Smart Bidding is now central to Google Ads performance strategy, especially for campaigns with reliable conversion tracking. New campaigns without enough conversion data can often start with Maximize Conversions, then move to Target CPA or Target ROAS once performance volume builds. The most common Smart Bidding failure is setting targets too aggressively, which limits the learning phase and restricts delivery. Clean conversion data, realistic targets, audience signals, and seasonality adjustments are the levers that separate well-performing automated campaigns from poorly performing ones.
Google Ads has shifted decisively toward automation. Manual CPC — once the default for hands-on advertisers — has been removed for new Search campaigns in many account types, and Google has continued to reduce its availability across the platform. For most advertisers, Smart Bidding is now the primary operating model.
The question is no longer whether to use Smart Bidding, but how to use it well. This guide covers how each strategy works, the conditions where each performs best, and the most common mistakes that cause Smart Bidding campaigns to underperform.
Smart Bidding Strategies Explained
Google offers four main Smart Bidding strategies for Search and Shopping campaigns. Each optimizes for a different objective, and choosing the wrong strategy for your campaign’s stage or goal is a common source of wasted spend.
| Strategy | Optimizes For | Best Used When | Minimum Data Threshold |
|---|---|---|---|
| Maximize Conversions | Most conversions within budget | New campaigns without CPA target; campaigns scaling volume | No strict minimum, but performs better with some conversion history |
| Target CPA | Conversions at or near a target cost per acquisition | Campaigns with stable conversion history and a defined acceptable CPA | Google recommends 30–50 conversions in the past 30 days as a starting point |
| Target ROAS | Revenue relative to ad spend at a target ratio | E-commerce campaigns with reliable conversion value tracking | Google recommends 50+ conversions in the past 30 days for stable performance |
| Maximize Conversion Value | Highest total conversion value within budget | Campaigns where revenue maximization matters more than efficiency ratio | Works with limited data but improves significantly with volume |
The conversion thresholds listed above reflect Google’s general guidance as a starting point for stable optimisation. Actual requirements vary by campaign type, bidding strategy, and account characteristics — they are recommendations rather than hard minimums. (Google Ads Help: About Smart Bidding)
Enhanced CPC (eCPC) was a transitional Smart Bidding option that applied automated adjustments to manual bids. Google has been deprecating eCPC for most campaign types and it is no longer the recommended starting point for new campaigns.
The Learning Phase: What It Is and How to Manage It
When a Smart Bidding strategy is first applied, or when significant changes are made to a campaign, Google’s system enters a learning phase. During this period, the algorithm evaluates how different bid levels perform for your specific account and conversion goal.
Learning phase duration varies. Google indicates it typically lasts around one week, but this depends on conversion volume. Campaigns with low conversion frequency can remain in learning phase longer, and campaigns with very low volume may struggle to exit it at all.
The most important rule during the learning phase: avoid making significant changes. Significant changes to targets, budgets, audiences, or ad structure may reset or extend the learning phase. Each reset costs time and may allow the algorithm to serve in less efficient ways while it relearns. (Google Ads Help: About the learning period for bidding strategies)
Signs the Learning Phase Is Stuck
If a campaign remains in learning phase for more than 2–3 weeks, look at:
- Conversion volume: If fewer than 10–15 conversions occurred during the learning window, the algorithm has insufficient data. Consider temporarily switching to Maximize Conversions without a CPA target to build volume, then add a target later.
- Target aggressiveness: A CPA target set far below historical performance will limit delivery as the system struggles to find conversions at that price. Start with a target 20–30% above your current average CPA and lower gradually.
- Budget constraints: If the campaign is frequently budget-limited, the algorithm cannot explore bid ranges effectively. The appropriate adjustment depends on whether the campaign is budget-limited or target-limited — a budget-limited campaign may need more spend headroom, while a target-limited campaign may need a less aggressive CPA target before budget utilisation improves.
Target CPA: How to Set Targets That Actually Work
Target CPA is the most commonly misconfigured Smart Bidding strategy. The algorithm needs realistic targets to function: if you set a target significantly below what your historical performance supports, delivery drops, the campaign struggles to spend budget, and the apparent “efficiency” is an artifact of reduced volume rather than better performance.
A practical approach for setting a Target CPA on an existing campaign:
- Pull your average CPA for the past 30–60 days from Search campaigns with similar conversion actions.
- Set your initial target CPA at or slightly above (10–20%) this historical average.
- After 2–3 weeks of stable performance, begin reducing the target CPA in 10–15% increments, waiting at least two weeks between each adjustment.
- Monitor conversion volume alongside CPA. A lower CPA achieved by drastically reducing conversion volume is not a win — you are paying less per conversion but getting far fewer of them.
Target ROAS: When It Works and When It Does Not
Target ROAS tells Google to optimize for conversion value rather than conversion count at a specific return ratio. It works well for e-commerce campaigns where products have meaningfully different values and the algorithm can prioritize higher-value orders.
Target ROAS performs less well when:
- Conversion values are uniform (all conversions assigned the same value), in which case Target ROAS behaves identically to Target CPA.
- Conversion volume is low (fewer than 50 per month). Low volume means high variance in the algorithm’s estimates, which can cause large swings in delivery and efficiency.
- The ROAS target is set too high relative to historical performance, which causes the algorithm to become overly selective and drive down volume without improving actual efficiency.
If your e-commerce campaigns pass conversion value through to Google Ads accurately, Target ROAS is worth testing for your highest-volume campaigns. For campaigns with mixed conversion types (purchases and lead forms, for example), model-level value assignments may be needed to make Target ROAS meaningful.
Portfolio Bidding Strategies
Portfolio bidding allows you to apply a shared Smart Bidding strategy across multiple campaigns. This is useful when individual campaigns have too few conversions to support their own learning phase but aggregate conversion volume across campaigns is sufficient.
Portfolio bidding with a shared Target CPA or Target ROAS pools learning across campaigns and can reduce time-in-learning-phase for lower-volume campaigns. It is particularly useful for accounts with multiple campaigns targeting similar products or audiences at different funnel stages.
Campaigns sharing a portfolio strategy should have broadly similar business objectives and conversion behaviour — pooling unrelated campaign types can produce unstable optimisation as the algorithm tries to reconcile conflicting signals. Monitor individual campaign performance regularly even when using portfolio strategies, and consider separating campaigns with significantly different conversion rates or values into separate portfolios.
Audience Signals and First-Party Data
Smart Bidding algorithms can use audience signals to adjust bids for users more or less likely to convert. Adding relevant audiences — Customer Match lists, website visitors, similar audiences — to campaigns in observation mode gives the algorithm additional inputs without restricting delivery to those audiences.
As third-party cookie availability decreases and browser privacy changes continue, first-party data is increasingly important for Smart Bidding quality. Accounts that upload Customer Match lists with email addresses from existing customers give the algorithm a better signal about what converting users look like.
Customer Match requires a Google Ads account in good standing and meeting minimum user volume thresholds for the lists to be active in campaigns. For details on managing audience lists alongside your bidding strategy, see our guide on Google Ads tactics that work in 2026.
Seasonality Adjustments
Smart Bidding algorithms use historical data to predict future conversion rates. During periods where conversion rates change suddenly — a flash sale, a product launch, a seasonal event — the algorithm may lag the actual conversion rate shift and bid suboptimally.
Google provides seasonality adjustments for Search, Shopping, and Performance Max campaigns using eligible Smart Bidding strategies. They let advertisers tell the Smart Bidding system to expect a temporary change in conversion rate. (Google Ads Help: About seasonality adjustments) For example, if you expect a 40% increase in conversion rate during a two-day sale, you can set a seasonality adjustment so the algorithm bids more aggressively during that window without treating the spike as a permanent trend.
Seasonality adjustments are most useful for:
- Short promotional events (2–7 days) where conversion rates are expected to spike above historical averages.
- Planned traffic periods (product launches, media appearances) where volume will change significantly from normal patterns.
Do not use seasonality adjustments for extended periods. They are designed for short windows. For longer structural changes to campaign strategy or conversion expectations, adjust your CPA or ROAS targets directly.
Performance Max and Smart Bidding
Performance Max campaigns run on Smart Bidding by default — you set a target CPA or ROAS, and the algorithm allocates budget across Google’s inventory including Search, Display, YouTube, Gmail, Maps, and Discover.
The same Smart Bidding principles apply to PMax campaigns: realistic targets, sufficient conversion volume, and clean conversion tracking are the foundation. PMax has additional complexity because budget allocation across channels is less transparent. Reviewing asset group performance and using campaign-level negative keywords (now available in most accounts; availability may differ depending on account configuration and Google Ads rollout) are important for managing PMax alongside Smart Bidding goals.
For a detailed breakdown of what to keep, cut, and adjust in Google Ads campaigns in 2026, see our Google Ads tactics guide.
Tracking What Smart Bidding Is Actually Optimizing For
Smart Bidding optimizes for whatever conversion action is set as the primary conversion goal in your campaign. If that conversion action does not represent real business value, the algorithm will optimize for the wrong thing efficiently.
Audit your conversion actions before trusting Smart Bidding results:
- Primary conversion actions should represent actual business outcomes: purchases, form submissions, calls with duration thresholds, appointments booked.
- Secondary conversion actions (page views, scroll depth, video views) should be observation-only and not drive bid optimization.
- Conversion attribution windows should match your sales cycle. A short attribution window for a long-consideration product undervalues conversions and causes the algorithm to underbid.
Understanding Google Ads Quality Score alongside Smart Bidding is also worth reviewing — Quality Score affects which auctions you compete in, which in turn affects the conversion data the Smart Bidding algorithm has to learn from.
Smart Bidding Strategy Selection Guide
Use this decision path to choose the right starting strategy for a campaign:
Need more conversions or conversion value?
│
├── No conversion values tracked (lead gen, calls, form fills)
│ │
│ ├── New campaign / no conversion history
│ │ → Maximize Conversions (no target)
│ │
│ └── Stable conversion history, defined acceptable CPA
│ → Target CPA
│
└── Conversion values tracked (e-commerce, revenue)
│
├── Maximize total revenue within budget
│ → Maximize Conversion Value (no ROAS target)
│
└── Maintain a specific return on ad spend
→ Target ROAS (requires sufficient volume history)
Start with a less constrained strategy (Maximize Conversions or Maximize Conversion Value) when the account lacks sufficient history. Add a CPA or ROAS target once stable performance data exists. Avoid setting targets before the campaign has enough data to validate them.
Common Smart Bidding Mistakes
| Mistake | What Happens | Fix |
|---|---|---|
| Setting an unrealistic CPA or ROAS target | Delivery drops as the algorithm cannot find enough qualifying auctions | Set initial target at or above historical average; reduce gradually |
| Making frequent structural changes | Learning phase resets repeatedly; campaign never reaches stable performance | Make one change at a time; allow 2+ weeks between significant edits |
| Optimising for the wrong conversion action | Algorithm delivers efficiently toward the wrong business event | Audit primary conversion actions; ensure they represent real outcomes |
| Running budget-limited campaigns with a tight CPA target | Dual constraints prevent the algorithm from exploring bid ranges | Diagnose whether the campaign is budget-limited or target-limited; address the binding constraint first |
| Pooling unrelated campaigns in one portfolio strategy | High-volume campaigns skew targets for lower-volume ones | Group campaigns with similar objectives and conversion behaviour |
| Evaluating performance during the learning phase | Daily variance misleads optimisation decisions | Wait until the learning phase ends before assessing results |
Frequently Asked Questions
Which Smart Bidding strategy should I start with?
For new campaigns without a conversion history, Maximize Conversions is usually the right starting point. It builds volume without requiring a CPA target the algorithm cannot yet validate. Once you have stable conversion data, you can add a Target CPA based on actual average performance from that period. For campaigns with conversion value tracking, Maximize Conversion Value is the equivalent starting point before moving to Target ROAS.
How long does the Smart Bidding learning phase last?
Google indicates the learning phase typically lasts around one week, but this varies by conversion volume. Low-volume campaigns can remain in the learning phase significantly longer. Significant changes to targets, budgets, or campaign structure may reset or extend it. Avoid making major changes during the learning phase and resist drawing conclusions from daily performance variance during this period.
Why is my Target CPA campaign not spending its full budget?
The most common cause is a CPA target set too far below historical performance. The algorithm becomes very selective and restricts delivery to only auctions it considers highly likely to convert at your target price. Diagnose whether the campaign is budget-limited or target-limited first — then set the initial CPA target at or above your current average and reduce it gradually in 10–15% increments with at least two weeks between each adjustment.
Does Smart Bidding work without audience signals?
Smart Bidding can function without audience signals, but first-party data and observation audiences give the algorithm additional context about what converting users look like. Adding Customer Match lists, website visitor audiences, and in-market segments in observation mode improves signal quality without restricting delivery. As third-party cookie availability continues to decrease, first-party audience signals become increasingly important for Smart Bidding quality.
When should I use Target ROAS instead of Target CPA?
Use Target ROAS when your campaigns track meaningfully different conversion values — typically e-commerce accounts where product prices vary — and you want the algorithm to prioritise higher-value orders. If all conversions are assigned the same value, Target ROAS behaves identically to Target CPA. Target ROAS requires more conversion volume to function stably than Target CPA, and setting the target too high relative to historical performance will reduce delivery significantly.
Sources
TL;DR Smart Bidding is now central to Google Ads performance strategy, especially for campaigns with reliable conversion tracking. New campaigns without enough conversion data can…
