is the share of visitors who become leads. The Experiment Designer agent records the conversion rate of your homepage, pricing page and top landing pages. The Experiment Designer agent designs a test of one change. The Results Analyst agent puts the winning version live.

Published Updated

What you get
Conversion experiment program · Saltmarrow Studio
First test complete
4
key pages with enough traffic to test
0.88%
homepage conversion rate with the winning button
Conversion rate before any test · homepage, pricing page and 3 landing pages 5 of 5 recorded
Set aside · retainer landing page, 250 visits a month Until its traffic grows
Test 1 · homepage call to action button Version B won
Homepage conversion rate · version A 0.29%, version B 0.88% Version B is live
Test log · result and learning of test 1 1 entry
Test 2 · homepage headline Starts on October 19

The Results Analyst agent runs one test at a time, as you chose.

See it run

First conversion test complete

Website visits a month
6,200 · 18 inbound leads a month
Testable pages with a finished test 1 / 4
Schedule
One test at a time
Each new test starts when the last test ends
Run log
  1. You
    “Set up A/B tests on the Saltmarrow Studio website. Turn more website visitors into leads.”
    Your instruction
  2. Baselines for 5 key pages Experiment Designer
    You counted a contact form submit as the conversion on each page. The Experiment Designer agent took the last 90 days of data from Google Analytics. A baseline is the conversion rate of a page before any test. The homepage has 2,400 visits a month and a baseline of 0.25%. The pricing page has 600 visits a month and a baseline of 0.83%. Four pages have 500 or more visits a month. You set aside the retainer landing page, with 250 visits a month, until its traffic grows.
  3. Homepage button test designed Experiment Designer
    Of the 4 pages, the homepage has the most traffic and the lowest conversion rate. The Experiment Designer agent chose the call to action button as the one element to change. Version A keeps the current button text, "Contact us". Version B has the button text "Book a 20-minute brand call". The hypothesis is that more visitors send the form when the button names a clear next action. You set 95% as the confidence level a winner must reach. Confidence is how sure you are that the result is not chance. The Experiment Designer agent set the test to run 6 weeks, so enough homepage visitors see each version.
  4. Version B won at 98% confidence Results Analyst
    The test ran from August 31 to October 11. Each version had 1,700 visitors. Version A had 5 form submits, a conversion rate of 0.29%. Version B had 15 form submits, a conversion rate of 0.88%. The Results Analyst agent confirmed 98% confidence, above your level of 95%. The Results Analyst agent named version B the winner.
  5. You
    Your approval of the program needed You
    On October 13, the Results Analyst agent put version B live for all homepage visitors. The Results Analyst agent added the hypothesis, result, gain and learning to the test log. The Results Analyst agent designed test 2, which changes the homepage headline. The Results Analyst agent finished a test on 1 of 4 pages with enough traffic. Approve the conversion testing program, so test 2 starts on October 19.
  6. Website analytics setup used Upstream
    Your website analytics setup comes from Set Up Website Analytics.
  7. Landing pages used Upstream
    Your landing pages come from Create Landing Pages.
  8. Website audit and fixes used Upstream
    Your website audit and fixes come from Audit the Website.
How it works

One change at a time, with a baseline for each page

You choose which action counts as a conversion on each page. The Experiment Designer agent records a baseline for each key page. You confirm which pages have enough traffic to test. The Experiment Designer agent writes a hypothesis about one change. The Results Analyst agent names a winner only when the result reaches your confidence level. The Results Analyst agent puts the winner live. The Results Analyst agent designs the next test.

What it needs
Conversion data from Google Analytics 4 (GA4) for the baselines
Traffic data to confirm that each page has enough traffic to test
Access to your website CMS or an A/B testing tool
Heat map data (where visitors click and scroll) from Hotjar or Clarity, if you have it
What you get

A conversion testing program with a written baseline for each key page and a first test with a clear hypothesis. The test result is confirmed at statistical significance. The winner is live, and the next experiment is queued in a running test log.

Run sequence

Start to finish

You start from your website analytics setup, your landing pages and your website audit and fixes. Then you, the Experiment Designer agent and the Results Analyst agent set up your conversion experiment program in four parts.

  1. 1
    Conversion Baselines

    Measure your key pages

    Experiment Designer records each page's conversion rate
    Experiment Designer records each page's traffic
    You confirm the pages with enough traffic
    Experiment Designer + You
  2. 2
    First Experiment Design

    Design the first test

    Experiment Designer picks the page with most to gain
    Experiment Designer picks one element to change
    Experiment Designer writes a one-change hypothesis
    Experiment Designer
  3. 3
    Test Result

    Read the test result

    Runs until 2 weeks or 200 conversions per version
    Results Analyst checks the sample size
    Results Analyst names the winner at your confidence level
    Results Analyst
  4. 4
    Live Winner and Next Test

    Act on the winning version

    Results Analyst puts the winner live
    Results Analyst updates the test log
    Results Analyst designs the next test
    Results Analyst + You
Before it starts

What Works asks you

A few questions in plain words. The answers set up the run for your business.

Which action counts as a conversion on each page: a form submit, a signup, a demo request, or a purchase?
YouForm submit or signup as the lead action
Which period gives a fair baseline: the last 30 days, the last 90 days, or a full season that evens out short spikes?
YouLast 90 days to even out swings
Pages with too little traffic for a test are set aside until their traffic grows. Is the list of pages to set aside correct?
YouYes, set aside the low-traffic pages for now
Which element most often changes your conversion rate: the headline, the call to action button, or the form length?
YouMain call to action text or button
What confidence level (how sure you are that the result is not chance) must a winner reach before you trust it: 90%, 95%, or higher?
You95% as the standard level
Should the next test stay on the same page to add more gains, or move to the next page in priority?
YouSame page, add another win
Who runs it

The agents behind this workflow

The Experiment Designer agent designs each test. The Results Analyst agent names each winner.

Playbook specialist

Conversion Copywriter

Owns this area

Writes and improves your homepage and landing pages, and runs the conversion experiments that raise their conversion rates.

On this workflow

Experiment Designer

Reads the analytics to choose the page with the most to gain, and designs a test that changes one element.

What it can do
Record the conversion rate and traffic of each pageFind the best page to test firstWrite a one-change hypothesis a test can prove wrongSize the test for statistical significanceSet up the test version in the testing tool

Results Analyst

Watches the running test, names the winner only at significance, and queues the next experiment.

What it can do
Watch a live test for sample size and significanceConfirm statistical significance before calling a winnerPut the winning version liveMaintain a running test log of learningsDesign the next experiment from what was learned
Jeanie, your right hand Works Crew

Jeanie leads a core crew of seven that runs your company day to day. They direct the specialist and workflow agents in each playbook, including the ones on this workflow.

Connect your tools

Bring what you already run

Works connects to the apps you already use.

  • Google Analytics The Experiment Designer agent connects to Google Analytics.
  • Unbounce The Results Analyst agent connects to Unbounce.
  • WordPress This workflow uses WordPress.
Other apps you can use
Your analytics tool

or any analytics tool you use

Your landing page tool
  • Leadpages
  • Instapage
  • Webflow
  • Carrd

or any landing page tool you use

These are starting points, not a limit. Connect the tools you already use.

Connect your work

The workflows before your conversion experiment program

Linked cards are live pages. The others are part of the Marketing playbook in Works.

Works · AI business OS

One workflow inside your whole business

In Marketing, Run Conversion Experiments is part of the Digital Presence work.

Prebuilt playbooks, plus your own
StrategyMarketingSalesCustomer SuccessPartnershipsProducts & ServicesOperationsFinancePeopleLegal & Compliance + Build your own
The Marketing playbook
74
Workflows
20
Specialist agents
147
Workflow agents
226
Templates
You're here
Run Conversion Experiments
One of 74 Marketing workflows · run by 2 workflow agents
Templates

What comes with it

Built for this workflow
A/B Test Results Model XLSX
A sheet where you calculate the results of an A/B test. It opens in Works after you sign up.
Experiment Backlog XLSX
A sheet where you list the experiments that you plan to run. It opens in Works after you sign up.
Experiment Hypothesis Canvas DOCX
A document where you write the hypothesis for an experiment. It opens in Works after you sign up.
Also included here
Homepage Conversion Checklist XLSX
A checklist to raise the conversion rate of your homepage. It opens in Works after you sign up.
Conversion Event Map XLSX
A sheet where you list the actions that count as conversions. It opens in Works after you sign up.
Fix Prioritization Matrix XLSX
A sheet where you rank the fixes for your website. It opens in Works after you sign up.
Website Audit Scorecard XLSX
A sheet where you score the results of your website audit. It opens in Works after you sign up.
Priority Page Tracker XLSX
A sheet where you track your priority pages. It opens in Works after you sign up.
Landing Page Conversion Checklist XLSX
A checklist to raise the conversion rate of a landing page. It opens in Works after you sign up.
Landing Page Wireframe Canvas DOCX
A document where you sketch the layout of a landing page. It opens in Works after you sign up.
Message Test Tracker XLSX
A sheet where you track your message tests. It opens in Works after you sign up.
Test Hypothesis Canvas DOCX
A document where you write the hypothesis for a test. It opens in Works after you sign up.

Each template opens inside Works after you sign up.

Questions

What people ask

Which page should you test first?

The Experiment Designer agent records the conversion rate and traffic of your homepage, your pricing page and your top landing pages. You confirm which pages have enough traffic to test. You set aside the other pages until their traffic grows. The Experiment Designer agent picks the page with the most traffic and the lowest conversion rate.

How long should a test run?

You choose when a test can stop, for example after 2 weeks or after 200 conversions per version. The Results Analyst agent checks the result against your confidence level. You choose that level, for example 95%. Confidence is how sure you are that the result is not chance. You also choose the longest time one test may run.

What happens when a test gives no clear result?

You decide what the Results Analyst agent does when the result is not clear. For example, the agent can extend the test for more data. The agent can also redesign the test with a bigger change. Or the agent can move on to the next page. The agent records each result in the test log.

What happens after a test finds a winner?

The Results Analyst agent puts the winning version live. You choose whether the winner goes to all visitors at once or to a share of visitors first. You choose what each test log entry records, for example the hypothesis, result, gain and learning. Then the Results Analyst agent designs the next test.

Turn more of your website visitors into leads

Sign up to record the baselines of your key pages.

Sign Up