There is a moment near the start of every Google Ads account that nobody treats as a decision.
You have picked your goal. You have typed in your website. And Google shows you a screen of campaign types, laid out in a neat grid with an icon and a cheerful line of description under each one. Search. Performance Max. Display. Shopping. Video. Demand Gen. Smart. App.
They are presented as equals. Same size tiles, same tone of voice, same implied promise that you are choosing a flavor rather than committing your entire budget to a machine with requirements you have not been told about.
Some owners never even see the full grid. Google's own documentation is unusually candid about this: if your objective is Sales, Leads, or Local store visits, Performance Max shows up as a campaign type selection, and Google's help page adds that if Performance Max is "the only option appearing when you want to create a different campaign," you need to go back and pick "create a campaign without a goal's guidance" to reveal all available campaign types.
Read that again. The documented way to see your actual options is to tell Google you have no goal.
That is not a bug and it is not a conspiracy. It is a product decision that reflects what Google would prefer you to run. But it means the first real choice in your account is one many owners never knowingly make.
Google Shows You a Menu. It Does Not Tell You the Prices.
The tiles all look like features. They are not. Each one is a request.
Every campaign type needs something from you before it can work, and the types differ enormously in what they ask for. That difference is invisible on the selection screen, and it is the single most important thing about that screen.
Search wants a keyword. Display wants a guess about who your audience is, plus images sized correctly. Video wants a video you have already made and uploaded to YouTube. Performance Max wants conversion history, creative assets across several formats, and audience signals.
Google states these requirements plainly, just never in one place and never side by side. On Search, the help page says the setup is easy because you "write text ads and pick keywords" and "skip special files or assets." On Display, the same family of pages tells you it is helpful to use an image editing tool to make square and landscape versions of your ad at the right dimensions. On Video, the tip is blunter: you will first need to create your video ad and upload it to YouTube. On Shopping, you will first need a Merchant Center account with your product inventory loaded.
Line those up and the grid stops looking like a menu of equals.
The Two Things a Small Account Cannot Manufacture
Sort those requests and almost all of them collapse into two categories.
The first is conversion data. Automated campaign types do not target; they predict. Prediction needs history. A campaign type that decides for itself who to show your ad to, on which surface, at what bid, is building a model, and it builds that model out of your conversions.
The second is ad creative. Video, images at multiple aspect ratios, several headline and description variants. Not a stock photo and a logo. Assets good enough to represent your business to a stranger who was not looking for you.
Both cost money. Not ad budget, different money. Conversion data costs time and spend you have not made yet. Creative costs a videographer, an editor, a designer, or the weeks of your own life you spend failing to be one.
This is the whole decision rule, and everything below is just applying it:
A campaign type that demands data or creative you do not yet have will not perform. It will spend.
That distinction matters. These campaigns do not error out or refuse to run. They launch cheerfully, serve impressions, generate clicks, and quietly produce nothing, because the thing they needed to be good was never supplied. Nothing in the interface tells you this is happening. You find out at the end of the month.
What $2,500 a Month Actually Buys You
Run the numbers, because they settle the argument faster than any opinion.
Take $2,500 a month in a typical small business search market at around $8 a click. That is a little over 300 clicks a month. Convert those at a respectable five or six percent and you have somewhere in the region of 15 to 19 conversions a month.
Now hold that against what automated bidding needs to learn. The working figure practitioners use, and the one our own account experience keeps confirming, is somewhere around 15 to 30 conversions per campaign per month before a bidding model has enough evidence to be worth trusting.
So $2,500 a month buys you approximately one campaign's worth of learning, and only just. One.
That is the sentence to sit with, because it reframes the entire selection screen. The question was never "which of these eight is best." Almost any of them can work for somebody. The question is which single one you can afford to feed, because you are not choosing between eight options. You are choosing where to put the only decent pile of evidence you are going to generate this quarter.
Split that $2,500 across a Search campaign and a Display campaign and you have not diversified. You have built two half-starved models instead of one adequately fed one. We wrote about the structural version of this in You Organized Your Google Ads Account. You Starved It., and the same arithmetic governs campaign types as governs campaign counts. The unit of starvation is just bigger here.
Search Is the Only Type That Works Before You Have Data
Here is why Search sits alone at the bottom of the requirement ladder, and why practically every experienced practitioner tells small budgets to start there.
A keyword is not a prediction. It is an instruction.
When you add the keyword "emergency plumber Houston," you have not asked Google to work out who wants a plumber. You have told it. The targeting decision was made by you, in advance, using knowledge about your business that no amount of machine learning was going to derive from sixteen conversions. Google's only remaining job is to match a search to your instruction and run the auction.
This is the property that makes Search survivable on a small budget. It is the one type where the account works on day one, before any conversion history exists at all, because the intelligence came from you rather than from your data.
Everything else on that grid asks Google to figure out the targeting. That is a reasonable thing to ask when you have thousands of conversions. It is an unreasonable thing to ask when you have sixteen.
There is a second, quieter advantage. Search is the only major type that needs no produced creative. Text, a landing page, and the keyword. For a business without a creative budget, that is not a minor convenience. It is the difference between launching this week and launching never.
Shopping Is the Exception, and Only If You Sell Products
One type genuinely joins Search at the low end of the ladder, and it is worth knowing about because a lot of small retailers are told to skip it.
Shopping works on a small budget for the same structural reason Search does: the targeting instruction comes from you. Your product feed is the instruction. Titles, attributes, categories and prices tell Google what each item is and which searches it belongs in, which is precisely the job a keyword does in a Search campaign. We took that apart in Google Shopping Ads Have No Keywords. Your Product Feed Is the Campaign.
The requirement is real but it is a one-off: a Merchant Center account with your inventory properly loaded. That is work, and for a lot of stores it is unglamorous work. It is not, however, an ongoing creative production cost, and it is not a demand for conversion history you do not have.
If you sell physical products, Shopping belongs in the same conversation as Search. If you sell services, it does not exist for you, and the conversation is shorter.
Performance Max Is Not a Beginner Campaign, However It Is Sold
Performance Max is the type most small advertisers get pushed toward, and the reasoning offered is almost always the same: it is simple to set up, and Google's AI does the hard part.
Both halves of that are true. Neither is a reason for a small budget to start there.
Performance Max is the most automated type on the grid, and automation is not a substitute for data. It is a consumer of data. Handing the targeting, bidding, placement and creative assembly to a model does not remove the need for evidence, it concentrates the need. A campaign that makes more decisions on your behalf needs more history to make them from, not less.
The people most enthusiastic about Performance Max tend to agree with this, which is the part worth noticing. Ben Heath, whose small-budget material is among the most useful published anywhere, describes himself as a big fan of Performance Max and still puts "stick to Search campaigns" as his first recommendation for small budgets, framing Performance Max as the thing you expand into once you have reached a certain level of scale. That is not a rejection. It is a sequencing argument, and sequencing is the whole game when your evidence is scarce.
There is a narrower version worth flagging honestly, because the creative objection has an exception. Google's documentation notes that a Performance Max campaign running purely on a Merchant Center feed does not require you to supply creative assets, though to keep it feed-only you have to skip adding text, image and video assets entirely during setup, and once manual assets are added they cannot be removed later to get back to that format. So the creative barrier can come down for retailers. The data barrier does not.
We went further into what that campaign type is actually built to do in Performance Max Isn't Your Smartest Campaign. It's Google's.
The Ones You Can Refuse Without Guilt
That leaves the rest of the grid. Each gets one honest sentence, because each deserves one and not more.
Display. Enormous reach, no intent. Nobody browsing a recipe site is looking for you, so you are paying to interrupt rather than to answer, and interruption needs both creative that earns attention and volume that a small budget cannot buy.
Video. Same intent problem, plus the highest production cost on the board. A video ad that works is a real piece of film-making, and a video ad that does not work still costs you the money you spent making it.
Demand Gen. Built for the top of the funnel, which is the part of the funnel a small budget should deliberately ignore. Demand creation is a strategy for advertisers who have already saturated demand capture. You have not.
Smart. The one that looks purpose-built for you, and the one to be most careful with. It asks for almost nothing and gives back the least control: you supply a few keyword themes and a budget, and Google places the ads across Search, Maps, YouTube, partner sites and the Display Network on your behalf. That is the whole Display objection above, applied automatically, with less visibility into where the money went. It also lives in Smart Mode, a separate simplified version of the account, so choosing it quietly forecloses most of the controls the rest of this article assumes you have.
App. Only if you have an app. If you do not, the tile is not a decision you need to make.
None of these are bad campaign types. Display and Video do things Search structurally cannot, and there is a point in a growing account where refusing them becomes the expensive choice. That point is not now.
Google Will Keep Asking
Having made this decision, expect to be asked to unmake it roughly once a month.
It will arrive as a recommendation in your account, as a drop in your optimization score, or as a phone call from a Google representative with a specific suggestion about expanding into additional campaign types or turning on additional networks. The suggestions are not random and they are not malicious. They are the product working as intended, and the intent is not identical to yours.
The useful stance is neither compliance nor blanket refusal. It is knowing which decision is being reopened and whether the thing that would justify changing it has actually happened. We laid out how to read that pressure in A 100% Optimization Score Just Means You Agreed With Google.
What Has to Be True Before You Add the Second Type
Graduating off a single campaign type is not a date on a calendar and it is not a budget number. It is a set of conditions, and you will know when they arrive.
Your Search campaign is consistently profitable. Not breaking even, not promising. Producing customers at a cost you would happily pay again tomorrow, sustained across more than one month.
You are producing enough conversions for a model to learn from, with room to spare. If your one campaign is still hovering at the bottom of that 15 to 30 range, a second campaign does not add reach. It halves the evidence.
You are limited by impression share rather than by ideas. If you are already winning most of the auctions your keywords are eligible for and the money has nowhere left to go, that is genuine expansion pressure. If you still have obvious keywords unfunded, the cheaper growth is still inside the campaign you have.
You can produce the creative the next type needs. Not intend to. Have it, or have the budget to commission it.
When all four are true, expand. Until they are, adding a campaign type is not diversification. It is division.
The Order Nobody Gives You
So the grid is not eight options. It is a sequence, and almost every small account should walk it in the same direction.
Start with Search. Add Shopping alongside it if you sell products and your feed is in order. Move into Performance Max when your Search campaign is profitable and generating enough conversions that the model has something real to learn from. Consider Display, Video and Demand Gen when you have both the volume to feed them and the creative to fill them, which for most businesses is considerably later than they expect and considerably less urgent than Google suggests.
The instinct that gets small accounts into trouble is not laziness. It is ambition, expressed as breadth. Running five campaign types feels like taking the channel seriously. It is the most reliable way we know to spend twenty-five hundred dollars a month and learn nothing from any of it.
One type, funded properly, generating enough evidence to actually improve, beats a full grid of starved experiments every single month. The account that looks the least impressive on the campaigns screen is very often the one making money.
If you would rather not spend a quarter finding that out by experiment, that is a reasonable position, and it is most of what we do. Our Google Ads management work usually starts by turning a busy account into a boring one, and the boring version is where the profit tends to be hiding.
