Meta’s latest update introduces a pivotal shift. Advertisers can now run Value Optimization (VO) campaigns for non-purchase events.

This might sound like a small change but, for many performance teams, it’s a long-awaited unlock. Until now, VO was limited to purchases, which worked well for e-commerce but left a major gap for brands with different growth models.

If your most valuable customers don’t convert right away, if your business depends on long-term value to recoup CAC, or if you optimize toward upper-funnel actions like installs or signups – this update is for you.

Now, you can tell Meta’s system what really matters earlier in the journey. You’re no longer forced to choose between optimizing for an easy-to-measure proxy or waiting weeks, or months, to optimize for revenue.

This is more than just a new campaign setting. It’s a meaningful step toward aligning ad spend with real business value at the moment it matters most.

What Changed and Why It Matters

Historically, Meta’s VO campaigns were restricted to purchase events. That made them a strong fit for direct-to-consumer brands with short sales cycles. But for everyone else, it created a misalignment between optimization targets and actual business outcomes.

For example:

• In subscription or B2B models, high-LTV customers often take longer to convert.
• Many brands need to recoup marketing spend through retention and expansion, not a single purchase.
• If you optimize for a proxy like Signup or Install, Meta treats all events as equal – even when some users are far more valuable than others.

Now, Meta has opened the door for you to optimize toward any meaningful event – not just purchases. That includes:

• Sign-ups
• App installs
• Trial completions
• Or any event you define 

“In an era of AI-enabled advertising where the best performing campaigns are those that have flexibility to find the most relevant customers, our goal is to help advertisers guide our ads system to optimize for the outcomes that are most important to their business.”
— Meta Business

That means you can tell Meta, “This earlier action is worth something, and we want to optimize for it.” 

But here’s what makes this update different. Previously, with binary optimization, you could ask Meta to optimize for an early event like a signup or install, but it treated every conversion the same. All signups looked equal, even if some users were far more valuable than others.

With VO for non-purchase events, you can now signal not just that an event occurred, but how valuable that user is likely to be. That allows Meta’s algorithm to bid more aggressively for high-value users and more conservatively for low-value ones. You can now bring greater efficiency and strategic depth to your upper-funnel efforts.

Your New Responsibility: Defining Your Signals 

While this change gives you greater influence over how Meta’s AI targets users and values different actions, it also places a new responsibility on you to define and deliver the right signals. To make VO campaigns work, Meta needs two critical inputs from you:

1. The event itself: What action are you asking Meta to optimize for?
2. A value for that event: How important is that action to your business?

Meta won’t supply that value. You have to define it, and the more predictive it is of long-term outcomes, the better Meta will optimize towards your goals.

Under the Hood – What’s Technically Possible 

With this update, you can get more hands-on with how your campaigns learn and improve. Here’s how:

1. Test different early-stage events: With this update you can now run VO vs. VO tests, allowing you to optimize two campaigns for different early-stage events (such as app install vs. onboarding complete) and see which yields stronger long-term results.
2. Compare value vs. cost optimization: Run campaigns using the same event, but optimize one for value (VO) and one for cost (tCPA) to understand where you gain efficiency.
3. Clean up legacy VO workarounds: In the past, you may have sent non-purchase actions (like signups or installs) to Meta’s Purchase event just to unlock VO. This creates messy learning behavior and reduced performance. Now that VO is supported on non-purchase events, you can assign the right signal to the right event, resulting in cleaner optimization and better results.
   
   1. Some newer Meta features (such as ASC campaigns) first roll out only to the purchase event (much like VO was only Purchase for a very long time). That means that you might still want to maintain your standard Purchase event as an actual purchase only, and allowing vo on different signals then allows you to not send non-purchase actions.
4. You always have the legacy option, of setting up more pixels, which will still be beneficial if you don’t have enough traffic to test 100 conversions a week on the non-purchase event.

However there are technical considerations to keep in mind:

• Mobile vs desktop behavior diverges: VO works reliably on desktop and mobile web, but native mobile apps require more deliberate setup. You’ll need proper SDK or MMP (mobile measurement partner) integration to pass the event and its value.
• Signal quality is key: Feeding Meta accurate, high-quality values tied to meaningful events is what makes VO work. That’s where predictive modeling comes into play. If your signals are noisy or misaligned with downstream value, you risk steering Meta in the wrong direction, which can tank performance. 
• Non-purchase VO needs more volume to learn: When you optimize for anything other than the Purchase event, Meta typically needs more data to train effectively – roughly 100 conversions per week instead of the usual 30. If you have enough traffic, that’s not a blocker. But it does mean your tests may require a larger budget or longer ramp to deliver clear, stable results.

Putting This Into Practice: Strategic Advice for Growth Leaders

This update isn’t just a tactical feature change – it’s a shift in how performance teams can think about value. You’re no longer confined to last-touch conversions or to a single VO strategy without a clean way to test what’s working. Instead, you can define the moments that matter earlier in your funnel and build campaigns around them.

To successfully respond to this shift, you and your team should: 

• Re-evaluate your most predictive early signals: Which actions signal long-term value? While it’s tempting to focus on a single event like Signup or Install, that alone may not tell the full story. The most effective strategies rely on predictive signals – ideally powered by ML models trained to identify which users are likely to mature into high-value customers.
• Intentionally test different VO strategies: With support for non-purchase VO, you can now run side-by-side campaigns optimized for different early signals and finally get clean, actionable readouts on what drives the best downstream outcomes.
• Make sure your event values reflect true business impact: Whether you’re using modeled predictions or actual observed values, Meta’s algorithm needs clear, high-fidelity signals. The more your inputs align with long-term growth, the more the system can work in your favor.
• Stay aware of platform limitations across web and mobile: What works on desktop or mobile web may require extra setup for native mobile apps. Make sure you’ve accounted for SDKs, MMPs, and evolving platform support before launching new campaigns.

Bridging the Gap: The Role of Predictive Signal Engineering

To truly leverage VO campaigns for non-purchase events in Meta, advertisers face the crucial task of defining and delivering high-quality, predictive value signals. As we covered earlier, you have to define that signal for Meta, and the more predictive it is of long-term outcomes, the better Meta will optimize towards your goals.

This is where predictive modeling and signal engineering enter your online media buying strategy. It's about turning your raw behavioral data into actionable, predictive value signals that Meta's AI can understand and optimize against. 

Voyantis helps advertisers bridge that gap. We specialize in signal engineering, transforming your rich, untapped data into predictive value signals that are AI-ready and speak Meta’s language.

We help you:

• Identify the early actions that correlate with customer LTV
• Build predictive models that assign value in real time
• Automatically feed those values to Meta with accuracy and confidence

Check out some of our success stories.

Signal engineering with Voyantis means you’re not guessing which signals to send or hoping the right users get targeted. Together, we’re guiding Meta’s AI with high-fidelity predictive signals that reflect your actual growth strategy. 

We’re building the predictive and prescriptive AI infrastructure that forward-thinking marketers need to gain the most from shifts like these from Meta and Google. If you're unsure where to begin, a conversation with our team can help you identify which signals to test, how to structure them, and what values to assign. Connect with our team today – as more platforms adopt value-based optimization, the advertisers who win will be those who master predictive signal engineering now. 

Book a strategy call with our experts