Every inherited Google Ads account tells you something about the person who built it. Some tell you they were in a hurry. A few tell you nobody was ever really driving. But the most common story is the strangest one: the account looks immaculate, and it is quietly failing.
Eleven campaigns, one per service, each named properly. Ad groups nested underneath in neat rows. Budgets split evenly across the top. It has the look of a job done carefully, and that appearance is exactly why nobody questions it for eighteen months.
Then you open the conversions column. Four conversions here. Two there. One campaign at zero for the quarter. Spread across eleven campaigns, a business generating a perfectly healthy forty leads a month has managed to make every single campaign look like a failing one.
Nothing in that account is broken in the way owners expect broken to look. The tracking fires. The keywords are relevant. The ads are approved. What went wrong is structural, and structure is the part almost nobody thinks of as a decision at all. It feels like filing. It feels like housekeeping you do so the account is tidy for whoever looks at it next.
It is not housekeeping. Your account structure decides where money is allowed to go and how much your bidding gets to learn from. Those are the two things that determine what Google Ads costs you.
Every Campaign Is a Separate Wallet and a Separate Student
Two facts about the Google Ads hierarchy explain nearly every structural mistake we see, and neither one is complicated.
The first: your budget and your bidding strategy are set at the campaign level. Not at the account level, not at the ad group level. A campaign is a wallet with its own rules about how aggressively to spend.
The second: your keywords, your ads, and your landing pages live at the ad group level, inside that campaign. Ad groups are how you organize what you are saying to whom. They do not get their own budget. They compete for the campaign's.
Put those together and the consequence falls out immediately. When you create a second campaign, you have not organized anything. You have opened a second wallet with its own spending rules, and you have started a second student who has to learn the business from scratch using only the conversions that happen to flow through it.
That second part is the one that costs money. Smart Bidding is a model. It watches which searches, devices, times of day, and audiences turn into conversions, and it adjusts your bids in every auction accordingly. It builds that model from conversion data. And it builds it inside the campaign.
So when an owner splits a service line into its own campaign so they can "see it more clearly," they get their clean reporting row. They also get a model trained on a fraction of the evidence. The reporting improves and the performance degrades, and because those two things arrive together, the split usually gets credit for the visibility and never gets blamed for the cost.
Smart Bidding Doesn't Grade on a Curve
Here is where it gets concrete. Google publishes actual minimums for its bidding strategies, and they are worth knowing precisely, because the numbers that circulate in the industry are usually somebody's rule of thumb repeated until it sounded official.
For Target ROAS on Search and Shopping campaigns, Google states you need at least 15 conversions in the past 30 days at the conversion tracking level. Display needs 15 with valid conversion values across all campaigns combined. Demand Gen wants 50 in the past 35 days, with 10 of those in the last 7. Video Action campaigns want 30 in 30 days.
Note the phrase "at the conversion tracking level" on that first one. That is an account-wide floor for turning the strategy on, not a per-campaign guarantee that it will model well once it is running. Those are different questions, and the second one is the one that decides your results.
On the per-campaign question, the practitioners we watch tend to land in the same range from experience rather than from documentation: roughly 15 conversions a month in a single campaign before the bidding has much to work with, and closer to 30 before it settles down. More is better. Treat that as the collective judgment of people who manage a lot of accounts, not as a published rule. It matches what we see, which is that campaigns starved of conversion volume behave erratically no matter how good the keywords are.
Now run the arithmetic on a real small business. Say you generate 30 conversions a month. That is a genuinely healthy number for a local service business spending two or three thousand dollars.
In one campaign, that is 30 conversions of evidence. The model has something to work with.
Split three ways, it is 15, 10, and 5. The first campaign is marginal. The second is guessing. The third is adjusting your bids in live auctions based on the behavior of five people. Five. And it will do it with total confidence, because that is what the system does.
This is the part that surprises owners: the campaigns do not fail evenly. Watch a segmented account over a quarter and you will usually find return on ad spend declining right alongside conversion volume, campaign by campaign, down the list. The smallest campaigns perform worst. That is not coincidence and it is rarely because those services are less profitable. It is the data.
Merge those three back into one and nothing about your business changed. Your ads did not improve. Your landing pages are the same. But the model now sees 30 conversions instead of 5, and it starts making better bids, which is the entire mechanism by which Google Ads gets cheaper.
Start at One Campaign. Make Every Other One Argue for Itself.
The instinct most owners bring to this is additive. A new service, a new campaign. A new city, a new campaign. Something you want to watch closely, a new campaign.
Flip it. Start from the assumption that you have one campaign, and make every additional one justify its existence. Not "would it be nice to see this separately," but "is there a reason this money cannot share a budget and a bidding target with everything else."
That is a high bar on purpose, and only a handful of reasons clear it.
Brand versus non-brand. People searching your company name are already yours. Mixing them with cold traffic corrupts every number you look at. This one is close to mandatory, and it gets its own section below.
Geography. If you serve genuinely different markets with different costs, different competition, or different value per customer, they need separate budgets. A campaign covering a metro where a lead is worth $400 should not share a bid target with one where it is worth $120.
Different margins or efficiency targets. The one owners most often miss, and the one that costs the most. Also its own section below.
Genuinely different categories. Not different keywords. Different businesses-within-the-business, where the customer, the buying cycle, and the intent have little in common. A company selling office chairs and houseplants is one business on paper and two audiences in practice.
That is close to the whole list. Notice what is not on it. "So I can see the numbers separately" is not on it, because you can segment your reporting by ad group without splitting your budget. "Because it is a different keyword" is not on it. "Because the last agency set it up that way" is definitely not on it.
There is a real tension here worth naming honestly, because you will run into it the moment you search for advice on this. One school says consolidate almost always, because data density beats everything. Another says segment, because otherwise Google picks a few favorite keywords, pours the budget into them, and you never learn what the rest could have done.
The second argument is not stupid. It is a measurement argument, and it is true that a single bucket hides which parts are pulling weight. But it is affordable only when you have volume to spare. On 30 conversions a month, buying visibility with your data is a trade that loses. On 300, it is a reasonable thing to pay for.
Structure follows volume. That is the rule underneath both schools of thought, and it is why copying a structure from someone whose budget is ten times yours goes wrong so reliably.
Two Margins, One Bid Target, and Somebody Loses
This is the one that quietly drains profit, and it never shows up as a problem on the dashboard, because on the dashboard it looks like success.
Budget and bidding are set at the campaign level. So when two services share a campaign, they share a target. If you are running Target ROAS, they share one ROAS target. If you are running Target CPA, they share one cost per acquisition.
Now suppose you run a home services company. Routine maintenance calls carry a 20% margin. System replacements carry 40%. You put both in one campaign with a single target CPA of $80.
At $80 a lead, the replacement work is enormously profitable. The maintenance work is barely breaking even, and on the bad weeks it is losing money once you count the truck roll.
Google does not know any of this. It cannot see your margins. It sees conversions, and it optimizes toward getting more of them at your stated cost. Maintenance leads are cheaper and more plentiful, so that is where the budget flows. Your cost per lead looks great. Your campaign looks like a winner. Your bank account disagrees, and it takes most owners a year to notice, because every metric they were told to watch is green.
The fix is not clever. Separate the campaigns so each one can carry a target that reflects what that work is actually worth. Replacements can afford a $150 lead. Maintenance cannot afford $80. Once they have their own budgets and their own targets, both can be profitable, and you can decide deliberately how much of your money goes to each instead of letting an algorithm decide by accident.
The same logic applies to price tiers within one category. If you sell a $200 product and a $4,000 product, they do not belong to the same bid target even though they sit in the same aisle. The unit economics are different, so the instruction has to be different.
Your Own Name Is Not a Marketing Channel
If someone types your company name into Google, you did not win that customer with an ad. You won them earlier, somewhere else, and the ad is collecting a toll on the way in.
There is still a case for bidding on your own brand. Competitors do bid on your name, and an unanswered competitor ad sitting above your organic listing costs you real business. The exception is narrow: if genuinely nobody bids on your terms and you already hold the top organic result, you can skip it.
But brand traffic must live in its own campaign, and every other campaign must be told to stay away from it. Otherwise brand conversions leak into your cold-traffic campaigns and flatter results that do not deserve it. A campaign quietly harvesting people who were already looking for you will always look like your best performer, which means it attracts more budget, which means the actual waste elsewhere stays hidden behind it.
On the mechanics, Google now offers a better tool than the one most accounts are still using. Brand exclusions block your ads from showing on searches for your brand across Performance Max and Search, and they cover common misspellings and related subsidiary brands automatically. Google explicitly recommends them over negative keywords for this job, because a negative keyword list only blocks what you thought to write down. Worth noting if you do use negatives: there is a limit of 1,000 negative keywords at the account level.
We have written about how badly branded traffic distorts reporting in Performance Max Isn't Your Smartest Campaign. It's Google's. The short version is that until brand is separated, you cannot trust a single number on the screen.
One Keyword Per Ad Group Made Sense When You Wrote the Ad
Inside your campaigns, there is a second structural question, and a lot of accounts are still answering it with a strategy that expired years ago.
There used to be a well-regarded approach called single keyword ad groups: one keyword, its own ad group, its own ad. It worked, and it worked for a specific reason. Back then you wrote a static ad. Three headlines, a description, in the order you chose. You controlled exactly what a searcher saw, so pairing one ad to one keyword gave you perfect relevance.
That is not how ads work now. A responsive search ad takes up to 15 headlines and 4 descriptions, and Google assembles them into combinations on the fly. There are thousands of possible versions of a single ad, and the system needs a real volume of impressions to work out which combinations perform. An ad group bidding on one keyword will never deliver that volume. You have built a test that cannot finish.
The replacement is the single theme ad group. One coherent topic per ad group, with roughly 10 to 20 keywords that all mean the same kind of thing, and ads written to that theme.
For a plumber, that is not one ad group per keyword. It is general plumbing, emergency plumbing, drain cleaning, water heater installation. For an auto shop: oil changes, brake service, AC repair, diagnostics. Each theme gets keywords that belong to it and ad copy that speaks to it, because someone searching for an emergency at 11pm needs a different sentence than someone pricing an oil change.
Two practical rules that go with this. Keep one landing page per ad group; if two ad groups point at the same page, they probably are not different themes. And watch for overlap between themes, particularly if you use broad match. Loosely-worded keywords across adjacent ad groups will end up competing in the same auctions, and you will be bidding against yourself with your own money.
Now, the honest counterargument. Some experienced practitioners split further than this, particularly for local service work, separating "plumber," "plumber near me," and "plumber Denver" into their own ad groups. Their reasoning is sound: dumped in one bucket, Google will favor a handful of terms and you will never discover that the city-name searches convert twice as well as the generic ones.
They are right that you learn something. The question is what it costs. Splitting ad groups this way spreads your data thinner and, if you also split ads, slows every test you are running. On a large budget, that is a fine price for the insight. On a small one, you are paying with the only thing that makes your bidding work. Start themed, and split further when volume gives you room.
Sometimes the Right Move Is to Split
None of this is an argument that consolidation is always correct. It is an argument that splits should be earned. Sometimes they are.
The clearest signal is an ad group doing well and being held back. If one ad group has a noticeably better return than the rest of its campaign, and its search impression share is low, that is a service line with demand you are not capturing. It is losing the internal competition for a shared budget, not losing in the market. Pull it into its own campaign, give it its own budget and its own target, and let it grow to the size the demand supports.
The other honest signal is simply outgrowing the setup. When a campaign is well past 30 conversions a month and still climbing, the case for consolidation weakens, because you now have data to spare. That is the point where segmenting buys you control and costs you little. Most small businesses never reach it in every campaign, and there is nothing wrong with that. It just means the structure that fits you is simpler than the one in the case study you read.
The mistake is not segmenting. The mistake is segmenting on a budget that cannot feed what you built.
What to Do Monday
None of this requires a rebuild. It requires about an hour and a willingness to act on what you find.
Count your campaigns, then pull the last 30 days and look at conversions per campaign. Write down every campaign under roughly 15 conversions for the month. That list is your candidate pool for consolidation.
For each one on that list, ask the only question that matters: is there a real reason this cannot share a budget and a bid target with a sibling campaign? Brand separation, a genuinely different geography, a different margin profile, a different category. If none of those apply, it should probably be merged into the closest relative. When you merge, only combine work that earns you a comparable amount per sale, or you will have solved a data problem by creating a profitability one.
Then check margins across everything currently sharing a bid target. Any two services with meaningfully different profit per job in the same campaign is a leak, regardless of how the campaign looks.
Confirm your brand separation. One brand campaign, brand exclusions applied everywhere else.
Finally, look inside your campaigns at the ad group level. Anything that is one keyword in its own ad group should be folded into a theme with its relatives.
One warning worth taking seriously: consolidating campaigns resets learning. Performance usually dips before it improves, and in the case studies we have watched play out, the recovery took a few weeks rather than a few days. Do this when you can leave it alone for a month, not the week before your busy season.
Most small accounts we look at are not underfunded. They are over-divided: a reasonable budget cut into so many pieces that no single piece can learn enough to spend well. Fixing that costs nothing except the willingness to have a less tidy-looking account.
If you would rather have someone look at yours before you start merging things, that is what we do all day. We are always happy to walk through an account structure with a business owner and say plainly what we would change and what we would leave alone, whether or not you work with us afterward.
