In the Facebook advertising ecosystem, Value Optimization (VO) is often discussed as the ultimate tool for scaling ROAS. For performance advertisers chasing high returns, VO sounds like a shortcut to finding high-spending users. But what is the underlying logic of VO? Why is it locked for some accounts? And why do In-App Advertising (IAA) developers have a love-hate relationship with it?

The Core Logic of VO: From "Who Will Buy" to "Who Will Spend More"

Traditional conversion optimization (such as standard purchase conversions or app install optimization) focuses on conversion probability—identifying who is likely to make a purchase. VO, however, shifts the focus to transaction value—identifying who will spend the most.

Meta's machine learning models predict user value. When a user enters the auction pool, the system analyzes historical purchasing power, device model (e.g., the latest iPhone), and interest signals (e.g., finance, high-end gaming) to estimate predicted ROAS or purchase value.

Under VO, the system prioritizes bidding for high-predicted-value users. If User A is predicted to spend $50 and User B is predicted to spend $5, the system will aggressively bid higher to win User A, even if their conversion probabilities are identical.

The Hard Thresholds to Unlock VO

VO is highly data-dependent. You cannot simply turn it on without meeting strict technical and data requirements.

1. Technical Infrastructure

  • Web: Meta Pixel or Conversions API (CAPI) must be fully configured with value passback.
  • App: Facebook SDK or a Mobile Measurement Partner (MMP) like AppsFlyer or Adjust must be integrated.

2. Data Diversity Requirements

The algorithm needs to see a range of purchase values to distinguish high-value from low-value users. If your store only sells a single $9.99 product, VO is useless.

  • E-commerce & In-App Purchases (IAP): At least 30 purchases with at least two distinct purchase values in the last 14 days (Meta officially recommends 100+ for stability).
  • In-App Advertising (IAA): At least 15 ad_impression events with varying values in the last 28 days.

The Pitfalls of VO for IAA: The High-Churn Reality

To compete with Google's App Campaigns (UAC 3.0) for ad-monetized apps, Meta rolled out VO for IAA. However, performance marketers often refer to it as a flawed solution.

Why? Google's algorithm can optimize based on longer-term value windows (e.g., 7-day or 30-day LTV). Meta's VO, on the other hand, heavily favors immediate, short-term value.

The Retention Cliff: VO targets users who are highly sensitive to ads. They watch a lot of ads, which drives up immediate eCPM, but they also have extremely low tolerance for apps and uninstall quickly.

This results in rapid Day 1 payback but disastrous Day 1 Retention (D1R). While this model works well for short-cycle monetization products (like utility cleaners, simple tools, or certain health offers), it can severely damage apps that rely on long-term retention and community engagement (like social apps).

How to Maximize the Value of VO

If you have the data and resources to test VO, use these strategic guidelines:

1. Volume First, Value Second

Do not launch a new ad account directly with VO. Start with standard conversion optimization to accumulate data. Once the model stabilizes and has a diverse distribution of purchase values, transition to VO to scale ROAS.

2. Target High-Net-Worth Signals

  • Targeting: Pair VO with a 3% Lookalike Audience (LAL) or high-net-worth interest layers (e.g., personal finance, high-end devices).
  • Creatives: Avoid low-quality, spammy creatives. High-value users expect polished aesthetics. For Tier 1 markets like the US and Europe, premium UI designs and clean, high-end lifestyle backgrounds in your creatives will naturally filter out low-value audiences.

3. Use ROAS Goals with Caution

Setting a strict minimum ROAS goal right away can choke your delivery and cause traffic to plummet. It is safer to start with the "Highest Value" bidding strategy without a cap, monitor CPI and LTV, and introduce a ROAS goal only after performance stabilizes.

Conclusion

Meta's VO is essentially a regression prediction tool that often trades long-term retention for immediate payback. In Meta's massive distribution engine, VO is a highly sharp tool—it doesn't judge quality, only value. Understanding its limitations is key to unlocking its true scaling power.