Most underperforming Google Ads accounts I audit do not have a traffic problem. They have a separation problem. Everything is thrown into one or two campaigns, the algorithm is left to sort it out, and the reporting looks fine right up until you realise a large slice of the revenue would have happened anyway.
The structure below fixes that. It is built on four layers, and each layer exists to do one job. Once you understand what each layer is for, the settings stop feeling like arbitrary best practice and start feeling obvious.
The principle everything else hangs off
All paid search traffic sits somewhere on a scale of intent. At one end is somebody typing your brand name into Google, who has already decided to buy from you. At the other end is somebody typing a broad category term who may not have heard of you and may not buy today. Those two people are worth wildly different amounts to your business, and they should not be paid for at the same price.
Google's automated campaign types will not make that distinction for you. Left alone, they will spend your budget wherever conversions are cheapest to find, which almost always means the traffic that was going to convert regardless. Your job as the advertiser is to separate the traffic by intent, then set a different level of aggression against each segment. That is genuinely all this structure is doing. Four layers, sorted by intent, each priced according to how much incremental value it actually creates.
Layer one: Performance Max
Performance Max is your broad reach new customer engine. It is the layer with the widest net and the least manual control, so the discipline is in how you contain it.
Start by deciding how many Performance Max campaigns you need, because the answer is usually more than one. If you sell products with very different margins, break them out into separate campaigns so each one can carry its own ROAS target. A single blended target across a mixed catalogue will quietly produce sales that are nowhere near profitable, and because the campaign level ROAS still looks acceptable, nothing in the interface will tell you it is happening. Margin bands get their own campaigns for the same reason a business does not price every product the same way.
Next, run it feed only. No headlines, no descriptions, no images, no video, nothing. When you add assets, you open the campaign up to display and video inventory, and your budget starts drifting away from the placements that actually close sales. Feed only keeps the spend concentrated in high intent shopping placements, and that is where the return lives.
Then exclude brand, on every Performance Max campaign you run. This is the single change that most often turns a mediocre account around. Performance Max is designed to find the cheapest conversions available, and the cheapest conversions in any account are people already searching for you by name. Left uncontrolled, the campaign will lean on your brand terms, report a beautiful ROAS, and charge you for customers who were walking through the door anyway. Excluding brand forces the campaign to earn its keep on genuinely new demand, and the reported numbers get worse before they get honest. Expect that, and judge the change on total new customer volume rather than on campaign ROAS.
Layer two: standard shopping
This is the layer most advertisers skip. If Performance Max is already running feed only, why bother with standard shopping at all? Because standard shopping gives you priority levels in the campaign settings, and Performance Max does not. Those priority levels are what let you build something we call a keyword funnel, and it is the closest thing to real control you have over shopping traffic.
Priority levels are worth understanding properly, because they are the mechanism that makes this whole layer work. When the same product sits in more than one shopping campaign, Google does not pick the campaign with the highest bid. It picks the campaign with the highest priority, and only falls back to the next one down when the higher priority campaign is not eligible for that particular search. Add a negative keyword to a high priority campaign and you are not blocking the traffic, you are pushing it down a level. String three campaigns together that way and you have your keyword funnel, sorting shopping traffic by intent before it ever costs you anything.
Here is how the three campaigns fit together.
| Campaign | Priority | Target ROAS | Negatives |
|---|---|---|---|
| High intent | High | Aggressive, low target | Brand terms, plus any generic keywords that are not high intent |
| Generics | Medium | Higher target than high intent | Brand terms only |
| Branded | Low | Least aggressive of the three | None needed |
The high intent campaign sits at the top on high priority with an aggressive, low ROAS target. Its job is to win the searches from people who know exactly what they want, and because those searches convert well, you can afford to bid hard for them. You block brand terms so they fall through to the branded campaign, and you block generic terms that are not high intent so they fall through to the generics campaign.
The generics campaign catches everything the first campaign pushed down. Medium priority, brand terms as the only negative, and a slightly higher ROAS target. That looser target is deliberate. Generic search terms convert at a lower rate almost by definition, so if you bid on them with the same aggression you use on high intent traffic, this campaign will happily eat your budget. The higher target lets you stay present in that part of the market without funding it out of your profitable traffic.
The branded campaign sits at the bottom on low priority with the least aggressive bidding in the account. It needs no exclusions, because the low priority setting means it only ever picks up what the two campaigns above it have rejected. Its only purpose is defensive. When somebody searches your brand, you do not want a competitor's product sitting above yours in the shopping carousel. You are buying presence, not incremental demand, so pay accordingly.
So the next time you hear that standard shopping is just an outdated version of Performance Max, ignore it. Every account we manage runs both, and the control this layer gives you is the reason the account stays profitable as it scales.
Layer three: standard search
One rule applies to everything in this layer: keyword themed ad groups. Tight groups of closely related keywords, each with ad copy written specifically for them. This matters more than it looks. Google rewards relevance between the search term, the keyword, the ad and the landing page, and that relevance is what drives Quality Score. A better Quality Score means better ad rank at a lower cost per click, so tight themes are effectively a discount on every click you buy. Dumping fifty loosely related keywords into one ad group with generic copy means paying full price for worse positions.
Build this layer in two stages, in order.
Stage one is a broad match campaign, structured with those keyword themed ad groups. The point of this campaign is discovery. You do not know every way your customers phrase what they are looking for, and exact and phrase match can only ever show you the demand you already thought to target. Broad match gives you reach into the searches you have not imagined yet. Run it for a few weeks and treat the search terms report as the actual deliverable. Revenue from this campaign is a bonus. The real output is a list of the exact search terms that convert best and return the most.
Stage two is where that list gets used. Once you know your winners, launch a second campaign built only on exact match, using only the terms you have proven convert. Same structure, same tight keyword themed ad groups. The combination of exact match and tight themes is what pushes Quality Score as high as it will go, and that is how you end up holding the best positions for the lowest cost. Your discovery campaign keeps running underneath it, feeding new terms up as it finds them.
Both of these campaigns exist to win new customers, so both of them exclude brand. Same reasoning as layer one.
Layer four: branded search
This is your least important layer, and I would rather be honest about that than sell you on it. Most brands do not need it. Somebody typing your name into Google is already most of the way to buying, and if you rank first organically, paying for the click above your own free listing is mostly a transfer of margin to Google.
There are two situations where it earns its place. The first is competitor pressure. As you scale, and particularly with so many advertisers now running Performance Max, you will find competitors bidding on your brand terms far more often than they used to. At that point branded search stops being a vanity campaign and becomes a defensive one, because the sale you lose to a competitor sitting above you is a sale you had already paid marketing budget to create.
The second is weak organic visibility. If your SEO is poor and you are not appearing prominently when someone searches your name, you have no free listing to protect, and the paid click is doing real work.
Check both before you commit budget here. If neither applies, spend the money in layers one to three.
What comes after the foundation
Only once all four layers are running properly should you start looking at YouTube and Demand Gen. The reason is simple. Google is a demand capture channel. People arrive with intent already formed and Google's job is to route them, which is very different from Meta, where you are interrupting someone and manufacturing the intent yourself. Demand generation on Google can work well, but it only works when there is a well built capture system underneath it to catch the demand you create. Build it the other way round and you spend money generating interest that your account is not structured to convert.
Get these four layers right first. This is the core structure driving 8 figure revenue for the clients we work with, and in almost every case the gains came from separating traffic properly rather than from finding some clever new tactic.
