Google Optimize Is Gone. Now What?
When Google killed Optimize in September 2023, agencies scrambled. VWO starts at $200/month. Optimizely is enterprise-priced. Most small businesses just stopped testing.
But WordPress 6.5 introduced something that makes third-party A/B testing tools unnecessary for most use cases: the Block Bindings API. Combined with a simple variant selection system, you can A/B test from block markup with zero external dependencies.
How It Works
A standard heading block has static text. With the voyager/ab-test binding, that same heading randomly selects from variants you define — and the selection is sticky per visitor via a 30-day cookie.
The markup is simple. In the block’s metadata, you set the source to voyager/ab-test and pass your variants as pipe-separated values: "Get Started Today|Start Your Growth Journey|Let's Build Together". Each visitor gets one variant, deterministically assigned by a hash of their cookie and the test ID.
No JavaScript swap after page load. No flicker. The correct variant is in the server-rendered HTML before the browser even starts painting.
Setting Up Your First Test
Here is the actual block markup for a heading A/B test:
<!-- wp:heading {"metadata":{"bindings":{
"content":{
"source":"voyager/ab-test",
"args":{
"variants":"Get Started Today|Start Your Journey|Let's Build Together",
"test_id":"hero-headline-q2"
}
}
}}} -->
<h2>Get Started Today</h2>
<!-- /wp:heading -->
The test_id ensures the same visitor always sees the same variant for this specific test. Different test IDs on the same page can run independently — test the headline and the CTA button simultaneously.
You can bind any text attribute: heading content, paragraph text, button labels. You can even bind the button URL to test different landing pages.
What to Test First
Not everything is worth testing. Focus on high-traffic, high-stakes elements:
- Homepage hero headline — highest traffic page, biggest impact on first impression
- CTA button text — “Get Started” vs “Book a Call” vs “See Pricing” directly affects click-through
- Service page value props — which benefit resonates most with your audience
- Pricing page CTA — the closest touchpoint to conversion
Do not test body copy, footer text, or low-traffic pages. The sample size will never be large enough to reach significance.
Why This Is Better Than Most Testing Tools
- No flicker. Server-side rendering means the correct variant is in the HTML. No JavaScript swap after load.
- No script overhead. Zero KB of testing JavaScript added to your pages. Better Core Web Vitals.
- No monthly fee. The binding source is part of the WordPress plugin. Test forever for $0.
- Sticky assignment. Same visitor always sees the same variant for the same test. No inconsistency.
- Works with any block. Headings, paragraphs, buttons, images — any bindable attribute.
- No consent banner trigger. The cookie is functional (variant assignment), not tracking. No GDPR issues.
Measuring Results
Voyager Orbit tracks every lead with full attribution. When a visitor who saw Variant B fills out a contact form, Orbit records which variant they were assigned. Compare conversion rates in the Portal dashboard.
This is not as sophisticated as a full multivariate testing platform with statistical significance calculators. But for 90% of agencies testing headlines, CTA text, and hero copy, it is more than enough — and it costs nothing. See it working live on our block bindings showcase.
Key Takeaways
- A/B testing does not require a $200/mo tool. Block bindings handle the variant selection natively.
- Server-side rendering eliminates flicker — the #1 complaint about client-side testing tools.
- Start with 1-2 tests on high-traffic pages. Headlines and CTA buttons first.
- Combine with Orbit lead tracking for conversion attribution without additional tools.
- The test_id system lets you run multiple independent tests on the same page.
Leave a Reply