HubSpot A/B Testing: How to Test Emails, Pages and CTAs

HubSpot A/B testing lets you compare two versions of a marketing email, an automated workflow email, a website or landing page, or a CTA, then keep the version that performs better. Email tests need Marketing Hub Professional or Enterprise, page tests need Content Hub Professional or Enterprise, and HubSpot recommends sending an A/B tested email to at least 1,000 contacts.

Checked against the HubSpot Knowledge Base on October 20, 2026.

What you can A/B test in HubSpot, and the plan each needs

Two details catch teams out. First, HubSpot's page testing articles list only Content Hub Professional and Enterprise, not Marketing Hub, and that applies to landing pages too. Second, you cannot A/B test a CTA inside a marketing email. HubSpot recommends testing the whole email instead.

How to run an A/B test on a HubSpot marketing email

You need edit permissions for marketing email. Then:

1. In your HubSpot account, go to Marketing > Email (click More first if you see it).
2. Click the name of a drafted email, or click Create email.
3. In the top right of the editor, click A/B test.
4. In the right panel, enter names for version A and version B. Use names that tell you what changed, such as "Short subject" and "Question subject."
5. In How many recipients should receive the A/B test, enter a percentage. Everyone outside the test receives the winning version.
6. Open the Winning metric dropdown and choose Open rate, Click rate or Click through rate.
7. In Test length, enter the number of hours the test should run. When that time is up, the remaining recipients get the winner.
8. Open the Fallback version dropdown and choose the version to send if the results are inconclusive.
9. Click Create test.
10. Edit both versions. Use the dropdown in the top right to switch between them.
11. When both are ready, click Review and send in the top right.

To change the test settings before sending, open the top right dropdown while viewing version A, select Manage test, make your changes and click Update. To remove the test, choose Manage test, click Delete test, type "delete" and confirm. That permanently deletes version B.

To stop a test that is already running, open the email from Marketing > Email, then on its performance page click Actions and select End A/B test.

What to change between versions

HubSpot suggests varying one of these at a time:

• Offer: for example, an ebook versus a whitepaper.
• Copy: plain paragraphs versus bullet points, or longer versus shorter text.
• Sender: an employee's email address instead of a generic department address.
• Image: a different image to see how conversions respond.
• Subject line: a different length, or added personalization.

Change one thing per test. If version B has a new subject line and a new button, you will not know which change moved the result.

Which winning metric should you choose?

HubSpot defines the three metrics differently, and the choice decides which version wins.

Be careful with open rate. HubSpot counts an open when the images in the email load or when a recipient clicks a link, and it notes that opens may be excluded by bot filtering or privacy screening, or never registered if the recipient does not load images. For anything below the subject line, a click based metric gives you a cleaner signal.

How many contacts do you need?

HubSpot's email A/B testing article says that for best results you should send an A/B tested email to at least 1,000 contacts. That is product guidance, not a guarantee of statistical significance. HubSpot's Marketing Blog post on email A/B test sample size (updated January 2, 2026) explains that the size you need depends on your baseline conversion rate, the smallest lift you want to detect and your confidence level, with 95 percent described as the industry standard. Its worked example, a 2 percent baseline and a 20 percent lift at 95 percent confidence, needs about 20,000 recipients per version.

The same post says test cells are typically 10 to 20 percent of the list, with the other 80 to 90 percent receiving the winner. Smaller lists need a larger share in the test.

Here is how the arithmetic plays out. The numbers below are simple splits, not HubSpot defaults.

What if you only have 900 recipients?

With 900 recipients you are below HubSpot's 1,000 contact recommendation, so a small test group followed by a winner send leaves each version with too few contacts to trust. The better distribution is a 50/50 split of the whole list: 450 contacts get version A and 450 get version B. HubSpot's German Marketing Blog (last updated January 29, 2020) describes the same approach for lists under 1,000 contacts, sending variant A to one half and variant B to the other.

Treat that result as a lesson for your next send, not a final answer. HubSpot's English sample size post recommends running the same test across a series of sends and aggregating the results to build a bigger data set.

How long should the test run?

HubSpot's test length field takes hours, and the knowledge base does not state a maximum. HubSpot's sample size post says to expect 85 percent or more of email results within the first 24 hours, and suggests 48 or 72 hours for slower audiences. Check your own past sends to see when engagement levels off, then set the test length to cover it.

Where to find HubSpot A/B test results

1. Go to Marketing > Email and click the name of your A/B email.
2. The Performance page summarizes the test. The winning version and the winning metric are highlighted in green.
3. Use the Email version dropdown in the upper left to switch between results for the email overall, version A and version B.

One reporting limit to know: in custom reports, you can only filter by the Email content ID of version A of an A/B test email.

How to A/B test automated emails in workflows

Workflow emails work differently from one-time sends. There is no test window and no automatic winner. HubSpot sends the variations across enrolled records in a 50/50 split, and that split builds gradually as more records enroll.

1. Go to Marketing > Email, click Create email, and create a new automated email or clone an existing one.
2. In the email editor, enable A/B testing and configure your options.
3. Click Review and Publish, then Publish in the right panel.
4. Go to Automation > Workflows and open or create a workflow.
5. Click the + icon in the timeline, select Send email, and choose your A/B email.

Rules to plan around:

• You must set up the A/B test before publishing. An automated email published without a test cannot be used for A/B testing later, even if you edit or unpublish it.
• Each A/B email can be used in one workflow only. Clone it to use it elsewhere.
• If the workflow uses Delay until an event happens, HubSpot cannot check which version an open or click came from.
• Metrics appear on the email details screen. When you choose a winner, the workflow sends only that version to future enrollees and archives the losing version.

You need Super Admin or Edit permissions for workflows, and Super Admin or Publish permissions for marketing emails. To test different automation paths rather than email content, HubSpot points to Random distribution by percentage branches in workflows.

How to A/B test a website or landing page

You need Content Hub Professional or Enterprise and Marketing Access with Edit and Publish permission.

1. Go to Content > Website Pages or Content > Landing Pages (click More first if you see it).
2. Click the name of a published page.
3. In the left sidebar, click the Test icon, then A/B test. You can also open the File menu and select New > Run A/B test.
4. Name each variation and click Create variation.
5. Edit the variation. HubSpot suggests testing CTAs, copy format, images or video, and form length.
6. Click Publish, then Publish now. Both versions go live at the same URL, each with its own preview URL.

Half of visitors see each version. A visitor sees the same version for the whole session, and the version is randomized again in the next session. If the page belongs to a multi-language group, run a separate test for each language.

To review results, hover over the page name, click Performance, then open the Test results tab and set the date range and frequency. When you have a winner, hover over it and click Choose as winner. The winner becomes the only live version and the other variation reverts to draft. To try a losing variation again, click Rerun this variation, which publishes it immediately.

HubSpot's sample size post gives a rule of thumb for page tests: divide the required sample size by your average weekly visitors, run in full week increments, and do not stop early.

Adaptive tests: up to five page variations

An adaptive test is the multi-variation option. You can test up to five page variations at a time, every variation must include a form, and HubSpot shows the best performer more often over time. Start it from the same Test icon by choosing Adaptive test, or from File > New > Run Adaptive test.

Once it is live, you can keep editing variations, but you can no longer add, remove or rename them. The Test results tab shows the variation with the highest submission rate and a Traffic distribution report. To finish, click End test, pick the variation to keep, confirm, and click Yes, end test. The other variations are archived.

How to A/B test a CTA

CTA testing needs Marketing Hub or Content Hub Professional or Enterprise, Super Admin or Edit permissions for CTAs, and an opt in to the A/B Testing for CTAs beta. It works on website pages, landing pages and blog posts.

1. Go to Marketing > CTAs.
2. Hover over the CTA, click Actions, and select Run a test.
3. Name the CTA and the variant (the default is "[CTA name] (B)"), then click Save.
4. Edit the variant in the CTA editor, then click Review and publish or Review and update.

Visitors are split 50/50. To see results, choose Actions > View performance and open the Test Results tab for clicks, views, click rate and, if the CTA has a form, submissions. Click Select a winner when you are ready. You cannot change the winner afterward, so wait for enough views on both variants. For a deeper walkthrough, see our guide on how to A/B test CTAs in HubSpot.

A worked example (hypothetical)

Imagine a software company with 10,000 newsletter subscribers that wants better click through on its monthly product email.

• Variable: the main button text only, "See what's new" versus "Try the new feature."
• Winning metric: Click through rate, because the subject line is the same in both versions.
• Test group: 20 percent, so 1,000 contacts get version A and 1,000 get version B.
• Test length: 24 hours, based on the company's past sends.
• Fallback version: version A, the current wording.

After 24 hours, HubSpot sends the winner to the other 8,000 subscribers. The team opens the Performance page, checks the green highlight, and records the variable, metric and result in a shared test log. Next month it tests button placement, keeping the winning text.

Common mistakes to avoid

• Testing more than one variable. You will not know what caused the difference.
• Using open rate for body changes. Pick Click through rate when the subject line is identical.
• Publishing a workflow email before adding the test. It cannot be converted afterward, so clone it and start again.
• Picking a CTA winner too soon. The choice is permanent.

Getting help with HubSpot testing

INSIDEA is an Elite HubSpot Partner and the growth operating system for modern businesses. If you want a testing plan that ties email and page tests to pipeline, our growth marketing team can build and run it with you.