How to Set Up a Google Ads Campaign: A 2026 Step-by-Step Guide
The Google Ads setup flow asks for about forty decisions before it lets you publish anything. A handful of them shape everything that follows, two cannot be undone, and the interface gives you no indication which are which. Someone building their first campaign usually discovers the weighting afterwards, when a report will not reconcile or a campaign spends a month failing to spend its budget.
Key Takeaways
|
The decisions carry very different costs. A character limit you get wrong costs a rejected ad and ten minutes. A bid strategy you get wrong costs a month of underspend while the algorithm hunts for auctions inside a constraint it cannot satisfy. A permanent account setting you get wrong costs every hour-of-day report you will ever run on that account. The ones that hurt most are the ones that stay invisible from inside the interface, which is why the order you make them in matters.
What follows runs from account creation through the first thirty days, in the order you meet the decisions. Where a number reflects practitioner convention rather than a documented platform rule, it is labelled as such, because the difference matters when you are deciding how closely to follow it. Google’s documentation is linked directly at the points where the specifics are worth checking against the source.
Before You Start: Prerequisites
Gather everything before you open the account. The setup flow asks for decisions you cannot easily reverse, and making them under time pressure is how accounts end up misconfigured for years.
- A Google account dedicated to the business rather than an individual’s personal login.
- Business name, address, and contact details for verification.
- A billing method ready to enter, since campaigns will not serve without it.
- A landing page that is live, fast, and built to receive paid traffic.
- Clearly defined conversion actions, meaning you know whether success is a form fill, a call, a purchase, or a demo request.
- Brand assets, including a square logo, approved imagery, and any copy restrictions your legal team enforces.
When you create the account, switch to Expert Mode rather than accepting the simplified flow. Expert Mode gives you access to the full settings tree, including bid strategy selection and the granular targeting options the guided experience hides.
Two Settings You Cannot Undo
Time zone and currency are permanent. Google’s documentation is unambiguous: to manage ads in a different time zone you must create a new account and set it at creation, and the same applies to currency. There is no support path that changes either on a live account.
The practical damage from getting this wrong is not billing, it is analysis. If the account runs on Pacific time while your sales team works Eastern, every hour-of-day report and every lead-timing comparison is offset, and no amount of downstream reporting work fully corrects it. Pick the time zone that matches the business operation, not the agency’s office.
Define Your Objective and Primary Conversion
Your objective determines which bid strategies are available and what the system optimizes toward. Start from the business outcome rather than from a keyword list.
More important than the objective label is your primary conversion action. This is the specific event Smart Bidding optimizes against, and if it is defined loosely, everything downstream inherits that looseness. A lead-generation account that counts every form submission equally will train the algorithm to find people who fill in forms, which is not the same population as people who buy.
For long sales cycles, this matters more than any setting in the campaign. Bidding toward a form fill in a business where the meaningful event happens six weeks later teaches the system to optimize for the wrong thing. Offline conversion imports from your CRM are the fix, and they are worth setting up before launch rather than after you have trained the algorithm on weak signal.
Choose Your Campaign Type
Two campaign types are worth weighing at setup, and one feature is worth knowing about because its name makes it sound like a third.
Search
Search captures existing demand from people typing queries. You control keywords, match types, and negatives, and you can see what people searched before they clicked. For a first campaign in a new account, this is almost always the right starting point, because it produces interpretable data. When something works or fails, you can tell why.
Performance Max
Performance Max runs across Search, Display, YouTube, Gmail, and Discover from a single campaign, with Google assembling asset combinations automatically. It needs substantially more conversion signal than Search to perform, and it gives you less visibility into what drove results.
Running it as a first campaign in an account with no conversion history is the common expensive mistake. Establish clean tracking and a working Search campaign first, then add Performance Max once there is enough signal for the automation to learn from.
AI Max Is a Feature, Not a Campaign Type
AI Max sits inside a Search campaign rather than alongside one. Google’s developer documentation puts it plainly: AI Max is not a new campaign type, but rather an optimization layer you can enable within an existing Search campaign, described elsewhere as a continuous optimization layer for Search campaigns built from targeting and creative features. You will not find it in the campaign type selector, because it is a switch you flip once a Search campaign already exists.
It bundles search term matching using keywordless technology, text customization, and final URL expansion. Enabling it activates those settings together, though you can switch individual components off. It moved out of beta in April 2026 and is generally available.
The reasonable position is that AI Max is a lever you pull on an existing Search campaign once that campaign has stable conversion data, not a thing you choose instead of Search. Note also that it will not help a campaign that is limited by budget, since it has no room to act.
Budget and Bidding
Two recent changes to the bid strategy menu affect what you can select at setup, so it is worth knowing what is actually there before you go looking for something that is not.
What Actually Exists Now
Enhanced CPC was retired from Search and Display on March 31, 2025, following its earlier removal from Shopping. Standalone Target CPA and Target ROAS have since been consolidated, so new campaigns created in 2026 do not see a standalone tCPA or tROAS option, only the Maximize variants with an optional target field. The underlying bidding behaviour is identical. Only the labelling and the entry point changed, which is worth knowing if you go hunting for a strategy by name and cannot find it.
So the practical menu at setup is Manual CPC, Maximize Clicks, Maximize Conversions with an optional CPA target, Maximize Conversion Value with an optional ROAS target, and Target Impression Share.
Manual CPC is also easier to reach than it was. As of January 2026, it appears directly under the Conversions goal as “Manually set bids”, rather than being buried behind a not-recommended link. It remains a legitimate choice for low-volume accounts and early testing, where you want control and there is not enough conversion data for automation to work with.
Sequencing Targets
Launch unconstrained, then add a target. Starting a new campaign on Maximize Conversions without a CPA target lets the system gather signal; adding a target before it has enough data is a common cause of underspend, because the algorithm cannot find enough auctions inside the cost constraint.
On volume, Google’s recommended floor is 30 conversions in the past 30 days at campaign level before adding a Target CPA, with 50 or more before a Target ROAS. Those are floors rather than comfortable operating levels, and practitioner consensus sits meaningfully higher. Below the floor, stay on Maximize Conversions, or pool campaigns into a portfolio bid strategy so the algorithm learns from aggregate volume rather than from several starved campaigns.
Set targets against what the account has actually achieved, not what you want it to achieve. A target below historical performance produces underspend rather than efficiency.
The August 17, 2026 Change
If you are setting up campaigns this month, plan around this. Google is changing how target-based bidding behaves so that campaigns hold their targets more consistently even when budgets are adjusted, starting August 17, 2026.
The affected population is specific: campaigns that are limited by budget and running a target-based strategy. Those campaigns often currently overperform their targets because the budget, not the target, is the binding constraint. After the change, they will optimize more consistently toward the target, which means a campaign beating its CPA target today may see that CPA rise toward the number you set. Google will not adjust targets or budgets automatically, so the correction is yours to make.
Manual CPC and Target Impression Share are explicitly unaffected. If you are launching a new campaign now on a target-based strategy, set the target deliberately rather than aspirationally, because after August 17 the system will hold you to it more literally than it does today.
Location, Language, and Audiences
Set geographic targeting to match what the business can actually serve, and apply exclusions as deliberately as inclusions. A service business that cannot deliver beyond a two-hour radius should exclude the surrounding region rather than relying on the radius setting alone.
Layer audiences in Observation mode rather than Targeting mode at launch. Observation collects performance data by segment without restricting who can see your ads, which tells you whether a segment is worth acting on before you narrow reach around it.
One caveat shapes how much any of this is worth doing. Under any Smart Bidding strategy, device, location, audience, and ad-schedule bid adjustments are ignored. The only adjustments that still bite are full device exclusions, seasonality adjustments, and data exclusions. So if you are on Smart Bidding, audience layering earns its place as a reporting and exclusion tool rather than as a bidding lever.
Ad Groups, Keywords, and Match Types
Build ad groups around a single intent, not a product catalogue. The reason is message match: an ad group covering one intent can carry ad copy that speaks directly to it, and copy that matches the query earns the click.
On keyword counts per ad group, the widely quoted five to fifteen range is practitioner convention rather than a Google rule, and Black Propeller treats it as a rule of thumb rather than a requirement. The real test is whether every keyword in the group can be served by the same ad copy without the copy becoming generic. If it cannot, split the group.
Match type selection now interacts with your bid strategy. Broad match works when paired with Smart Bidding and clean conversion data, because the algorithm has the signal to sort useful queries from wasteful ones. Broad match on Manual CPC in a new account with no conversion history is how budgets disappear. Phrase and exact carry the load where control matters, including brand terms and regulated categories.
Negatives are not a launch-day task you complete once. Review the search terms report weekly, add negatives for queries that are semantically close but commercially irrelevant, and keep doing it. A software business will need to exclude research and free-tier language early; a premium service will need to exclude bargain language.
Responsive Search Ads and Assets
Responsive search ads are the only text format still accepted. The specification is up to 15 headlines and 4 descriptions per ad, with headlines capped at 30 characters and descriptions at 90. The minimum to publish is 3 headlines and 2 descriptions.
Google recommends filling all 15 headline slots, and its guidance is to run at least 2 RSAs with Good or Excellent Ad Strength per ad group with a unique final URL for each. Ad Strength is a more useful target than any particular headline count, because it measures the diversity of what you have supplied rather than the quantity. Fifteen variations on the same claim will score worse than eight genuinely different ones.
Write the first three headlines as though they will appear together, since they often do. Pin only what genuinely must hold position, such as a legal disclaimer or a brand name, because every pin reduces the combinations available.
Assets
Assets, called extensions until the 2022 rename, expand what your ad occupies on the page and give people more ways into your site. The ones worth configuring at setup:
| Asset | Setup note |
|---|---|
| Sitelinks | Minimum 2 to serve, but add 4 or more; up to 6 can show on desktop and 8 on mobile |
| Callouts | Short non-clickable benefit statements; write them to survive being read in any order |
| Structured snippets | Category lists that qualify the click before it happens |
| Call assets | Direct click-to-call; verify the number and country code or it will disapprove |
| Location assets | For businesses driving physical visits; requires a linked Business Profile |
| Image assets | Square and landscape; 1:1 and 1.91:1 ratios, 300×300 pixels minimum |
Two sitelinks is the serving minimum rather than a target worth aiming at. Up to six can show on desktop and up to eight on mobile, so the number you supply sets a ceiling on how much of the results page your ad can occupy. Google recommends creating sitelinks at account level first for faster coverage, keeping link text short so more links can display, and adding descriptions, which unlock larger formats without reducing how many links show.
Reusing the same link text across sitelinks is not allowed even when each points to a different destination, which is a frequent cause of asset disapprovals.
Conversion Tracking
Nothing else in this guide matters if this step is skipped or done badly. Smart Bidding optimizes toward the conversion actions you define, so a tracking setup that fires inaccurately does not merely misreport, it actively trains the system toward the wrong outcome.
- Install the Google tag across the site, not only on conversion pages.
- Create conversion actions for each meaningful event, and mark one as primary.
- Link Google Ads to GA4 so audience and conversion data flow between them.
- Use Google Tag Manager for event tracking rather than hardcoding each trigger.
- Enable enhanced conversions to recover attribution lost to browser restrictions.
- Apply UTM parameters consistently so cross-channel reporting reconciles.
- Fire a test conversion and confirm it appears in the interface before spending.
Assign conversion values wherever you can, even estimated ones. Value-based bidding needs them, and a lead-generation account that treats a demo request and a newsletter signup as equivalent is telling the algorithm something untrue about the business.
Quality Score: What It Does and Does Not Do
Quality Score is the number people reach for when a campaign costs more than expected, and it is worth understanding precisely, because the shorthand version of how it works points you at the wrong fix.
Google’s own position is that Quality Score is not an input in the ad auction; it is a diagnostic tool for understanding how your ads affect user experience. The 1 to 10 number you see is a summary of past performance on that keyword. What actually operates in the auction are the real-time quality signals underlying it, which is why your reported Quality Score can lag the behaviour it describes.
The practical consequence is unchanged: improving ad relevance, expected click-through rate, and landing page experience does reduce what you pay for a given position, and the component ratings tell you which of the three to fix. But do not treat the number itself as a lever, and do not confuse it with Ad Strength, which is an ad-level diagnostic with no impact on auction performance.
Pre-Launch Review
Work through this before publishing. Every item here represents a failure that is cheap to prevent and expensive to discover a week later.
- Test conversions fire and appear in the interface.
- Auto-tagging is enabled and the GA4 link is confirmed.
- Location targeting and exclusions match what the business can serve.
- Daily budget matches the intended monthly spend, divided correctly.
- Bid strategy matches the account’s actual conversion volume.
- Negative keyword lists are applied at the right level.
- Every final URL loads, matches its ad, and carries correct UTM parameters.
- Ad copy is checked for spelling, claims, and brand compliance.
- Assets are attached at account or campaign level and are approved.
Google’s review step before publish is the last checkpoint, and it catches formatting problems rather than strategic ones. A campaign can pass review cleanly and still be pointed at the wrong country with the wrong budget.
The First 30 Days
New campaigns enter a learning period while the system calibrates, and performance during it is volatile by design. The discipline is to avoid changing things that reset it. Editing targets, pausing ad groups, or moving budgets significantly all restart learning and discard accumulated signal.
Keep any single change to a target or daily budget modest rather than dramatic, and leave several days between adjustments so you can attribute the effect. What to watch, roughly in order of usefulness:
- Search terms report, weekly, for negatives and for what the matching is actually finding.
- Conversion tracking integrity, since tags break silently.
- Impression share lost to budget versus lost to rank, which tells you whether the constraint is money or relevance.
- Device and location performance, for exclusions rather than bid adjustments.
- Ad Strength and asset-level performance for creative iteration.
Resist restructuring in week two. Most early performance noise is the learning period rather than a structural problem, and rebuilding mid-learning guarantees you never find out which it was.
Where Automation Fits
The through-line in every change described above is that Google has moved decisions from the advertiser to the system, and given the advertiser fewer, higher-leverage inputs in exchange. Conversion signal quality, campaign structure, and target setting now matter more than granular bid management, because granular bid management largely no longer exists.
That shifts where an operator adds value. It also explains why third-party optimization layers such as Pixis have become more common alongside in-house account management rather than in place of it, which is the sensible division of labour when the platform’s own automation is opaque about why it did what it did.
Whatever tooling sits on top, the input quality determines the output. Clean conversion data and a defensible account structure are what automation amplifies. Neither is something automation can supply for you.
Frequently Asked Questions
How do I set up a Google Ads campaign for the first time?
Create the account, switch to Expert Mode, and confirm time zone and currency before anything else, since neither can be changed later. Install conversion tracking and verify it fires. Then build a Search campaign with tightly themed ad groups, responsive search ads, and assets, and run the pre-launch review before publishing.
What campaign type should I choose?
Search, for a first campaign in a new account. It produces interpretable data, which matters more early than reach does. Add Performance Max once conversion tracking is proven and there is enough volume for the automation to learn from.
Is AI Max a campaign type?
No. It is an optimization layer enabled inside an existing Search campaign, covering search term matching, text customization, and final URL expansion. It left beta in April 2026. Enable it once a Search campaign has stable conversion data, and note it will not help a campaign that is limited by budget.
Do I need conversion tracking before launching?
Yes. Smart Bidding optimizes toward defined conversion actions, so without tracking there is nothing to optimize toward. Inaccurate tracking is worse than none, because it trains the system toward the wrong outcome while appearing to work.
Because standalone Target CPA and Target ROAS were consolidated into the Maximize variants. Select Maximize Conversions and add an optional CPA target, or Maximize Conversion Value with an optional ROAS target. The bidding behaviour is the same as the old standalone strategies.
What is the best bidding strategy for a new campaign?
Maximize Conversions without a target, until the campaign clears roughly 30 conversions in 30 days. Manual CPC remains reasonable for low-volume accounts or early testing where control matters more than automation. Add a CPA or ROAS target only once there is enough data to support it, and set it against actual performance.
How many headlines should a responsive search ad have?
Up to 15 are accepted and Google recommends using all of them. The minimum to publish is 3 headlines and 2 descriptions. Aim for Good or Excellent Ad Strength across at least two RSAs per ad group rather than targeting a specific headline count.
What changes on August 17, 2026?
Budget-limited campaigns on target-based bid strategies will hold their targets more consistently. Campaigns currently overperforming because budget is the binding constraint may see performance move toward the target you set. Google will not adjust targets or budgets for you. Manual CPC and Target Impression Share are unaffected.
Getting Help
If you would rather have this built and managed by a team that does it daily, Black Propeller’s paid search management and Google Ads management services cover account structure, conversion infrastructure, and ongoing optimization.
