{"id":4250,"date":"2026-07-27T22:57:28","date_gmt":"2026-07-27T19:57:28","guid":{"rendered":"https:\/\/ivitskiy.com\/blog\/google-ads-bid-strategies\/"},"modified":"2026-07-28T15:45:31","modified_gmt":"2026-07-28T12:45:31","slug":"google-ads-bid-strategies","status":"publish","type":"post","link":"https:\/\/ivitskiy.com\/blog\/en\/google-ads-bid-strategies\/","title":{"rendered":"Google Ads bid strategies: why the target you typed is not a spending ceiling"},"content":{"rendered":"\n
In short:<\/strong> most advertisers pick a bid strategy once and never revisit it. A bid strategy is the rule Google follows when it places a bid on your behalf in each individual auction. The thing worth knowing before you choose: a target cost per conversion is a goal, not a spending ceiling. Across the campaigns I have reviewed, only 34% with a target set stayed within plus or minus 10% of it, and the median campaign came in 10% above target. Below: a plain map of the strategies, that data in full, Google’s official conversion thresholds, a decision tree for your data volume, and separately what changes inside the automated strategies on 17 August 2026.<\/div>\n\n\n\n

You open campaign settings, see the list of strategies, pick the one Google suggests by default, and never go back. Or the other version, just as common: you never open the account at all, an agency typed the target in, and all you see is a report where a lead somehow costs more than you agreed. Both stories end the same way.<\/p>\n\n\n\n

I looked at a slice of 32 real accounts I have reviewed: the median advertiser keeps 89.7% of budget on a single bid strategy, and 15 of the 32 hand one algorithm more than 90% of the money. The numbers show where money sits, not why. But when I ask owners why that particular strategy, there is usually no answer: the interface picked it at launch and nobody went back in.<\/p>\n\n\n\n

In order: what is actually on the list, what my data says about how accurate automated bidding is, what Google officially requires in conversions, how to pick yours from a decision tree, and what changes in August.<\/p>\n\n\n\n

The map: what you are actually choosing<\/h2>\n\n\n\n

Google ads are sold at auction: every time, the system decides whether to show your ad to a specific person and what to bid on your behalf. You pay for the outcome your campaign is built around, most often a click. A bid strategy is who makes the bidding decision and on what criteria.<\/p>\n\n\n\n

The map below covers Search and standard campaigns where you pay per click. Video and reach campaigns have their own strategies with their own payment logic and are not in this map.<\/p>\n\n\n\n

Manual.<\/strong> You set the maximum cost per click yourself, keyword by keyword. It is a maximum: the actual cost per click usually lands lower, because at auction you only pay what it takes to edge past the next advertiser. Full control, full responsibility, a lot of routine.<\/p>\n\n\n\n

Volume strategies.<\/strong> Maximize Clicks and Target Impression Share. You are telling the system «bring me more traffic» or «keep me at the top of the page». Leads and money are not part of the conversation: the system honestly does what you asked and promises no conversions at all.<\/p>\n\n\n\n

Outcome strategies (this is what Smart Bidding<\/a> means).<\/strong> Maximize Conversions, Target CPA, Maximize Conversion Value, Target ROAS. Here the algorithm optimizes toward your conversions, on one condition: you pass it the conversion data, and you pass it correctly. If measurement is broken, a smart strategy will optimize toward noise, and it will do so very diligently.<\/p>\n\n\n\n

\n
Strategy<\/th>What you are asking the system for<\/th>Who it suits<\/th><\/tr><\/thead>
Manual CPC<\/td>«Never pay more per click than the maximum I set»<\/td>Launch without data, narrow niches, tight control<\/td><\/tr>\n
Maximize Clicks<\/td>«Bring the most traffic the budget allows»<\/td>Collecting first data, always with a CPC limit<\/td><\/tr>\n
Target Impression Share<\/td>«Keep me at the top of the page»<\/td>Defending brand terms, visibility goals<\/td><\/tr>\n
Maximize Conversions<\/td>«Squeeze the most leads out of the daily budget»<\/td>You have conversions but no stable cost yet<\/td><\/tr>\n
Target CPA<\/td>«Hold the average cost per lead near this number»<\/td>The main working mode for service businesses<\/td><\/tr>\n
Maximize Conversion Value<\/td>«Maximize the value I send you, not the count of actions»<\/td>Advertisers who pass Google the real value of each order<\/td><\/tr>\n
Target ROAS<\/td>«Hold the return per dollar near this bar»<\/td>Online stores with correct order values<\/td><\/tr>\n<\/tbody><\/table><\/figure>\n\n\n\n

The last two rows hide the trap that costs online stores money. The algorithm optimizes toward exactly the value you send it. Send order revenue and it chases revenue, not profit, and on products with different margins those are two different things. If you want it to chase profit, pass profit as the conversion value.<\/p>\n\n\n\n

One more confusion worth clearing immediately. Maximize Conversions and Target CPA are not one strategy with a checkbox, they are two different contracts. In the first you name no price at all, and the strategy is designed to spend the full daily budget<\/a>, as Google’s own help page puts it. In the second you name a number. That number is where it gets interesting.<\/p>\n\n\n\n

The promise corridor: what 1,950 campaigns show<\/h2>\n\n\n\n

You set a target cost per conversion and expect the actual number to land near it. I took campaigns from accounts I have reviewed where a target was genuinely set, and compared the actual cost per conversion over the period against the target that stood in the campaign at the time of the export.<\/p>\n\n\n\n

34%<\/span>of campaigns with a target cost per conversion stayed within plus or minus 10%<\/strong> of their own target. Another 36%<\/strong> ran more than 20% above it. Median ratio of actual to target: 1.10<\/strong>, meaning the typical campaign came in 10% over.<\/span>\n
My slice of accounts I have reviewed: 1,950 campaigns with a target set, $42.2M in spend, 7 categories. September 2024 to February 2025. Method and limits at the end of this article.<\/div><\/div>\n\n\n\n

Read that again. Not «the algorithm sometimes misses» but only one campaign in three stayed anywhere near its own target. And formally the system broke nothing: in Google Ads a target cost per conversion is an optimization goal, not a spending limit. A thermostat holds the average temperature around the number you set and never promises the room will not get warmer than that. A circuit breaker cuts the power at a threshold. You thought you were setting a circuit breaker.<\/p>\n\n\n\n

Hence the name I use with clients: the promise corridor<\/strong>. The number you typed is not a ceiling, it is a reference point inside a corridor. Two figures from my slice describe the width of it honestly: the median campaign runs 10% over target, and roughly one in three runs more than 20% over. One practical conclusion follows: build your economics not around the target, but around what happens if your campaign lands in that upper third.<\/p>\n\n\n\n

Target ROAS behaves differently, and it misses in the opposite direction. Only 27% of campaigns reach or beat their return target, with a median of 86% of the bar. The typical campaign falls roughly 14% short of the return it promised.<\/p>\n\n\n\n

Share of campaigns that meet their own target<\/strong>\n
Cost per lead target<\/span><\/span><\/span>34%<\/span><\/div>\n
Return target<\/span><\/span><\/span>27%<\/span><\/div>\n
Scale 0 to 100%. The criteria differ and are not directly comparable: for cost per lead it is landing within plus or minus 10% of target (1,950 campaigns, $42.2M), for return it is meeting or beating target (622 campaigns, $16.9M). September 2024 to February 2025.<\/div><\/div>\n\n\n\n

And let me immediately rule out a conclusion these numbers do NOT support: «turn automation off». In accounts spending over a million dollars, 72.7% of spend runs on target based strategies, 3.6% on manual and enhanced CPC, and 1.5% on targetless Maximize Conversions. That is an observed distribution of money, and it says exactly one thing: where budgets are large, manual bidding stopped being the main mode a long time ago.<\/p>\n\n\n\n

What changes on 17 August 2026<\/h2>\n\n\n\n

Everything above describes how the system behaved before August 2026. Here is news worth factoring in before you change anything in the account.<\/p>\n\n\n\n

Google has announced that from 17 August 2026, campaigns that are limited by budget and run on a target based bid strategy will perform more consistently toward the target you set<\/a>, including when you adjust budgets. The exact wording is «will more consistently perform toward your bid target».<\/p>\n\n\n\n

What that means in practice. Previously a budget constrained campaign could beat its own target noticeably, and the owner got used to that actual number. After 17 August those campaigns get pulled toward the target instead. If your target sits high «just in case» while you have been getting cheaper leads for years, that gap may disappear. One thing worth checking now: whether your campaign is limited by budget, and whether it carries a target you have never revisited.<\/p>\n\n\n\n

How many conversions you need: Google’s official thresholds<\/h2>\n\n\n\n

The most repeated piece of advice is «you need at least 15 conversions a month, ideally ten a day». I repeated something like it for years myself. The problem is that no single threshold exists: Google sets different requirements per strategy and even per campaign type.<\/p>\n\n\n\n

\n
Strategy and campaign type<\/th>What Google officially requires<\/th><\/tr><\/thead>
Target CPA<\/td>No conversion history required at all, you can start from zero<\/td><\/tr>\n
Maximize Conversions<\/td>No hard requirement; recommended baseline is at least 15 conversions in 30 days<\/td><\/tr>\n
Target ROAS, Search and Shopping<\/td>At least 15 conversions in the past 30 days<\/td><\/tr>\n
Target ROAS, Display<\/td>At least 15 conversions with valid values in 30 days across all campaigns combined<\/td><\/tr>\n
Target ROAS, Video Action<\/td>At least 30 conversions in the past 30 days<\/td><\/tr>\n
Target ROAS, Demand Gen<\/td>At least 50 conversions in 35 days, 10 of them in the past 7<\/td><\/tr>\n<\/tbody><\/table><\/figure>\n\n\n\n
Source: Google Ads Help pages on Target CPA, Target ROAS and Maximize Conversions. Verified 27 July 2026.<\/div>\n\n\n\n

Look at the first row, it is the counterintuitive one. Google states plainly that advertisers can start using Target CPA with no conversion history<\/a>. Formally nothing stops you from switching it on day one. But the same page recommends measuring performance over at least 30 days including at least 30 conversions. Those are two different numbers answering two different questions: can I switch it on, and what can I draw a conclusion from.<\/p>\n\n\n\n

The practical takeaway is not «switch it on immediately». If you are still counting conversions in single digits, you can technically run a smart strategy, but you will have no reliable basis for a verdict for another month, and you pay for the algorithm’s learning the whole time. Google itself recommends in that case to build traffic and conversion data on Maximize Clicks first<\/a>.<\/p>\n\n\n\n

The decision tree: which strategy to run right now<\/h2>\n\n\n\n

This tree applies to Search campaigns chasing leads: services, B2B, local business. Shopping, Video and Demand Gen follow different logic. The thresholds in the branches are not Google requirements (you can see above that the official ones differ), they are my risk marker: the data volume at which I am personally willing to hand bidding to the algorithm without paying more for its learning than the learning is worth.<\/p>\n\n\n\n

\n
0<\/span>Conversions are not set up, or they are passed with errors<\/strong>\n

No outcome strategy works here: it will optimize toward garbage. Measurement first, automation second. This is not a step you can skip.<\/p><\/div>\n

1<\/span>Fewer than 15 conversions in 30 days<\/strong>\n

Maximize Clicks with a mandatory CPC limit, or manual bidding. Formally some strategies start from zero history, but the point of this stage is not leads, it is data: which queries bring enquiries and what a click costs in your niche.<\/p><\/div>\n

2<\/span>15 to 30 conversions in 30 days, cost per lead still swinging<\/strong>\n

Maximize Conversions without a target. Remember the strategy is designed to spend the entire daily budget, so the budget is your circuit breaker here.<\/p><\/div>\n

3<\/span>Over 30 conversions in 30 days, cost per lead is stable<\/strong>\n

Target CPA. Set the number you can survive in the upper third of the corridor, not the number you would like. A target set too low cuts volume: the system simply will not find auctions where it is achievable.<\/p><\/div>\n

4<\/span>Orders differ in value and you actually pass that value<\/strong>\n

Maximize Conversion Value, then Target ROAS. The condition is strict: value must be passed correctly, otherwise the algorithm optimizes toward invented numbers.<\/p><\/div>\n

5<\/span>Brand terms and defending your own name<\/strong>\n

Target Impression Share in a separate campaign. Do not mix brand defence with selling services in one campaign: different goals, different economics.<\/p><\/div>\n<\/div>\n\n\n\n

One rule sits above the whole tree: one campaign, one change, one evaluation window. If you switch the strategy, raise the budget and rewrite the ads on the same day, you will never learn which move worked, and you will reset learning for certain.<\/p>\n\n\n\n\n\n\n\n

What happens right after you switch<\/h2>\n\n\n\n

The most expensive mistake after a strategy change is not made at the moment of the change, it is made on day five. Numbers dip, nerves go, the advertiser switches the old strategy back on and restarts learning all over again. Here is how it looks in the official documentation.<\/p>\n\n\n\n

How long learning lasts<\/h3>\n\n\n\n

Google writes that it can take up to 3 weeks or 1 to 2 conversion cycles for the bid strategy to calibrate to the new objective<\/a>, and it can be faster when there is plenty of data. A conversion cycle is the typical time from click to enquiry in your business: a few days for services, noticeably longer for expensive purchases. There is an important footnote on the same page: the algorithms keep learning even after the «Learning» status has disappeared from the interface.<\/p>\n\n\n\n

What actually restarts learning<\/h3>\n\n\n\n

Per Google’s documentation, four things trigger «Learning» status: a new or reactivated strategy, a change to its settings, a composition change, and in some cases an ad group target change in Shopping campaigns. «Composition change» sounds abstract, so here it is as an action: it is when you or your agency add or remove campaigns from a portfolio strategy, create a new ad group, or clean out two hundred irrelevant keywords in a single day. Formally you did a good thing, in practice you restarted learning. That is why big clean-ups are better done in one deliberate pass before a planned restart, rather than in small doses every Tuesday.<\/p>\n\n\n\n

How much and how often you can change the target<\/h3>\n\n\n\n

One popular piece of advice is worth dispelling here. There is no rule in Google’s documentation saying «do not change the target by more than 20% at a time», I checked. What is there is different: Smart Bidding reacts to target changes within minutes, but reaching the new target takes 1 to 2 conversion cycles<\/a> because of conversion delay. And there is an explicit warning: avoid making multiple ROAS target changes within a single conversion cycle. The unit of caution here is time, not a percentage.<\/p>\n\n\n\n

Where Performance Max fits<\/h2>\n\n\n\n

Performance Max is often presented as a separate bid strategy. Not quite: it is a campaign type that runs on the same target based strategies but picks the Google surfaces to appear on by itself. Per the documentation that means YouTube, Display, Search, Discover, Gmail and Maps<\/a> at once.<\/p>\n\n\n\n

The most common question about PMax is whether it will eat the brand traffic you already pay for in your Search campaign. The official answer is specific: if a query matches an exact match keyword in your Search campaign, Google prioritizes the Search campaign over Performance Max. But the same page carries an honest caveat, and it is worth quoting: your PMax campaign may sometimes show for branded keywords even on exact match «under certain conditions, such as when your Search campaign has a limited budget or uses stricter targeting». Brand exclusions and campaign level negative keywords exist for exactly that case.<\/p>\n\n\n\n

In practice: if you are still building the system and do not yet have a steady flow of leads from Search, PMax is not step one. Managed Search with a bid strategy you understand comes first, expansion second.<\/p>\n\n\n\n

What happens to your $10 target<\/h2>\n\n\n\n

Instead of restating the above, let me run one example all the way through, because that is where the pieces click together.<\/p>\n\n\n\n

A workshop owner wants leads at $10 and types 10 into the target. Meanwhile: he gets 22 conversions a month, which puts him in branch 2 of the tree, not branch 3, so a target is premature. He sets it anyway. For the first two weeks the campaign is learning, leads get more expensive, his nerve goes and he switches Maximize Clicks back on. Learning restarts, another two weeks go into calibrating back. A month later he concludes «automation does not work», having never once let the system reach a stable state and never once evaluated results over a window that means anything.<\/p>\n\n\n\n

Now the same owner, played differently. He works out his break-even: a lead above $13 makes the advertising unprofitable. He sets the target at $10, with headroom, knowing the median campaign runs 10% over and a third run 20% over or more. He spends the first month on Maximize Clicks with a limit, collects up to 35 conversions, and only then switches the target on. Then he touches nothing for three weeks. When the actual cost lands at $11, he does not panic: $11 is inside his corridor and inside profitability. Both stories are about the same number, 10. The difference is that in the second one it was calculated from break-even rather than from hope.<\/p>\n\n\n\n