Google Ads Bidding Strategies: Which One to Use
Every guide lists the strategies. Almost none list the conditions each one needs — which is the only part that decides whether it will work in your account.
Bidding is where most accounts get switched to something clever too early.
Target ROAS sounds like the sophisticated choice. Manual CPC sounds like something you have outgrown.
But bid strategies are not a ladder you climb. They are tools with entry requirements, and using one before you meet them produces worse results than the "basic" option you skipped.
So here is each one with what it actually needs.
What bidding actually decides
Not whether you appear — that is your keywords and match types, covered in Google Search Ads.
Bidding decides how much you are willing to pay for a given auction, and which auctions are worth entering at all. Automated strategies make that decision per auction using signals you cannot see or set manually: device, time, location, browsing history, query specifics.
That is a genuine advantage — but only when the system has enough of your outcomes to learn what a good auction looks like.
The strategies, and the conditions each needs
| Strategy | Optimises for | Needs | Use when |
|---|---|---|---|
| Manual CPC | Whatever you tell it | Nothing but your attention | New account, low volume, or you need hard cost control |
| Maximise clicks | Volume of clicks | A budget cap so it cannot run away | Early on, to build traffic and discover search terms |
| Maximise conversions | Number of conversions | Working tracking; some conversion history | You want volume and do not yet have a firm cost target |
| Target CPA | A cost per conversion | Working tracking; steady conversion volume | Lead generation with a known acceptable cost per lead |
| Target ROAS | Revenue relative to spend | All of the above, plus accurate conversion values | E-commerce where your revenue values are genuinely correct |
| Maximise conversion value | Total revenue | Accurate values; no fixed efficiency target | You want revenue growth and can accept variable efficiency |
| Target impression share | Position on the page | A reason that is not vanity | Brand defence, or a competitor bidding on your name |
Read the "needs" column before the "use when" column. Most bad bidding decisions come from picking a row by ambition rather than by eligibility.
Conversion volume is the gatekeeper
This is the honest version of the "how much data do I need" question.
Google has published thresholds over the years and has softened them since, so any hard number quoted as fact would be out of date. The useful test is behavioural rather than numerical:
Is this campaign producing conversions most days, or a handful a month?
Most days means there are patterns to find, and automation will find them faster than you. A handful a month means you are asking an optimisation system to detect signal in noise — and it will confidently optimise toward whatever randomness it saw.
Tracking has to be right before any of this matters
Automated bidding optimises toward the conversion event you defined. If that event is wrong, it will pursue the wrong outcome with total commitment.
The failures we see most:
- Counting every form view as a conversion — the system learns to find people who load pages.
- Double-counting across GA4 import and the Google Ads tag, so performance looks twice as good as it is.
- Optimising for a micro-conversion like a newsletter signup while expecting revenue.
- Conversion values left at zero or set to a flat placeholder, which makes target ROAS meaningless.
- Tracking that broke months ago and nobody checked, so the strategy is optimising against a frozen picture.
Verify with a real test transaction before changing anything. This is the same prerequisite Performance Max has, for the same reason.
How to switch without a crash
- One campaign at a time. Switching the whole account removes your ability to tell what caused what.
- Set the first target at what you already achieve — not at what you want. An aggressive first target throttles delivery and the campaign starves.
- Expect a dip. Every switch restarts a learning period. Performance gets worse before it settles, and that is normal rather than a sign to revert.
- Do not touch it for two weeks. Editing mid-learning restarts the clock, exactly as it does on Meta.
- Then tighten in small steps — move the target by modest increments and let each one settle before the next.
- Keep a control. Leave one comparable campaign on the old strategy so you have something honest to compare against.
When manual is still the right answer
Manual CPC is not obsolete. It is correct when there is not enough data for automation to be smarter than you.
- A brand new account with no conversion history at all.
- Genuinely low-volume campaigns — a handful of high-value conversions a month.
- Hard cost ceilings where a click above a certain price is never acceptable.
- Diagnostic periods where you need to isolate whether a problem is bidding or something else.
The right progression is usually: manual or maximise clicks to build data → maximise conversions once tracking is solid → target CPA once volume is steady → target ROAS only if your conversion values are genuinely accurate. Skipping steps is what causes the crashes people blame on automation.
Setting the target itself
For target CPA, work backwards from what a customer is worth and what proportion of that you can spend acquiring one.
For target ROAS, work from margin. A 4× ROAS is excellent at one margin and loss-making at another, which is why a good ROAS is a business calculation rather than a benchmark you can borrow. In Shopping especially, where price is visible in the ad, that margin pressure is the whole game.
Frequently asked questions
Which Google Ads bidding strategy is best?
There is no best one — there is only the one your account currently has the conditions for. Smart bidding needs verified conversion tracking and steady conversion volume. Without both, manual or maximise clicks will outperform it, because automation cannot optimise against a signal that is not there.
How much conversion data does smart bidding need?
Enough for patterns to exist rather than a fixed number. As a working rule, if a campaign is producing conversions most days rather than a handful a month, automated bidding has something to learn from. Below that, you are asking an optimisation system to find signal in noise.
What is the difference between target CPA and target ROAS?
Target CPA optimises toward a cost per conversion and treats every conversion as equal. Target ROAS optimises toward revenue relative to spend and needs accurate conversion values to work. Use target CPA for lead generation, and target ROAS only for e-commerce where your values are genuinely correct.
How do I switch bidding strategy without tanking performance?
Change one campaign at a time, set your first target close to what the account already achieves rather than to what you want, and leave it alone for a couple of weeks. Every switch restarts a learning period, so performance dips before it settles. Switching everything at once removes your ability to tell what caused what.
Is manual CPC still worth using?
Yes, in specific cases. New accounts with no conversion history, very low volume campaigns, and situations where you need hard control over what a click can cost are all legitimate reasons. Manual is not outdated — it is the right tool when there is not enough data for automation to be smarter than you.
We fix the signal before we change the strategy
Conversion tracking verified end to end, targets set from your margin, and migrations run one campaign at a time with a control left standing.