If you’re running A/B tests, multivariate tests, or personalizations, targeting the right audience is half the battle. This guide is for digital marketers, product folks, and anyone trying to get real insights—not just pretty reports—from their experiments. You’ll learn how to actually use Google Optimize’s audience targeting features without wasting time on stuff that doesn’t move the needle.
Let’s cut through the fluff and get into what really works.
Why Audience Targeting Matters (and When It Doesn’t)
Before you jump into segments and conditions, remember: more targeting isn’t always better. Testing on “all users” is fine if you’re validating a site-wide change. But if you’re making changes for new users, mobile visitors, or a certain country, targeting is essential.
But—don’t get clever for the sake of it. Over-targeting means fewer people see your test, results take forever, and your insights are shakier. Only get specific with your audience when you have a real reason.
Step 1: Get Clear on Who You’re Targeting (and Why)
Don’t start with the tool—start with your question. Ask:
- Who do I actually want to see this experiment?
- Is this change for everyone, or just a slice (e.g., mobile users, logged-in customers, folks from a certain ad campaign)?
- Do I have enough traffic in that audience to get a clear result in a reasonable time?
Pro tip: If your audience is too niche, you’ll wait weeks (or months!) for enough data. Sometimes it’s better to test on a broader group, then segment your results later.
Step 2: Know Your Targeting Options in Google Optimize
Google Optimize gives you several ways to pick your audience. Here’s what’s actually useful—and what you can skip.
The Most Reliable Targeting Methods
- URL targeting
For changes on specific pages or sections. Easy, fast, and dependable. - Device category
Mobile vs. desktop. Good for mobile-first designs or troubleshooting. - Geography
Country, region, or city. Useful but beware of VPNs or misreported locations. - Behavioral targeting (new vs returning users)
Great for onboarding tweaks or retention experiments.
Advanced, But Worth It (Sometimes)
- First-party cookies & JavaScript variables
For logged-in states, shopping cart status, user roles, etc. Powerful but needs dev help. - Google Analytics Audiences
Build segments in GA, then use them in Optimize. More flexible, but there can be a delay syncing audiences (sometimes minutes, sometimes hours).
What to Ignore (Mostly)
- Query parameters for deep targeting
Works, but can get messy fast. Only use when you control the parameters. - Referrer targeting
Unreliable. Many browsers and privacy plugins block or modify referrers.
Step 3: Set Up Your Audience in Google Optimize (The Smart Way)
Let’s walk through creating a targeted experiment without shooting yourself in the foot.
1. Start Broad, Then Narrow Down
- Don’t pile on conditions out of the gate. Start with the minimum targeting you need.
- Example: Want to test a new homepage banner for mobile users in Canada? Use “device = mobile” and “country = Canada.” Don’t add more unless you have a good reason.
2. Use Preview and Debug Tools
- Always preview your experiment as the targeted audience before launching.
- Use Chrome DevTools, VPN, or device simulators to test geo and device targeting.
- For cookie or JavaScript-based segments, check your targeting logic in the browser console.
3. Watch for Sample Size Problems
- The more you target, the smaller your test audience.
- Do the math: If you get 10,000 visitors a week, but only 5% are in your target group, that’s just 500 a week. If your test needs 5,000 per variant, you’ll be waiting a while.
4. Keep Track of Overlapping Experiments
- If you run multiple tests on different audiences, make sure they don’t overlap unless you’re OK with cross-contamination.
- Google Optimize will warn you about some conflicts, but you’re the final line of defense.
Step 4: Use Google Analytics Audiences (But Don’t Rely on Them Alone)
You can import Google Analytics audiences into Optimize. It’s handy for:
- Targeting people who hit a certain funnel stage
- Segmenting by traffic source (e.g., only Facebook ad visitors)
- Reaching “high-value” users based on past behavior
But:
- Audience syncing isn’t instant. Some users won’t get bucketed right away.
- If your audience is too small or complex, you’ll wait forever for results.
- There’s a lag. Don’t use this for “real-time” reactions (like cart-abandonment popups).
Pro tip: Use GA audiences for bigger, persistent groups (e.g., all users who bought in the last 30 days), not for micro-moments.
Step 5: Avoid These Common Pitfalls
Even smart teams make these mistakes:
- Over-segmenting.
More conditions slow down experiments and can give you dodgy stats. - Not testing targeting logic.
If your audience definition is broken, your whole experiment is toast. - Ignoring privacy and consent.
Some targeting uses cookies or personal info. Make sure your use is above board—especially in the EU. - Blindly trusting Optimize’s audience estimates.
Always check your actual numbers in GA or another analytics tool.
Step 6: Analyze Results by Segment—But Don’t Cherry-Pick
After your experiment runs, it’s tempting to slice and dice the results. That’s fine, but:
- Stick to your initial hypothesis. If you said you’d test for new users, don’t claim a win for “users from Toronto on Samsung phones.”
- Beware false positives. The more segments you check, the more likely you’ll find a “winner” by accident.
- Use Analytics for deeper dives. Google Optimize’s built-in reporting is basic; use Google Analytics for serious segmentation.
Step 7: Iterate—Don’t Overthink It
You won’t get perfect targeting on your first try. That’s fine. The goal is to get clear, actionable insights—not to build the most complicated targeting setup.
If your first experiment is too narrow and takes forever, go broader next time. If your results are muddy, tighten your audience. It’s a cycle.
Real-World Tips from the Trenches
- Keep your targeting logic documented. If you leave, someone else should understand what you did and why.
- Limit active experiments per audience. Overlapping tests can screw up your data and make it impossible to know what worked.
- Don’t obsess over tiny segments. Unless you have millions of users, testing for “returning users on iPhone 12 in Belgium who came from LinkedIn” just isn’t worth it.
- QA before you launch. Targeting bugs are hard to spot until too late—always double-check.
Bottom Line: Make Targeting Work for You, Not Against You
Audience targeting in Google Optimize is powerful, but it’s easy to overcomplicate things. Pick the smallest set of targeting rules that match your real-world business goal. Make sure your experiment runs fast enough to get results. If in doubt, start broad, measure, and adjust.
Don’t let “fancy” targeting slow you down. The best experiment is the one you actually finish—and learn from. Keep it simple, keep it honest, and keep iterating.