On this page
Value-based bidding is a campaign strategy where you tell the ad platform how much each conversion is worth, so it can prioritize users likely to generate high revenue rather than just any user likely to convert at all.
The core idea is simple: not all customers are equal. A user who pays $99 for an annual subscription is worth more than one who downloads your app for free and churns after three days. Target CPA bidding treats both as identical conversion events. Value-based bidding signals to the platform's machine learning system which type of user it should be chasing.
Why Standard Conversion Optimization Leaves Revenue on the Table
When you run a campaign on target CPA, the platform's algorithm learns to find people likely to take any action you define as a conversion. If your conversion event is "purchase," it optimizes for purchase probability. It does not care whether that purchase is worth $5 or $500.
This creates a systematic problem for businesses with high LTV variance. If your average customer generates $40 in lifetime value but your top 20% generate $200+, a CPA-optimized campaign will happily fill your acquisition funnel with the $40 customers. The algorithm is doing exactly what you told it to do.
In practice, teams often try to compensate by setting separate campaigns with different CPA targets for different user segments. That approach fragments your data, slows the learning phase for each campaign, and requires constant manual management.
Value-based bidding solves this at the signal level. Instead of targeting a specific cost, you give the platform a revenue figure for each conversion and let the algorithm learn what predicts high-revenue customers across all your campaigns simultaneously.
How Value-Based Bidding Works on Google and Meta
The implementation differs by platform, but the underlying logic is the same: attach a monetary value to your conversion events, and the platform's AI factors that value into its real-time bidding decisions.
Google UAC and Performance Max with Target ROAS. On Google, value-based bidding uses the Target ROAS bid strategy paired with value-enriched conversion tracking. You configure your conversion action to include a value parameter, and Google's algorithm bids to hit your stated ROAS target across all auctions. The system considers hundreds of real-time signals for each impression: the user's query context, device, location, time of day, historical behavior, and audience memberships. It adjusts bids up for users who look like your high-value customers and down for users who look like your low-value ones.
Meta Value Optimization. Meta surfaces this as "Value Optimization" bidding at the ad set level. Instead of optimizing for purchase count, you tell Meta to optimize for purchase value. Meta then builds a model of your high-revenue customers based on the value signals you pass through its Conversions API or SDK, and targets users whose profiles most closely match. You can set a minimum ROAS floor to control spend efficiency.
BeFreed, a language learning app, switched from target CPA to value-based bidding on Google UAC after identifying that LTV varied by a 4x factor between their lowest and highest segments. Over 60 days, that change dropped their blended CAC from $24 to $15 while holding ROAS steady at 3.2x. The campaign was acquiring fewer installs total, but far more high-value users.
What Signals to Pass and How to Pass Them
The quality of your value signals determines whether value-based bidding outperforms standard conversion bidding. A few practical rules:
Use actual revenue, not predicted LTV, when you have it. If your conversion event is a purchase, pass the purchase amount. Do not pass a fixed proxy value like your average revenue per user. Variable values give the algorithm meaningful differentiation. Fixed values make value bidding functionally identical to CPA bidding.
For subscription products without an upfront payment, use 90-day LTV estimates. You will need to build a model that predicts 90-day revenue at the time of the trial start or first engagement event, then pass that predicted value. This is more complex but captures the real economic signal. Teams using Appsflyer's predictive analytics module or Adjust's LTV prediction can pass these values directly through the SDK.
For app campaigns, use Appsflyer or Adjust to relay values. Both MMPs support passing revenue values from in-app purchase events back to Google UAC and Meta. In Appsflyer, you configure revenue events under "In-App Events" and map the af_revenue parameter. Adjust surfaces similar configuration through its Event Token settings. Without MMP integration, you're relying on the platform's own SKAdNetwork signals, which are delayed and aggregated.
Pass $0 for non-revenue events. If you want to capture trial starts in your conversion funnel but don't want the algorithm to optimize for them at full value, pass a $0 value. The event still registers for attribution, but the platform deprioritizes it in value optimization relative to paid events.
When Value-Based Bidding Outperforms Target CPA
Value-based bidding generates the most lift when two conditions exist: high LTV variance in your customer base, and sufficient conversion volume for the algorithm to learn.
High LTV variance means your customers differ meaningfully in how much they spend. Subscription products with annual tiers, games with heavy-spender segments, and B2B SaaS with seat-based pricing all have this characteristic. If 80% of your customers generate roughly the same revenue, there is less for the algorithm to optimize.
Sufficient conversion volume means the platform has enough data to build a reliable model. Google's guidance to reach 50+ conversions per month at the campaign level before switching to target ROAS is not arbitrary. Below that threshold, the algorithm's value model is based on too few data points and will overfit to noise.
In practice, teams often introduce value-based bidding at the account level using portfolio bid strategies rather than at the campaign level, which pools signals across campaigns and helps smaller-volume campaigns benefit from the combined data.
The Transparency Tradeoff
Value-based bidding is a black-box optimization. You observe the outcomes but not the mechanism. The algorithm does not explain why it targeted a particular user segment, which creatives it weighted higher for high-value users, or which audience signals it found most predictive.
This is a real limitation for teams that need to understand their customer acquisition, not just its cost. You can see that high-value users came from certain placements or time windows in your reporting, but you cannot extract the model's decision logic.
Hellyeah's Managed Growth layer addresses this by running performance attribution across value tiers simultaneously with the platform's optimization. Instead of relying on the platform's reporting alone, Hellyeah tracks which creative treatments, audience seeds, and campaign configurations correlate with high-LTV user acquisition across your historical data. That signal supplements what the platform's algorithm is doing with an independent view of what's working.
Common Implementation Mistakes
Setting a ROAS target too aggressively at launch. If your historical target CPA campaign delivered a 2x ROAS and you immediately switch to a 4x ROAS target with value bidding, the algorithm will under-serve because it cannot find enough inventory to hit that threshold. Start 10-15% above your current achieved ROAS and tighten over time.
Inconsistent value definitions. If you pass purchase price for some events and predicted LTV for others, the algorithm's model becomes unreliable. Pick one value definition and apply it uniformly across all campaigns using value optimization.
Resetting the learning phase too frequently. Every time you make a significant change to a campaign running value-based bidding, the algorithm enters a learning period of 1-2 weeks where performance fluctuates. Changing bid targets, audience configurations, or conversion definitions restarts this cycle. Plan changes carefully and avoid stacking multiple changes in the same window.
Ignoring the relationship between value-based bidding and attribution. The values you pass to the ad platform are only as accurate as your attribution model. If your MMP is over-attributing conversions to last-touch paid channels, your value signals are inflated and the platform's model will chase users who look like your attributed high-value customers, not your actual ones.
Related Concepts
Value-based bidding sits at the intersection of bid strategy and ROAS optimization. It is most effective when paired with a reliable LTV model, accurate attribution setup, and sufficient conversion volume. Teams moving from CPA marketing to value-based approaches should plan a 4-6 week transition period to allow the algorithm's learning phase to complete before evaluating results.
Related Guides
Frequently asked questions
How is value-based bidding different from target CPA bidding?
Target CPA treats all conversions as equal, optimizing to hit a specific cost per any conversion event. Value-based bidding tells the platform that some conversions are worth more than others by passing a revenue value with each conversion. The platform's AI then prioritizes impressions likely to generate high-value conversions, not just any conversion. The result is often a higher average LTV from paid users at the same or better overall CAC.
What conversion values should I pass for value-based bidding?
Pass the actual revenue value of the conversion: purchase price, subscription monthly fee, or predicted LTV at a fixed window such as 30-day or 90-day. For apps with free trials, pass $0 for trial starts and the actual subscription value when a user converts to paid. The key is consistency. The value you pass must correlate with what you actually want the platform to optimize for.
How much conversion volume do I need to run value-based bidding?
Google recommends at least 50 conversions per month at the campaign level with relatively stable values before switching to value-based bidding. Meta's Value Optimization requires sufficient volume to exit the learning phase, typically 50+ optimization events per week per ad set. Below these thresholds, the algorithm does not have enough signal to distinguish high-value from low-value users reliably.
Does value-based bidding work for mobile apps?
Yes, but it requires passing a value parameter through your mobile measurement partner. Appsflyer and Adjust both support revenue values attached to in-app events. You configure the event to include the purchase amount, and that value flows into Google UAC or Meta's Value Optimization via the SDK integration. Without that value field populated, the platform cannot distinguish a $2.99 purchase from a $49.99 subscription.
Related Terms
Smart Bidding: Google's Automated Bid Strategies and When to Use Them
Smart Bidding is Google's suite of machine-learning bid strategies. Learn how Target CPA, Target ROAS, Maximize Conversions, and Maximize Conversion Value differ and when each one fits.
Bid Strategy: How to Choose the Right Approach for Every Campaign
Bid strategy determines how you pay for ad placement. Learn the difference between manual, automated, and hybrid approaches and how to pick the right strategy for your campaign's goals and data volume.
ROAS: What It Means and How to Improve It
ROAS (Return on Ad Spend) measures how much revenue you earn for every dollar spent on advertising. Learn how to calculate it, what benchmarks actually mean, and how to improve it.
Customer Acquisition Cost (CAC): Definition and Benchmarks
CAC is the total cost to acquire one paying customer, covering ad spend, team, and tools. Learn how to calculate it accurately, what drives it up, and how to bring it down.

Co-founder
Co-founder of Hellyeah. Writes about building durable growth loops that compound over time.

