In the world of digital advertising, it’s easy to get caught up in 'Last Click' attribution. However, just because someone clicked an ad before buying doesn't mean the ad caused the purchase—they might have bought from your Brisbane storefront or online shop anyway. Incrementality testing allows you to measure the 'lift' your ads actually provide, ensuring your marketing budget is being spent on growth, not just claiming credit for existing customers.
Why Incrementality Matters for Australian Small Businesses
With rising costs of living and tighter competition in the Australian market, every dollar of your marketing budget needs to work harder. Incrementality testing (often called 'Lift Testing') helps you identify which channels are driving new revenue and which are simply 'cannibalising' organic traffic. This guide will walk you through setting up a scientific test to find your true Return on Ad Spend (ROAS).
---
Prerequisites: What You’ll Need
Before you begin, ensure you have the following ready:
- Active Ad Accounts: Access to Meta Ads Manager or Google Ads.
- Conversion Tracking: A functional Meta Pixel or Google Tag Manager setup with enough historical data (at least 50+ conversions per week).
- Control over Budget: The ability to pause or segment ads for a specific period.
- Statistical Patience: A minimum test window of 14 to 30 days.
---
Step 1: Define Your Hypothesis
Every good experiment starts with a question. For an Australian e-commerce brand, it might be: "If I turn off my Google Brand Search ads, will my organic sales drop by the same amount, or stay the same?"
Write down exactly what you are testing. For example: "I believe Meta Advantage+ campaigns are driving a 15% incremental lift in total sales compared to a control group with no ads."
Step 2: Choose Your Testing Methodology
There are two primary ways for small businesses to test incrementality:
- Platform-Based Lift Tests: Using built-in tools like Meta’s 'Experiments' tool.
- Geo-Testing (The 'Aussie' Approach): Splitting your audience by geography. For example, showing ads to users in Sydney and Melbourne (Test Group) while suppressing ads for users in Brisbane and Perth (Control Group).
Step 3: Segment Your Audience (The Control vs. Test Group)
To get clear results, you must ensure your Control Group (the people who won't see ads) and your Test Group (the people who will) are as similar as possible.
If using Geo-Testing, choose regions with similar population sizes and purchasing habits. Don't compare a high-intent metro area like Sydney CBD with a rural town in the NT, as the baseline conversion rates will be too different.
Step 4: Determine Your 'Holdout' Percentage
A standard holdout is usually 10% to 20%. This means 10% of your potential audience is intentionally kept in the dark to see how they behave without your influence. While it feels like you are 'losing' sales, this is the only way to prove the value of the other 90%.
Step 5: Establish a Baseline Period
Before launching the test, look at your data from the previous 30 days. Note the conversion rates and total sales for both your intended Test and Control regions. This ensures that any change you see during the test is actually caused by the ads, not just pre-existing trends.
Step 6: Set Up the Experiment in Ads Manager
If using Meta:
- Go to Experiments.
- Click Create Test and select Brand Lift or Conversion Lift.
- Select your specific campaign or the entire account.
- Set your schedule (we recommend 28 days for the Australian buying cycle).
If using Google Ads:
- Go to Campaigns > Experiments.
- Create a Custom Experiment.
- Set your split to 50/50 or your desired ratio.
Step 7: Launch and 'Hands Off'
Once the test starts, the most important rule is: Do not touch it. Changing your creative, budget, or targeting mid-test will 'pollute' the data. In Australia, we often see seasonal spikes (like EOFY or Black Friday); try to run your tests during 'normal' trading periods to get the most accurate baseline.
Step 8: Monitor Data Integrity
Check your dashboard every few days to ensure the 'Control' group isn't accidentally seeing ads. This can happen if you have other 'Always On' campaigns running that weren't included in the experiment.
Screenshot Description: Look for a 'Confidence Level' percentage in your experiment dashboard. You want this to be above 90% before making decisions.Step 9: Calculate the Incremental Lift
Once the test concludes, use this formula:
Incremental Lift = (Conversions in Test Group - Conversions in Control Group) / Conversions in Control GroupIf your test group had 120 sales and your control group (adjusted for size) had 100 sales, your incrementality is 20%. This means 20% of your sales were directly caused by the ads, while 80% might have happened anyway.
Step 10: Analyse the ROAS vs. iROAS
Standard ROAS (Return on Ad Spend) is what the platform tells you. iROAS (Incremental ROAS) is the truth.
If your Meta dashboard says you have a 5x ROAS, but your incrementality test shows only 50% lift, your true iROAS is 2.5x. Use this figure to decide if the channel is actually profitable after COGS (Cost of Goods Sold) and Australian GST are factored in.
---
Pro Tips for Success
- Test One Variable: Don't test a new audience and a new creative at the same time. You won't know which one caused the lift.
- Consider the 'Halo Effect': Sometimes paid ads drive organic searches. Check your Google Search Console during the test to see if organic brand searches drop in the control regions.
- Seasonality Check: In Australia, retail peaks in November/December. Running an incrementality test during this time may give you inflated results that won't apply in February.
Common Mistakes to Avoid
- Ending the Test Too Early: Small businesses often panic after 3 days if they see a dip. Statistical significance takes time.
- Overlapping Audiences: If you are running Meta and Google ads simultaneously, a user in your 'Control' group for Meta might still see your Google ads, which spoils the purity of the test.
- Ignoring Offline Impact: If you have a physical shop in Brisbane, remember that digital ads drive foot traffic. Incrementality tests often underestimate this unless you use 'Store Visits' tracking.
---
Troubleshooting Common Issues
"My results show 0% incrementality!" This usually means your ads are targeting people who were already going to buy (e.g., heavy remarketing to existing customers). Try testing top-of-funnel 'Prospecting' audiences instead. "The platform says 'Not enough data'." You likely need more conversions. If your budget is small, try testing at the 'Add to Cart' level instead of 'Purchase' to get a higher volume of data points. "The Control group is outperforming the Test group." This is rare but happens. It suggests your ads might actually be annoying your customers or driving them to competitors. Review your ad creative—is it too aggressive or irrelevant to the Australian audience?---
Next Steps
Once you’ve mastered incrementality for one channel, the next step is Marketing Mix Modelling (MMM) to see how all your channels work together.
Need help setting up a scientific test for your Brisbane business? Our team at Local Marketing Group can help you audit your current spend and find the 'waste'. Contact us today to get started.