Choosing the wrong bidding strategy in Google Ads doesn't just cost you money on the wrong clicks. It can trap a perfectly good campaign in a learning loop it never escapes, or set the algorithm chasing a target so aggressive it stops spending entirely. The bid strategy decision is often treated as a setup checkbox. It should be treated as one of the most consequential choices in account management.
This guide covers every Smart Bidding strategy available in 2026, explains what each is actually optimising toward, and gives you the specific conditions under which each performs — and where each fails. No generic "it depends." Just the framework.
What Smart Bidding Actually Means
Smart Bidding is Google's term for bid strategies that use AI to optimise for conversions or conversion value at auction time. The key phrase is "auction time." Unlike older automated strategies that adjusted bids periodically or used simple rules, Smart Bidding sets a unique bid for every single auction based on dozens of real-time signals: device, location, time of day, browser, search query, remarketing list membership, recent site behaviour, and more.
As of 2026, there are four core Smart Bidding strategies:
- Maximize Conversions — get as many conversions as possible within your budget
- Maximize Conversion Value — prioritise higher-value conversions over sheer volume
- Target CPA — a goal layered onto Maximize Conversions to hit a specific cost per acquisition
- Target ROAS — a goal layered onto Maximize Conversion Value to hit a specific return on ad spend
One important technical clarification that confuses many advertisers: Target CPA and Target ROAS are not independent strategies. They are constraint layers applied on top of the Maximize Conversions and Maximize Conversion Value frameworks respectively. Maximize Conversions with a Target CPA set behaves exactly like a Target CPA strategy. Maximize Conversion Value with a Target ROAS set behaves exactly like Target ROAS. The naming in the Google Ads interface can make this feel more complicated than it is.
Also worth noting: Enhanced CPC (ECPC), the hybrid between manual and automated bidding, was deprecated for Search and Display campaigns in March 2025. If you had campaigns using ECPC and didn't proactively migrate them, they're now effectively running on Manual CPC.
Maximize Conversions: The Starting Line
Maximize Conversions tells Google: spend my daily budget, and get as many conversions from it as you can. There is no efficiency target. There is no cost constraint. The algorithm bids aggressively when it sees conversion opportunity and pulls back when it doesn't.
This sounds dangerous — and for poorly set up campaigns, it can be. But for well-structured campaigns at the right stage of their lifecycle, Maximize Conversions is the correct choice.
When Maximize Conversions is right:
Your campaign is new or has recently had its bidding strategy reset. Maximize Conversions lets the algorithm explore the conversion landscape without an efficiency constraint artificially limiting what it can learn from.
Your conversion volume is between 15 and 50 per month. Below this range, any Smart Bidding strategy is making decisions with insufficient data. Above it, you have enough signal to add a CPA or ROAS target. In between, let Maximize Conversions build the data that will make a target meaningful.
You want to test a new campaign or ad group. Constraining a new test with an aggressive efficiency target before it has any performance history is one of the most common ways to get misleading results from Google Ads experiments.
When Maximize Conversions becomes a problem:
Without a Target CPA set, Maximize Conversions will spend your entire daily budget regardless of the CPA it achieves. If your budget is large relative to the available conversion volume, the algorithm will pay increasingly for marginal opportunities just to exhaust the budget. This is when you see the CPA creeping up — not because the campaign is broken, but because it's doing exactly what you asked.
The fix is straightforward: either set a Target CPA to give the algorithm an efficiency guardrail, or reduce the budget to match actual conversion opportunity.
Maximize Conversion Value: The E-Commerce Version
Maximize Conversion Value is Maximize Conversions' sibling for businesses where conversions have different monetary values. Rather than optimising for the number of conversions, it optimises for the total value of conversions generated within your budget.
The practical difference matters enormously for e-commerce. If you sell products ranging from $15 to $2,000, optimising for conversion count will push the algorithm toward the easy, low-value purchases. Optimising for conversion value tilts it toward the higher-revenue transactions, which is almost certainly what you want.
Using Maximize Conversion Value requires accurate conversion value tracking. If your conversions are all assigned the same static value (or no value at all), this strategy performs identically to Maximize Conversions. The value signal is what differentiates it.
Target CPA: Volume With Efficiency Guardrails
Target CPA is the mature version of Maximize Conversions. You tell Google the average cost you're willing to pay per conversion, and the algorithm works to maximise conversion volume within that constraint. Some conversions will cost more than your target, some less, but the average should trend toward your specified goal over time.
The word "average" is crucial. Target CPA does not mean every conversion will cost exactly your target. It means the algorithm aims for that average across all the auctions it participates in. Evaluating Target CPA performance over a single day or a handful of conversions will always produce misleading data.
The data requirements:
Target CPA needs a foundation of conversion history to work from. Google's current guidance suggests 15-30 conversions in the past 30 days as a minimum. That's genuinely a minimum — accounts with 50-100 monthly conversions per campaign see significantly more stable and reliable Target CPA performance. Below 15 conversions per month, the algorithm doesn't have enough signal to distinguish between efficient and inefficient auctions.
How to set your Target CPA:
This is where most advertisers make a mistake. They set their target CPA at the number they want to achieve rather than the number they currently achieve.
If your historical CPA is $80 and you set a Target CPA of $40, you're asking the algorithm to achieve a 50% efficiency improvement overnight with no operational changes. The algorithm will respond by dramatically restricting spend — bidding only on the highest-confidence conversion opportunities to protect the aggressive target. You'll see impressions collapse, clicks dry up, and conversion volume plummet. The campaign looks broken. In reality, it's rational: you gave it an impossible constraint.
The correct approach: set your Target CPA at or slightly above your current average CPA. Let the campaign find its footing and hit the target consistently. Then tighten the target by 10-15% every three to four weeks as long as performance holds. Gradual tightening gives the algorithm time to adapt without triggering a full reset of the learning period.
The transition from Maximize Conversions to Target CPA:
Don't make this change simultaneously with other significant account changes. Switching bid strategies resets the learning period. Adding new keywords, restructuring ad groups, or launching new ads at the same time compounds the instability. Make one significant change at a time and give each at least two weeks before evaluating.
Target ROAS: Value Efficiency for Data-Rich Accounts
Target ROAS is the efficiency-constrained version of Maximize Conversion Value. You tell Google what return on ad spend you want to achieve — expressed as a percentage — and the algorithm maximises conversion value while targeting that ratio.
If you set a Target ROAS of 400%, you're telling Google: for every rupee spent on advertising, generate four rupees in conversion value. The algorithm bids aggressively when it predicts a high-value conversion and pulls back when it predicts a low-value one.
The data requirements are higher than Target CPA:
Target ROAS needs not just conversion volume but conversion value diversity to work properly. It needs to see enough variation in conversion values to model the relationship between auction signals and likely conversion value. The minimum is 15 conversions with valid conversion values in the past 30 days, but in practice, Target ROAS tends to be unstable below 50 conversions per month.
When Target ROAS is the right choice:
You're in e-commerce with variable order values and you're tracking revenue accurately as your conversion value. This is the canonical use case for Target ROAS, and it's the environment where it performs most reliably.
Your profit margins are tight and you need to protect efficiency, not just volume. Target ROAS gives you a direct line between your advertising spend and your revenue return — something Target CPA cannot provide when conversion values vary.
You've already run Maximize Conversion Value and built sufficient data. The progression from Maximize Conversion Value → Target ROAS mirrors the progression from Maximize Conversions → Target CPA.
When Target ROAS is the wrong choice:
You're in lead generation with fixed or estimated lead values. If every lead is assigned the same static value, Target ROAS and Target CPA will produce similar results. Use Target CPA — it's simpler and easier to communicate to stakeholders.
Your conversion volume is under 30 per month per campaign. The efficiency constraint will strangle spend before the algorithm has enough data to work with.
You've just launched the campaign. Starting with Target ROAS on a new campaign with no historical data is asking the algorithm to optimise for a ratio it has no basis for calculating.
How to Set Targets That Don't Break Your Campaigns
Both Target CPA and Target ROAS are only as useful as the targets you set. Setting them based on aspiration rather than current performance is the single most common cause of Smart Bidding failure.
For Target CPA:
Calculate your acceptable CPA from your business economics, not from the Google Ads interface. If your customer lifetime value is ₹20,000 and you want a 4:1 return, your acceptable CPA is ₹5,000. If your current average CPA is ₹7,000, start your target there and work down gradually. Don't jump straight to ₹5,000.
Give the algorithm flexibility by starting 10-15% above your true target. A Target CPA of ₹7,700 when you really want ₹7,000 gives the algorithm room to find conversions while trending toward your actual goal. Tighten progressively.
For Target ROAS:
Calculate from your margins, not from your wishes. If your average margin is 35% and you want to break even on advertising after product cost, your minimum viable ROAS is roughly 285%. If you want advertising to be responsible for a third of profit after that, you might target 350-400%.
Smart Bidding won't hit your target immediately. Expect 30-60 days of averaging toward your goal. Don't evaluate performance in week one.
Set ROAS targets 10-15% lower than your true efficiency floor. Using flexible ROAS targets — targets with some room beneath your actual goal — consistently delivers access to more converting query volume while maintaining overall efficiency. Advertisers who set rigid targets at their exact efficiency floor systematically lock themselves out of incremental volume.
The Progression Framework: Which Strategy at Which Stage
The right bidding strategy isn't static. It should evolve with your account.
Stage 1 — New campaign with no history: Use Maximize Conversions or Maximize Conversion Value with no target set. Build at least 30 conversions per month before adding any constraint.
Stage 2 — 30-50 conversions per month: Add a Target CPA (or Target ROAS for e-commerce) at or slightly above your current average. Monitor for two to four weeks before making target adjustments. Avoid other significant changes during this period.
Stage 3 — 50-100 conversions per month: Begin gradually tightening your target by 10-15% every three to four weeks as long as performance holds. At this conversion volume, the algorithm has enough data to respond to incremental efficiency improvements.
Stage 4 — 100+ conversions per month: Consider portfolio bid strategies if you have multiple campaigns with similar goals. Portfolio strategies pool data across campaigns, which can significantly improve algorithm performance compared to individual campaign strategies. An account with 300 total monthly conversions split across six campaigns gives each campaign 50 conversions to learn from. A portfolio strategy gives the combined algorithm 300 conversions of signal.
The Common Failure Modes
Understanding why Smart Bidding fails is as important as knowing how to set it up.
Campaign not spending budget (underspending): Almost always caused by a Target CPA or ROAS that's too aggressive. The algorithm cannot find enough high-confidence conversion opportunities at that efficiency level and restricts spend. Solution: loosen the target and tighten gradually.
CPA rising over time: Either budget is outpacing available conversion opportunity (causing the algorithm to bid on lower-quality traffic), or your landing page or offer quality has changed. Diagnose by comparing impression share to previous periods — if impression share is holding but CPA is rising, the issue is conversion rate, not the bidding strategy.
Erratic week-to-week performance: Insufficient conversion volume. The algorithm doesn't have enough data to make stable predictions. Either build more conversion volume before using targets, or use portfolio strategies to pool data.
Learning period never ending: Significant account changes reset the learning period. If you're making frequent changes to keywords, ads, bids, or campaign settings, you may be perpetually resetting the learning cycle. Give campaigns a minimum of two weeks of stability.
Smart Bidding optimising toward the wrong conversions: The algorithm optimises toward whatever you've included in your "Conversions" column. If you've included micro-conversions (page views, scroll depth, time on site) alongside macro-conversions (purchases, form fills), the algorithm may be optimising for quantity of micro-conversions rather than quality of macro-conversions. Audit your conversion actions and ensure your bid strategy is learning from the actions that actually reflect business value.
The Bottom Line
Smart Bidding is not a black box that occasionally produces results. It's a systematic optimisation engine that performs in direct proportion to the quality of the inputs you give it: conversion volume, conversion value accuracy, realistic targets, and sufficient time to learn.
The four strategies form a logical hierarchy. You start with maximum volume collection (Maximize Conversions or Maximize Conversion Value), establish a performance baseline, add efficiency constraints only when you have the data to make them meaningful, and tighten those constraints gradually as the algorithm proves it can deliver.
Every shortcut — starting with aggressive targets, switching strategies without sufficient data, making changes before the learning period completes — adds weeks or months to the time before your campaigns perform reliably. Patience with Smart Bidding isn't passivity. It's the precondition for performance.
Portfolio Bid Strategies: When Individual Campaign Strategies Aren't Enough
As your account matures and you're managing multiple campaigns with similar objectives, individual campaign-level bid strategies become limiting. A portfolio bid strategy pools conversion data across multiple campaigns, giving the algorithm a larger data set to work from and often producing more stable, better-performing results than isolated strategies.
The math is straightforward. Six campaigns each generating 50 conversions per month gives each campaign's algorithm 50 data points. A portfolio strategy applied to all six gives the combined algorithm 300 data points. The difference in prediction accuracy at 300 data points versus 50 is substantial — more confident bid decisions, faster adaptation to changes, more stable week-to-week performance.
Portfolio strategies work best when the campaigns share the same conversion goals and target metrics. Mixing campaigns with fundamentally different objectives or wildly different CPAs into a single portfolio can produce poor results — the algorithm tries to optimise toward the average, which may not serve any individual campaign well.
To create a portfolio bid strategy: Tools and settings → Bid strategies → + (new). Select your strategy type, apply it to the relevant campaigns. Monitor portfolio-level performance alongside individual campaign metrics.
The New 2026 Option: Campaign Total Budget
In early 2026, Google introduced Campaign Total Budget — a fixed spending amount for the entire campaign flight rather than a daily budget that recurs indefinitely. This is distinct from daily budgets and has different pacing behaviour.
Campaign Total Budget is particularly useful for:
Time-limited campaigns (seasonal promotions, product launches, event advertising) where you want to guarantee a specific total spend over a defined period. Advertisers who have struggled with daily budget pacing inconsistencies — Google's allowance to spend up to twice your daily budget on high-traffic days can cause end-of-month surprises. Total Budget eliminates that variability by capping absolute spend.
For evergreen campaigns with no natural end date, Daily Budget remains the appropriate choice. Campaign Total Budget is a planning tool, not a replacement for ongoing campaign management.
Smart Bidding and Seasonality: The Seasonal Adjustments Feature
Smart Bidding learns from historical patterns. This is usually an advantage — the algorithm understands that your conversions peak on weekdays, that your best customers search in the evening, that certain times of year drive higher order values. But it can become a liability during predictable seasonal spikes that fall outside the algorithm's training window.
If you're running a major promotional sale, launching a new product, or approaching a known high-traffic period (major holidays, end-of-financial-year, industry events), Smart Bidding may underestimate conversion opportunity based on historical data that doesn't account for the spike.
Seasonality adjustments allow you to tell the algorithm to expect a temporary change in conversion rate. You can apply an adjustment for a specific date range, specifying the anticipated conversion rate change as a percentage. The algorithm temporarily recalibrates its predictions for that period, bidding more aggressively when it expects higher conversion rates.
This feature is underused and highly effective for accounts with predictable seasonal patterns. Apply it 24-48 hours before the seasonal event begins and set it to expire naturally when the event ends.
Reading Smart Bidding Performance: The Metrics That Actually Matter
Many advertisers evaluate Smart Bidding by the wrong metrics. Impression share, average position, and Quality Score are useful diagnostics but not the right scorecard for a Smart Bidding strategy. The metrics that matter are:
Conversion volume — is the campaign generating enough conversions to learn effectively? Below 30 per month, performance will be erratic regardless of strategy.
CPA trend over 30-day periods — not week over week (too much noise), but month over month. Is your CPA trending toward your target as the algorithm accumulates data?
Conversion value / cost (ROAS) — for value-based strategies, the aggregate ratio over 30-day windows. Not individual transaction ROAS, but the portfolio average.
Budget utilisation — is the campaign spending its full budget? Consistent underspending on a Smart Bidding campaign usually indicates your target is too aggressive. The algorithm isn't finding enough auctions where it's confident it can hit your target efficiently.
Lost impression share (budget vs. rank) — if you're losing impression share due to budget, add budget. If you're losing it due to rank, your bids are too low or your quality signals need improvement.
The Decision You Keep Avoiding
Most Smart Bidding problems come back to one avoidance: advertisers set aspirational targets rather than data-grounded ones, the campaign underperforms against those targets, they tweak and adjust before the learning period completes, they switch strategies prematurely, and they conclude that Smart Bidding doesn't work.
Smart Bidding works. The friction is almost always with patience and with the willingness to set targets that reflect current reality rather than desired future outcomes.
The discipline required is this: decide your target based on current account data and your business economics. Set it where the data supports it, not where you wish it were. Give the algorithm a full learning period — two weeks minimum, four weeks ideally. Then evaluate. Then make one change at a time and give each change its own evaluation window.
This process feels slow. It is slow, relative to the instant feedback loops that make digital advertising appealing. But it's the only process that produces reliable, compounding Smart Bidding performance. Every shortcut resets the clock.