The Smart Bidding Audit: When Maximize Conversions Beats Manual CPC (And When It Doesn't)

Most Google Ads guides treat smart bidding as the universal default. Pick a strategy, set a target, let the algorithm run. The honest framing is more useful. Smart bidding works when 3 conditions are all true, and most accounts at $20K monthly spend do not meet at least one of them.

Google's own documentation on Target return on ad spend (ROAS) states that Search and Shopping campaigns need at least 15 conversions in the past 30 days to use the strategy. Most direct-to-consumer (DTC) accounts I audit at $20K monthly spend fall below even that lower bar on at least half their campaigns. The brands paying Google's algorithm to find audiences without the data sufficiency to teach it are paying for noise dressed up as automation. The patterns repeat across $150M+ in ad spend managed on DTC, software-as-a-service (SaaS), and lead-gen accounts. The diagnostic is straightforward enough that any operator can run it on their own account this week.

The 4 smart bidding strategies, explained for operators

There are 4 major smart bidding strategies in Google Ads in 2026, plus manual CPC sitting outside the group. Each one optimizes for a different outcome.

Maximize Conversions. Google bids to get the most conversions within the daily budget. No specific cost or value target. The algorithm decides what to spend per click based on conversion probability. Best for accounts where conversion volume matters more than conversion cost, like lead-gen with a tight daily budget cap.

Maximize Conversion Value. Google bids to maximize total conversion value (revenue) within the daily budget. The algorithm bids more on clicks that look likely to drive higher-value orders. Best for DTC ecommerce with varying average order values where some orders are worth several times more than others.

Target CPA (tCPA). Google bids to acquire conversions at a specified average cost. The operator sets the cost target. The algorithm tries to stay near that target on average. Best for lead-gen accounts with stable conversion value where CPA is the controllable metric.

Target ROAS (tROAS). Google bids to achieve a specified average return on ad spend. The operator sets the ROAS target. The algorithm bids more aggressively on high-value clicks and less on low-value ones. Best for DTC ecommerce with consistent gross margin where ROAS is the controllable metric.

Manual CPC sits outside this group. Manual lets the operator set the maximum bid per keyword directly. No algorithmic interpretation. Full control, no automation. Manual CPC is the precision instrument for accounts that have not earned the right to use the automated one yet.

Google Ads documentation on automated bidding groups these strategies under the "Smart Bidding" label specifically because each one uses machine learning to set bids on a per-auction basis. The contrarian read on the same documentation: machine learning needs data to learn from. A campaign with too few conversions per month is not training the algorithm. It is paying Google to guess. The minimum data thresholds are not aspirational. They are structural.

The 3 conditions smart bidding needs to work

Smart bidding fails reliably when any one of these is missing.

Condition 1. Sufficient conversion data. Google's published thresholds vary by strategy and campaign type. For Search and Shopping campaigns, Target ROAS needs at least 15 conversions in the past 30 days per campaign. Target CPA needs roughly the same. Maximize Conversion Value performs best above 30 conversions per 30 days with value tracked. Maximize Conversions can work at lower volumes but the algorithm learns faster above 10 conversions per 30 days. Most DTC accounts at $20K monthly spend fall below these thresholds on at least half their campaigns. The algorithm under-fed makes statistical noise decisions.

Condition 2. Clean conversion tracking. Smart bidding only optimizes for the events the algorithm sees. If conversion tracking fires the wrong event (a "form view" instead of a "form submit," or a "view cart" instead of an "order complete"), smart bidding accelerates the wrong outcome at scale. The algorithm finds more traffic that looks like form-viewers, not form-submitters. Stage 1 of the Google Ads audit method covers the conversion tracking diagnostic in full.

Condition 3. A profitable target metric. Setting Target ROAS at 4x when your account historically delivers 2.5x means Google will throttle spend dramatically to find the 4x conversions, which often do not exist at scale. The target has to be defensible against historical data plus a reasonable upside. For the gross-margin math behind a defensible ROAS target, see the gross-margin math behind a good ROAS. Pulling a number out of the air and asking the algorithm to chase it is one of the most common ways to spend half a quarter wondering why the campaign throttled itself to zero.

When any one of the 3 conditions fails, the right move is to switch back to manual CPC, fix the underlying issue, then re-attempt smart bidding 30 to 60 days later. Not pause the campaign. Not raise the budget. Fix the prerequisite.

The decision tree by campaign type

Different campaign types call for different default bidding strategies. The same account often needs 3 or 4 strategies running at once, one per campaign.

Brand search (Self-Defense Campaigns). Manual CPC. Always. Smart bidding on brand keywords overspends because the algorithm reads the high click-through rate (CTR) on branded queries as a signal to bid more aggressively. The traffic was already coming to you. Manual CPC caps the bid inflation. For the full case on why brand keywords need their own campaign and their own manual bid, see the Self-Defense Campaign methodology.

Non-brand search at $20K+ spend with value tracking. Maximize Conversion Value if the campaign has 30 or more conversions per month and your conversion values pass cleanly. Below that volume, manual CPC. The structural decision here pairs with match-type strategy. Broad match needs smart bidding to function. Exact and phrase work fine with manual. See the Match-Type Discipline framework for the pairing rules in detail.

Shopping (Standard, not Performance Max). Maximize Conversion Value if you have sufficient conversion data and value tracking. Manual CPC for new product categories that do not yet have a conversion history. Standard Shopping behaves like Search in this respect. The same 3 conditions apply.

Performance Max. Smart bidding is mandatory. Performance Max does not allow manual CPC by design. Set Target ROAS based on historical Shopping performance, not on aspiration. Audit the placement breakdown monthly. The Performance Max audit checklist covers the placement reporting, the asset group audit, and the conditions under which Performance Max actually outperforms Standard Shopping.

Display and YouTube acquisition campaigns. Avoid these entirely as acquisition campaigns in 2026. They should be retargeting layers only. The bidding strategy on the retargeting layer is whatever maximizes frequency to a controlled cap. Smart bidding on Display acquisition burns budget on impressions that do not convert.

Retargeting (any platform). Target CPA works because the conversion data is denser per impression. Manual CPC also works if you want precision over impression frequency. Either is defensible. The wrong move is leaving retargeting on the same bidding strategy as the acquisition campaign it feeds.

The pattern: smart bidding works best on high-data, value-tracked campaigns. Manual CPC works best on low-data, precision-required campaigns. Most accounts have a mix of both and need different strategies per campaign, not one strategy across the whole account. For the underlying framework on what each account stage means structurally, see the 4 stages of a Meta account. The same stages apply to Google accounts and pair directly with the bidding strategy choices in the table below.

The 4-step audit for your current bidding setup

The diagnostic operators can run this week. Four steps. No paid tool required.

Step 1. Pull conversion volume per campaign. Last 30 days. List every active campaign with its conversion count. Identify which campaigns are above and below the threshold for their current bidding strategy. A campaign running Target ROAS with 9 conversions in the last 30 days is starved for data. The algorithm cannot optimize on noise.

Step 2. Verify conversion tracking accuracy. In Google Ads conversion settings, confirm each campaign is optimizing for the right event (Purchase, qualified Lead, etc.), not an intermediate signal (View Content, Add to Cart). Cross-check against your Shopify orders, customer relationship management (CRM) entries, or whatever system records the actual revenue or qualified-lead count. The number Google sees and the number your business records should match within 5 to 10%. A larger gap means the bidding algorithm is optimizing for the wrong thing.

Step 3. Compare your bidding target to historical performance. Pull the 90-day average ROAS or CPA per campaign. Compare to your set target. If you are running Target ROAS at 4x and the campaign's 90-day average is 2.8x, the target is fictional. The algorithm will throttle spend to chase conversions that do not exist at that return. Reset to the historical average plus 20% upside, or switch to Maximize Conversion Value while the data stabilizes.

Step 4. Identify campaigns over-using smart bidding. Campaigns with under 10 conversions in the last 30 days running Maximize Conversions are starved. Campaigns with under 15 conversions on Target CPA or Target ROAS are below Google's own documented threshold. Switch these to manual CPC, build conversion volume manually, re-attempt smart bidding when the data justifies it.

A composite example. Rohan runs a DTC supplements brand at $40K per month in Google spend. He had just finished restructuring his match types to exact and phrase per the Match-Type Discipline framework. His next question: switch the new campaigns to Target ROAS or hold on manual CPC. The audit pulled conversion volume per campaign. Average across the new campaigns was 22 conversions in the last 30 days. Below Google's stated 30-conversion threshold for Target ROAS at a value-tracked stable level, but above the 15-conversion floor. The campaigns were on the edge. The recommendation was to stay on Maximize Conversion Value for 60 days while volume built, then re-attempt Target ROAS once the data was denser. The brand had wanted to switch immediately. Doing so would have throttled spend by 40% as the algorithm chased a fictional ROAS target on insufficient data.

A typical audit at $20K monthly Google spend finds 30 to 50% of campaigns using the wrong bidding strategy for their data depth. The fix is reassignment, not new infrastructure. The campaign-level audit covered in the full Google audit method walks through the same diagnostic across the other 7 stages.

When manual CPC actually outperforms smart bidding

Five scenarios where manual CPC is structurally better in 2026, despite the industry consensus that smart bidding wins everywhere.

1. Brand campaigns (Self-Defense Campaigns). Manual CPC. Always. Smart bidding on brand keywords overspends because the algorithm reads the high CTR on branded queries as room to bid more aggressively. The traffic was already coming to you. A $0.40 manual CPC bid on a branded keyword usually wins position 1. Smart bidding will push the same auction to $1.20 or higher because the algorithm thinks there is more headroom in conversion probability.

2. Launch-stage accounts with under 30 conversions per month total. Smart bidding cannot learn from noise. Manual CPC for the first 90 days builds the data foundation that smart bidding later optimizes. The right sequence is manual first, smart bidding once the account has earned the data depth. The wrong sequence is starting on smart bidding because Google's onboarding flow recommends it, then spending 60 days paying for the algorithm's learning curve.

3. Highly competitive auctions where you need precise bid control. Trademarked categories, legal services, finance, regulated verticals. The price ceiling matters because a conversion exists at $80 CPC but does not exist at $200 CPC. Smart bidding does not respect the operator's price ceiling. It chases the conversion probability the algorithm sees. If the conversion looks possible at $200, smart bidding will bid $200. Manual CPC contains the bid to a defensible maximum.

4. Accounts with seasonal volatility. Bridal, holiday, tax season, school supply. Smart bidding optimizes against recent data. A sudden seasonal spike confuses the algorithm because last week's conversion rate is not this week's. Manual CPC lets the operator anticipate the season the algorithm is still learning about. The operator who runs manual CPC during the seasonal spike and switches to smart bidding during the calm periods usually outperforms the operator who keeps smart bidding on year-round.

5. Accounts with poor conversion tracking that cannot be fixed quickly. Smart bidding accelerates whatever you are tracking. If tracking is wrong and the fix is 60 to 90 days out (waiting on a developer, a re-platforming, a Shopify app rebuild), manual CPC contains the damage while the tracking gets cleaned up. Smart bidding plus broken tracking is the most expensive combination on Google.

The closing position. Smart bidding is not an upgrade from manual CPC. They are different tools for different conditions. The brands that win Google Ads in 2026 use both, applied correctly to the campaigns where each one structurally fits. The industry default of "always switch to smart bidding" is not analysis. It is convenience for Google and convenience for the agencies managing dozens of accounts at once. Convenience for the platform is rarely the same thing as profit for the brand.

Why Google's CSMs push smart bidding too early

Google's customer success managers recommend smart bidding for the same structural reason they recommend Performance Max. Their job is to maximize the platform's algorithmic reach, not your account's profitability. The two goals overlap on healthy accounts. They diverge sharply on accounts that have not earned the data depth that smart bidding needs to function.

The structural read. Google's CSM compensation and key result areas (KRAs) reward spend growth and product adoption. Smart bidding is a product. Performance Max is a product. Recommending these products is in the CSM's interest regardless of whether the account is ready. The CSM is not lying. They are doing their job. The mistake is reading the recommendation as account-specific advice. It is product-specific advice.

The economic mechanic. Smart bidding works when 3 conditions are all true: sufficient conversion data, clean tracking, profitable target. The CSM cannot easily verify any of these from outside the account. So the recommendation defaults to "use smart bidding" because that is the platform-aligned recommendation regardless of conditions. The verification has to come from inside. The operator running the campaigns is the only person who can confirm the 3 conditions. The CSM cannot.

The honest framing. Smart bidding is not an upgrade. It is a different tool with different prerequisites. Manual CPC is the precision instrument for accounts that have not earned the right to use the automated one yet. The brands at $20K monthly spend who switched to smart bidding in 2023 because their CSM recommended it, without verifying conversion data sufficiency, are still paying for that decision in 2026. The way you stop paying for it is to run the 4-step audit above on your own campaigns and reassign the bidding strategy per campaign based on the data, not based on what the platform-aligned recommendation defaults to.

A Self-Defense Campaign on manual CPC, a non-brand Search campaign on Maximize Conversion Value, a Performance Max campaign on Target ROAS, and a retargeting campaign on Target CPA can all run in the same account at the same time. They should. Different campaign types call for different bidding strategies because they have different data depth, different intent, and different price ceilings. One strategy across the whole account is the sign of an account being managed for the platform's convenience, not the brand's profit.