Bidding Strategies Explained: When to Use Target CPA vs Max Conversions

Why Your Bidding Strategy Is the Biggest Decision in the Account

Most Google Ads advice obsesses over keywords and ad copy, but the bidding strategy is the setting that decides what the entire account is actually trying to do. Every other choice you make, the keywords you target, the audiences you layer, the budget you set, gets filtered through the bid strategy before it touches the auction. Pick the wrong one and you can have perfect keywords, great ads, and a fast landing page, and still watch the algorithm spend your budget chasing the wrong outcome.

The confusion is understandable, because Google’s naming makes the options sound interchangeable. Maximize Conversions, Target CPA, Maximize Conversion Value, Target ROAS: they all sound like “get me more sales for less money.” In practice they give the algorithm fundamentally different instructions, and they behave very differently depending on how much conversion data your account produces. A strategy that works beautifully at 100 conversions a month can be actively destructive at 10.

This guide walks through each Smart Bidding strategy, what it actually optimizes for under the hood, and, most importantly, the data thresholds that determine when each one becomes safe to use. The thresholds matter more than the marketing names. Smart Bidding is a prediction machine, and prediction machines are only as good as the evidence you feed them.

How Smart Bidding Actually Works: Auction-Time Predictions, Not Magic

Before comparing strategies, it helps to understand what Smart Bidding is doing mechanically, because every strategy is the same engine with a different objective bolted on.

In every auction, Google estimates the probability that this specific click, from this specific person, at this specific moment, will convert. It draws on signals you cannot bid on manually: device, location, time of day, browser, operating system, search history context, audience list membership, even the actual query rather than just the keyword that matched it. It then sets a bid for that one auction based on the predicted conversion probability and whatever objective you gave it. High predicted probability, higher bid. Low probability, lower bid or no bid at all.

Two consequences follow from this. First, manual bidding cannot compete on granularity. A human sets one bid per keyword, maybe adjusted by device and location; the machine sets a different bid for every auction. Second, and this is the part advertisers underestimate, the predictions are built from your account’s conversion history. An account with hundreds of recent conversions gives the model rich evidence about who converts. An account with eight conversions last month gives it almost nothing, and the model falls back on broad averages and guesswork. That is why every strategy below comes with a volume threshold, and why ignoring those thresholds is the single most common Smart Bidding mistake we see in audits.

Maximize Conversions: The Starting Point for Most Accounts

Maximize Conversions gives the algorithm the simplest possible instruction: spend the daily budget in whatever way produces the most conversions. No cost ceiling per conversion, no efficiency target, just volume within budget.

That simplicity is both its strength and its danger. The strength is that it works with very little data. Because it is not trying to hit a specific cost target, it does not need a long conversion history to function; it just bids toward the highest-probability auctions it can find within your budget. This makes it the natural starting strategy for new campaigns, new accounts, and accounts coming off a tracking fix, where there is not yet enough history for a target-based strategy to behave sensibly.

The danger is the missing cost ceiling. Maximize Conversions will happily pay far more per click than you would ever choose manually if it predicts a conversion on the other side, and it is strongly inclined to spend the entire daily budget whether or not the auctions available that day are any good. The classic failure mode is launching it with a budget far above what the campaign realistically needs: the algorithm interprets the budget as permission and CPCs balloon. The practical guardrails are simple. Set the daily budget at what you genuinely intend to spend, not an aspirational number. Watch average CPC in the first two weeks, because that is where the trouble shows first. And treat Maximize Conversions as a phase, not a destination: its job is to accumulate conversion data efficiently enough that you can graduate to a target-based strategy with real evidence behind it.

Target CPA: Efficiency Control, but Only With Enough Data

Target CPA changes the instruction from “get the most conversions” to “get conversions at roughly this average cost.” You tell Google what a conversion is worth paying for, say 60 dollars, and the algorithm bids to keep your average cost per conversion around that number over time. Individual conversions will come in above and below it; the target is an average, not a cap.

When it has enough data, Target CPA is the workhorse of lead generation accounts. It gives you a real efficiency lever: lower the target and the algorithm gets pickier, entering fewer auctions and buying cheaper, higher-probability clicks at the cost of volume; raise it and volume opens up at the cost of efficiency. That trade-off, volume versus efficiency, dialed with a single number, is the entire game of bid management compressed into one setting.

When it does not have enough data, Target CPA is erratic. With thin conversion history, the model cannot reliably distinguish good auctions from bad ones, so it either throttles spend to almost nothing because few auctions clear its confidence bar, or it swings wildly as every individual conversion reshapes its picture of the account. The other classic mistake is setting the target at what you wish conversions cost rather than what they actually cost. If your trailing 30-day CPA is 90 dollars and you set a 45 dollar target, you have not made the account more efficient, you have told the algorithm to stop buying most of your traffic. The discipline that works: start the target within roughly 10 to 20 percent of your actual trailing CPA, then walk it down in small steps every couple of weeks, letting performance stabilize between moves.

Target CPA vs Max Conversions: The Data Thresholds That Decide It

So when do you use which? The honest answer is that it is less a stylistic choice than a function of conversion volume, and the typical industry guidance is fairly consistent.

Below roughly 15 conversions in the last 30 days at the campaign level, target-based bidding is on thin ice; Google’s own guidance has long pointed to a floor in this neighborhood for Target CPA, and most practitioners treat it as a bare minimum rather than a comfortable operating level. In this zone, run Maximize Conversions without a target and focus your energy on getting volume up: consolidating campaigns so conversions pool together, broadening to a slightly higher-funnel conversion action, or fixing tracking so the conversions you do get are all counted.

Between roughly 15 and 30 conversions a month, you are in the transition zone. Target CPA can work here, but expect volatility, and set the target generously, at or slightly above your actual trailing CPA, so the algorithm is not starved of auctions while it learns. Above roughly 30 to 50 conversions a month per campaign, Target CPA becomes the better tool for most lead-gen accounts, because the efficiency control starts being worth more than the raw volume push, and the model finally has enough evidence to hit a target without lurching.

These are typical ranges, not laws of physics, and they describe campaign-level volume, not account-level. Two other notes matter. First, conversion lag changes the math: if your leads take two weeks to register as conversions, the algorithm is always learning from stale data, and you should be more conservative with targets. Second, since Google merged the strategies in the interface, Target CPA is literally implemented as Maximize Conversions with a target CPA field. The practical workflow this enables is the cleanest path we know: launch on Maximize Conversions, accumulate volume, then add a target near your actual CPA once you cross the threshold, no strategy migration or learning reset required beyond the adjustment period the target itself triggers.

Maximize Conversion Value and Target ROAS: When Revenue, Not Leads, Is the Goal

Everything above treats every conversion as equal, which is fine when a lead is a lead. It breaks down the moment your conversions carry different values: an ecommerce store where orders range from 30 to 800 dollars, or a service business where a kitchen renovation lead is worth twenty times a faucet repair.

That is what the value-based pair is for. Maximize Conversion Value is the revenue analogue of Maximize Conversions: spend the budget to produce the most total conversion value, regardless of efficiency. Target ROAS is the analogue of Target CPA: hold a specific return on ad spend, expressed as a percentage, so a 400 percent target tells the algorithm to aim for four dollars of conversion value per dollar of spend. The same volume logic applies, but the bar is higher, because the model now has to predict not just whether a click converts but how much it will be worth. Typical industry guidance puts the comfortable floor for Target ROAS around 30 to 50 value-bearing conversions in the last 30 days, and accounts with highly variable order values need more history, not less, for the predictions to settle.

The prerequisite that gets skipped: value-based bidding requires real values flowing into the account. For ecommerce that means accurate transaction-value tracking. For lead gen it means assigning differentiated values to conversion actions, or better, importing offline conversion values from your CRM so the algorithm learns which leads became revenue. Feeding Target ROAS a single static value per lead just recreates Target CPA with extra steps. If you cannot yet measure value honestly, stay on the conversion-count strategies; a clean signal pointed at the wrong-but-honest goal beats a noisy signal pointed at the right one.

The Other Strategies: Where Manual CPC, Maximize Clicks, and Impression Share Still Fit

Smart Bidding gets the attention, but three other strategies still exist, and each has a legitimate, narrow use case.

Manual CPC, with Enhanced CPC now retired, is the last fully hands-on option, and its honest role has shrunk to diagnostics and special cases: brand-new accounts with zero conversion tracking where you want controlled data collection, tightly budgeted experiments where you need a hard ceiling on what any click can cost, and verification work when you suspect Smart Bidding is overpaying and want a baseline. As a long-term operating strategy for a lead or sales campaign, it concedes the auction-time signals to every competitor running automation, and that concession gets more expensive every year.

Maximize Clicks is the traffic strategy: most clicks for the budget, no opinion about conversions. Its defensible uses are short. New campaigns gathering initial data before any conversions exist, and situations where traffic itself is the goal. The trap is leaving it on past that phase, because it systematically prefers cheap clicks, and cheap clicks and converting clicks are usually different populations.

Target Impression Share bids for visibility, a percentage of auctions where your ad shows, optionally pinned to absolute top of page. It exists for one scenario worth paying for: defending your own brand terms, where you want to be present nearly every time someone searches your name and conversion-based logic might let a competitor slip above you. On non-brand campaigns it is almost always the wrong tool, since paying for visibility regardless of conversion probability is precisely what Smart Bidding exists to avoid.

Learning Periods, Target Changes, and the Mistakes That Reset Your Progress

Whichever strategy you choose, how you manage it week to week matters as much as the choice itself, and most self-inflicted damage falls into a few repeatable patterns.

The first is impatience with the learning period. After a strategy change or a significant target change, the algorithm spends roughly one to two weeks recalibrating, and performance during that window is noisy by design. Advertisers who panic on day four, change the target again, and restart the clock can keep a campaign in permanent adolescence for months. The working rule: after any meaningful change, wait at least two weeks, or one full conversion cycle if your sales lag is longer, before judging the result.

The second is oversized adjustments. Moving a target CPA or ROAS by 30 or 40 percent in one step forces a dramatic re-evaluation of which auctions qualify, and volume can crater before it recovers. Typical practitioner guidance is to move targets in steps of roughly 10 to 20 percent and let performance settle between moves.

The third, and the most damaging, is optimizing toward the wrong conversion. Smart Bidding is ruthlessly literal: if the campaign’s conversion goal includes page views, newsletter signups, or unqualified form fills alongside real leads, the algorithm will find the cheapest action in the set and flood you with it. Audit which conversion actions are set as primary before blaming the bid strategy, because no strategy can outperform a corrupted goal. And finally, judge changes against the right baseline: seasonality, budget changes, and new competitors all move CPA on their own, so compare like-for-like periods before crediting or blaming the bidding.

Choosing Your Bidding Strategy: A Simple Decision Framework

Pulling it all together, here is the decision path we apply to every campaign we take over, and it resolves most cases in four questions.

First: is conversion tracking accurate and trusted? If not, fix that before touching the bid strategy, running Manual CPC or Maximize Clicks at modest spend in the meantime, because every Smart Bidding strategy amplifies whatever the tracking says. Second: do conversions have meaningfully different values you can measure? If yes, you belong on the value side, Maximize Conversion Value first, Target ROAS once you clear roughly 30 to 50 valued conversions a month. If no, you belong on the count side. Third: how many conversions does the campaign produce in 30 days? Under about 15, run Maximize Conversions with a realistic budget and work on volume. Above roughly 30 to 50, add a Target CPA set near your actual trailing CPA and tighten it gradually. In between, either strategy is defensible; pick based on whether volume or efficiency is the more urgent problem. Fourth: is this a brand defense campaign? Then Target Impression Share is the one place it earns its keep.

The meta-point is that bidding strategy is not a set-and-forget choice but a ladder you climb as your data matures: clicks to conversions, conversions to a CPA target, a CPA target to value, value to a ROAS target, with offline conversion data improving every rung. Climb it at the pace your conversion volume allows, resist the urge to skip steps, and the algorithm becomes the most capable employee in your marketing department. Rush it, and you are paying machine-learning prices for coin-flip decisions. If you would rather have someone manage that ladder for you, this is exactly the kind of work a performance-focused team like SearchPod handles inside a flat monthly plan, but the framework above works the same whether you run it yourself or hand it off.