How to run an A/B test for your real estate marketing

A/B testing means showing two versions of one piece of marketing to similar audiences at the same time, then keeping whichever performs better. For an agency, that could be two subject lines on a property alert, two headlines on a Facebook ad, or two versions of an appraisal button. It replaces guesswork with evidence from your own audience.

This guide covers what is worth testing, a 10 step process you can follow, the tools to use, and the mistakes that make a result misleading.

Why test instead of copying what works elsewhere

Advice like "red buttons convert better" or "emails sent on Tuesday perform best" comes from someone else's audience. Your vendors, landlords and buyers in your suburbs may behave differently. A test gives you an answer for your market.

It also protects your budget. If you are paying to run ads or send campaigns anyway, small improvements in click rate or enquiry rate carry across every future campaign.

What is worth testing in real estate marketing

Start with elements that sit close to the action you want, such as an enquiry, a click or an open. Here are realistic examples.

Email

• Subject line: "New listing matching your property alert" against "[First name], a new listing matches your property alert".
• Sender name: your agency name against your own name.
• Send time: weekday early morning against weekday midday. This is the one case where you deliberately change the timing.

Facebook and Instagram ads

• Image: a lifestyle shot against a plain exterior photo.
• Headline: a question against a statement.
• Call to action: "Get a free appraisal" against "Find out what your home is worth".

Website

• Button wording or position: the appraisal button above the fold against further down the page.
• Page or blog headline: "How to prepare your house for sale" against "12 ways to prepare your house for sale".
• Form length: asking for name and phone only against adding address and timeframe.

The 10 step process

1. Pick one variable. If you change the image and the headline together, you cannot say which one moved the result. Small, specific changes are easier to read than a total redesign.
2. Set one goal. Opens, clicks, enquiries or booked appraisals. Choose before you build version B, because the goal decides how you set the test up.
3. Build the challenger. Version A is what you run now. Version B changes only the variable you picked.
4. Split the audience. Most tools do this automatically. If you split a list by hand, make the groups random and equal, not "A to Sydney contacts and B to Newcastle contacts".
5. Plan the sample size. The more people who see each version, the less a result is down to luck. A list of 1,000 contacts might be split into two groups of 250, with the winner sent to the remaining 500. A list of 150 is too small to trust a single test, so repeat the test over several sends before drawing a conclusion.
6. Decide what counts as a win. Write down, before you start, how big a gap you need and how long you will wait. This stops you calling a winner the moment one version edges ahead.
7. Choose the tool. See the next section.
8. Run both versions at the same time. If A runs this week and B runs next week, a school holiday or an interest rate announcement could be the real cause of the difference.
9. Collect the data and leave it alone. Early numbers swing wildly. Low traffic pages need weeks, not days.
10. Analyse and record. Use the winner going forward. If there is no clear winner, keep the original and note that the variable did not matter. Then test something else.

Which tools to use

Most email platforms include a built-in A/B option for subject lines and sometimes content. Check your platform's help section for how it picks the winner and how long it waits.

For paid social, Meta Ads Manager has an A/B test option that compares versions of an ad, audience or placement. It splits the audience so the same person is not shown both versions. The exact menu names change from time to time, so check Meta's current help pages before you set one up.

For web pages, be careful. Google Analytics 4 measures what visitors do, but it does not run page tests on its own, and Google retired its free testing tool, Google Optimize, in 2023. If you want to test a page, you will need a separate testing tool or a developer. A simpler approach for small sites is to change one element, record the date, and compare enquiry numbers over comparable periods, while accepting that this is weaker evidence than a true split test.

Mistakes that ruin a result

• Testing too many things at once. This includes running two linked tests together, such as an email test and a test of the landing page it points to.
• Stopping early. A lead after one day is rarely a real lead.
• Judging on the wrong metric. A headline that gets more clicks but fewer enquiries has not won. Always measure against your goal from step 2.
• Testing trivial things on tiny audiences. If only 200 people will ever see a page each month, test bigger changes and be patient.
• Not recording results. Keep a simple spreadsheet with the date, variable, versions, audience size, result and what you decided.

A first test you can run this week

Take your next property alert or market update email. Write two subject lines that differ in one way only, for example with and without the recipient's first name. Send each to a random half of the list, and note the open rate after your chosen wait. Log the outcome, then apply the winning style to the next send and test something new.

To make sure the enquiries from those campaigns are being tracked properly, see our guides on UTM tracking and writing email subject lines. If your tests are about getting more leads from your site, how to generate more leads from your real estate website shows where to focus first.

Need a hand with testing?

Testing works best when it is set up properly and tracked from the first click to the booked appraisal. If you would like help planning tests for your ads, emails or website, you can book a time to talk with the Lead Fleet team.