Smart Bidding: Max Conv vs Target CPA 2026

-14 to -22% CPA: that's the median gain observed on an account moving from Manual CPC to mature Smart Bidding, but only after 14 to 21 days of learning. The right choice between Maximize Conversions and Target CPA at the right moment drives most of this gain — a wrong choice at the same moment halves it or even negates it.

It's probably the most Googled question from Google Ads advertisers in 2026: Maximize Conversions or Target CPA? Both are Smart Bidding strategies, both use the same AI engine, both maximize conversions. And yet, they produce radically different results on the same campaign.

On our 2025 sector benchmark, the performance gap between "right strategy at the right moment" and "wrong choice" reaches -14 to -22% average CPA depending on vertical, after manual → automated migration. The friction point isn't Smart Bidding itself (a lever we cover in detail in our CPA reduction guide (lever 3)), it's the choice between these two variants.

This guide decides. We start by understanding how each algorithm decides, then compare number by number, give a 4-question decision tree, and finish with the migration procedure. Plan 10 minutes of reading and 30 minutes to apply on your account.

What is Smart Bidding and why is it more than an upgrade?

Smart Bidding refers to the full set of AI-driven Google Ads bidding strategies: Maximize Conversions, Target CPA, Target ROAS, Maximize Conversion Value, and more recently Maximize Clicks (which falls outside the "conversion" scope). The underlying engine is the same: a deep learning model that evaluates each auction in real time and adjusts the bid based on 70+ signals — device, time of day, precise location, user history, exact query, browser, OS, audience membership, and many others not publicly documented.

The difference from Manual CPC isn't marginal. In manual mode, you set a bid per keyword, possibly modulated by 3-4 bid modifiers (device, audience, hour). In Smart Bidding, each auction gets a unique bid, calculated in 50 milliseconds based on dozens of signals. On a $10,000/month account, that's the difference between 200 different bids and millions of different bids.

Maximize Conversions and Target CPA share this engine. What changes between the two: the objective function. One maximizes volume under budget constraint. The other maximizes volume under cost-per-conversion constraint. This difference in objective produces two radically opposed behaviors on the same account, and that's where 95% of advertisers go wrong.

How does the Maximize Conversions algorithm decide on bids?

Objective function: maximize the number of conversions over the period, spending up to 100% of the daily budget. No cost ceiling. No notion of "too expensive." Just: spend everything and bring back as many conversions as possible.

How the algorithm decides concretely, auction by auction: for each incoming query, it calculates a conversion probability based on available signals. If this probability is above the implicit breakeven threshold (budget / expected conversions), it bids. The higher the probability, the more aggressive the bid — the algorithm is willing to pay very high for the auction if it "senses" the conversion is nearly locked (user returning to the site, ultra-commercial query, qualified audience).

Major signals used:

• Remaining daily budget (priority: spend it)
• Account conversion history (last 30 days)
• User signal quality (audience, recency, engagement)
• Query and commercial intent
• Device + geo + hour + day
• Auction competition (how many other advertisers)

Ideal use cases: campaign launch, account below 30 conv/month, promo period where volume trumps unit cost, exploration phase after a major change (new landing, new funnel).

How does the Target CPA algorithm decide on bids?

Objective function: maximize the number of conversions while respecting a target cost per conversion. The budget isn't necessarily fully spent. If the algorithm can't find an auction where it can convert at ≤ target, it doesn't serve. It's a paradigm shift: budget becomes secondary, the CPA target becomes the alpha and omega.

See the official Target CPA documentation for Google's details. On the algorithmic side, the operation is subtle: the goal isn't "each conversion costs exactly the target," but "the average conversion over the period converges toward the target". Over a week, some conversions may cost 2× the target (ultra-qualified user the algorithm overbid), and others 0.5× the target (cheap long-tail opportunity). The average aims at the target.

What changes vs Maximize Conversions:

• The algorithm refuses auctions it judges above the implicit unit budget (target × error margin).
• Volume is constrained by the target. Lowering the target by 10% typically reduces volume by 15 to 25%.
• The algorithm learns to distinguish good conversions (qualified users, strong intent) from apparently good ones (high CTR but low conv rate). This distinction takes 14 to 21 days.
• Target changes greater than 15% restart a full 7-day learning period.

Recommended starting target: historical CPA + 10 to 15%. Never below historical CPA or the algorithm will collapse. Example: if your 30-day average CPA is $42, start Target CPA at $48 (≈ +14%). Once stability is reached (D+21), lower to $45. Then $42 two weeks later if volume holds. This is how you reach the 15 to 25% savings vs manual.

Ideal use cases: mature account with 50+ conv/month, budget-sensitive CPA (tight-margin sectors), progressive scaling where cost discipline trumps raw volume.

Quantified comparison: Maximize Conversions vs Target CPA

Here's the side-by-side table of the 10 criteria that matter. Numbers drawn from our internal analysis, cross-checked with the official Google comparison.

Reading the table: Max Conv and Target CPA aren't two levels of the same strategy, they're two different products. Max Conv optimizes for a world where cost per conversion isn't a constraint (launch, test, budget set elsewhere). Target CPA optimizes for a world where CPA is the key variable of your P&L. You must explicitly choose which of the two worlds matches your situation.

How many conversions do you need to activate Smart Bidding?

Official Google threshold for Target CPA: 30 minimum conversions over the last 30 days on the concerned campaign. Max Conv has no minimum threshold, but we advise 15 conv/month for learning to be useful (below that, you're in pure broad targeting).

Another often-forgotten prerequisite: your conversion tracking must be reliable. Smart Bidding optimizes toward the signal it receives. If a significant share of your conversions is poorly tracked, the algorithm optimizes toward a truncated signal and performs -16 to -26% below its potential depending on vertical. It's the #1 bug we see in audit.

Third prerequisite: an average Quality Score ≥ 5/10 on main keywords. Our Quality Score guide details why a low QS caps your minimum achievable CPA, regardless of bidding strategy. Smart Bidding can partially compensate for a medium QS, not a catastrophic QS.

Decision: which to choose across 5 scenarios

The decision tree below summarizes the logic in 4 questions. If you answer in order, you reach the right answer in 90% of cases.

Application to the 5 most frequent scenarios:

1. E-commerce that just launched (3 months, 20 conv/month): Maximize Conversions. Not enough signal for Target CPA. Stay on Max Conv until crossing 40-50 conv/month, then switch.
2. Mature B2B SaaS (2 years, 80 leads/month, CPA $112): Target CPA at $125 (+12%) for 21 days, then descent to $110 then $95 in 10% steps.
3. Local services lead-gen (old account but volatile CPA): Target CPA with a wide target (+20%) to start, then tighten. Alternative: Max Conv with tight budget.
4. Fashion e-commerce with variable cart $40 to $400: Target ROAS rather than Target CPA (see FAQ). Target CPA would treat a $40 sale like a $400 sale = suboptimal.
5. Black Friday / flash promo period: Maximize Conversions temporarily. CPA constraint is suspended, the objective is volume. Return to Target CPA D+7 after the promo.

Migrate from one to the other: step-by-step procedure

The Max Conv → Target CPA migration (the most frequent direction) is done in 5 steps over 4 weeks. Don't rush any step, or you'll restart learning and lose 2 to 3 weeks of data.

1. Week 0 (audit): verify the 3 prerequisites (30+ conv, reliable tracking, QS ≥ 5). Our free audit runs the 3 checks in 3 minutes.
2. Week 1 (target calculation): calculate the average CPA of the last 30 days, multiply by 1.12 → your starting target. No lower. No higher than 20% above either (useless).
3. Week 2 (switch): Campaign Settings → Bidding → Change strategy → Target CPA. Enter the target. Let it run 7 days without modifying anything.
4. Week 3 (stabilization): verify that volume held at ±15% of Max Conv level. If yes, lower the target by 10%. If no, leave 7 more days.
5. Week 4+ (optimization): progressive descent in 10% steps every 2 weeks as long as volume holds. Stop as soon as volume drops more than 20%.

For the reverse migration (Target CPA → Max Conv, e.g., before a promo period), it's faster: immediate switch, the algorithm recalibrates in 3-5 days. The "cost" is higher CPA variance in the first days.

Should you merge campaigns via Portfolio Bidding?

Portfolio Bidding lets you share a single bidding strategy (Target CPA or Target ROAS) across multiple campaigns. Google then pools the signal: 3 campaigns of 20 conv/month each become 1 portfolio of 60 conv/month = Target CPA threshold crossed.

When to use it: you have 3+ campaigns sharing the same CPA (or ROAS) objective, the same conversion type, and a similar audience (same sector, same broad geo). This is typical of e-commerce accounts with 1 campaign per product category.

When to avoid: if your campaigns have very different expected CPAs (e.g., SEO branded CPA $5 + cold acquisition CPA $80), the portfolio will level down and degrade both. Better to keep separate strategies. Rule of thumb: don't merge if historical CPAs vary by more than 2× across campaigns.

The 3 beginner mistakes to avoid

1. Setting a Target CPA 30% below historical CPA. It's the #1 mistake we see. The algorithm can't reach the target, it cuts serving, your volume collapses to -60% in 7 days. Absolute rule: never below historical CPA at start. +10 to +15% minimum, then progressive descent.
2. Changing the target every week. Each change greater than 15% restarts 7 days of learning. If you change every Monday, your campaign is permanently in the learning phase, therefore never at optimal performance. Leave at least 14 days between two changes.
3. Using Max Conv on a small budget without supervision. Max Conv can spend 100% of the budget in 2-3 "bad" days if signal is fuzzy. On an account under $50/day, switch to Target CPA as soon as possible, even with a wide target (+30% vs historical), rather than letting Max Conv run without a net.

Sources

Official sources consulted for this guide:

• support.google.com
• support.google.com
• support.google.com

FAQ