Account structure is a strategy, not a setup.
Most Google Ads accounts fail before a single bid is placed. The structure decides what data you collect, and the data decides what you can manage. This page is our methodology for building accounts that answer questions instead of hiding them. Part of our Google PPC practice.
Structure decides what you can see
You cannot manage what your structure hides.
An account where every keyword lives in one campaign gives you one budget, one blended average, one set of learnings. You cannot see that emergency keywords outperform research keywords. You cannot see that one city converts and another burns cash. The dashboard shows you a single number, and that number lies by averaging.
Google's machine learning organizes around the campaign. Budget, bid strategy, location targeting, ad schedule, audience signals: these are all campaign-level levers. So the campaign is the unit of decision. The ad group is the unit of relevance. Get that division wrong and every decision you make is built on blended, muddy data.
This is why we treat account structure as the first and most durable decision in any Google Ads engagement. Keywords can be swapped in an afternoon. Bids can change tomorrow. But a bad structure corrupts every report you will ever pull, and no amount of bid management fixes corrupted reports.
Segment by intent, not by whim
Most accounts are segmented by service line or by whatever the last agency felt like that week. We segment by intent: the distance between the searcher and the sale.
Search intent falls into tiers. The top tier is transactional: someone ready to hire. "Emergency plumber near me," "sell my junk car today." The next tier is commercial research: "best roofer in Fort Lauderdale," "how much do junk car buyers pay." Below that is informational: how-to searches, DIY questions, price curiosity. Then brand searches, then competitor searches.
Each tier gets its own campaign, because each tier deserves its own budget and its own bid strategy. Hire-intent gets the largest budget and the tightest controls, because that is where revenue lives. Research gets a measured test budget or gets excluded. Informational gets excluded from paid search almost always: paying for curiosity is a luxury most local businesses cannot afford.
Within intent tiers, we segment two more ways. First by geography: separate campaigns for separate markets, because we never blend cities we cannot manage separately. What converts in Fort Lauderdale may burn in Miami, and a blended campaign cannot tell you that. Second by margin: high-margin services get their own campaigns so their economics never get averaged down by low-margin volume.
Ad group design: one idea per ad group
If one ad cannot speak to every keyword in the ad group, the ad group is too big.
Our rule is simple: five to fifteen tightly related keywords per ad group, all expressing one idea. "AC repair" and "AC installation" are different intentions, different urgency, different price points, and different landing pages. They are different ad groups. "Water heater repair" and "water heater replacement" are different decisions the customer is making. Different ad groups.
Tight ad groups let us write ads that mirror the search almost word for word. That lifts click-through rate, which lifts Quality Score, which lowers what you pay per click. Loose ad groups force generic ads, generic ads get ignored, ignored ads cost more per click. Relevance is not a nice-to-have. It is a cost lever.
We do not need hundreds of keywords per ad group, because close-variant matching already covers the variations. What we need is the right clusters: each one tight enough that the ad, the keywords, and the landing page all say the same thing. When all three match, Google rewards you and the customer trusts you. Both of those show up in the numbers.
Match-type strategy
Exact and phrase do the managing. Broad does the discovering, and only when we say so.
Our builds run on exact and phrase match. Exact for the proven core: the keywords we know convert, controlled tightly. Phrase for controlled expansion: new variations we want to test without opening the floodgates. That is the whole doctrine for the account that spends real money.
Broad match is a discovery tool, not a management tool. It spends your money to teach Google what you want, and you can get that education cheaper from search-term reports and a small, fenced discovery campaign. We never let broad run loose inside a hire-intent campaign, because broad's job is to find new things and a hire-intent campaign's job is to convert known things. Mixing the two poisons both.
Negative keywords are the third match type, and they are the walls that keep phrase and exact honest. Every build ships with negative lists from day one: job seekers, DIY, free, cheap, wrong services, wrong locations. A match-type strategy without negatives is a fence with no rails. The search-term report tells us what new rails to add, on a schedule, forever.
Budget allocation logic
Budgets follow intent and margin, not equality.
The most common budget mistake is equality: splitting the budget evenly across campaigns as if every campaign were equally valuable. They are never equally valuable. The hire-intent campaign gets the lion's share, because that is where the money comes back first. Research gets a test budget sized to answer a question. Brand gets a defensive token. Informational gets nothing until the profitable tiers are fully funded.
Small budgets change the shape of the build. A small account with twelve campaigns starves every campaign of learning data, and Google's bidding needs volume to learn. Better three campaigns with enough volume each than twelve campaigns with none. We would rather dominate one intent tier in one market than spread thin across everything and learn nothing.
Scaling is mechanical. Winners earn more budget in steps. Losers get cut. No sentimentality, no sunk-cost attachment to the campaign somebody spent a week building. The data decides, and the data is only trustworthy because the structure kept it clean. That is the loop: clean structure, honest data, mechanical scaling.
Naming conventions are infrastructure
A name you cannot read at a glance is technical debt.
Our convention encodes the four things that matter: market, intent, service, and match type. A campaign named FL-Broward_HireIntent_ACRepair_Exact tells anyone, instantly, what it does, who it serves, and how tightly it is controlled. An ad group named ACRepair_Emergency_Exact tells the same story one level down. Ads are labeled by test variant so we know which creative won six months later.
This matters more than it sounds. When we audit accounts named "Campaign 3" and "Campaign 3 copy" and "New campaign final FINAL," we burn days decoding someone else's improvisation before we can do a single useful thing. Naming is how an account stays manageable across people and across time.
The test is simple: six months from now, a stranger should be able to run the account from the names alone. If they cannot, the names failed, and the account will rot the moment the person who built it stops touching it.
Clean build vs messy build
Most accounts we audit look like the left column. All of ours look like the right one.
The messy build
- One campaign holding hundreds of keywords
- All match types mixed together, including loose broad
- One generic ad trying to speak to everyone
- Homepage as the landing page for everything
- No negative lists, or a handful of afterthoughts
- One blended average that hides every winner and loser
The clean build
- Campaigns segmented by intent tier, market, and margin
- Exact and phrase only, broad fenced into discovery
- Tight ad groups with ads that mirror the search
- Dedicated landing pages matched to each intent
- Negative lists from day one, maintained on a schedule
- Data that answers: which intent pays, which market pays, which ad wins
The messy account cannot be optimized. It can only be spent in. Every adjustment is a guess, because the data underneath is blended beyond recognition. The clean account is a question-answering machine: raise the budget here, cut it there, test this ad against that one. Management becomes a series of small, evidence-based moves instead of one big hope.
Our own junk-car operation, Instant Money Junk Cars, runs on this exact architecture: intent-tiered campaigns, tight ad groups, exact and phrase control, budgets that follow returns. That operation bought 3,000+ cars in a year and grew operations and profit 4x. We do not teach structure we have not lived inside.
When architecture matters most
Four situations where structure is the whole ballgame.
1. After bad agency work
The most common rebuild trigger. The old account is a tangle of broad match, blended campaigns, and mystery settings. Sometimes we restructure; sometimes the honest math says a clean rebuild costs less than untangling the mess. An audit decides, and we show our work.
2. When spend is scaling
Small spend hides structural sins. At scale, every hidden loser costs real money every day. Scaling is when blended averages become expensive, and when intent-tiered budgets start paying for themselves.
3. Multi-service businesses
High-margin services and low-margin services cannot share campaigns. The cheap, high-volume service will eat the budget while the profitable service starves. Separate campaigns, separate budgets, separate economics.
4. Multi-market businesses
Different cities convert differently, cost differently, and compete differently. Blending them into one campaign means your best market subsidizes your worst one and you never find out. Separate markets, separate campaigns, honest comparisons.
Want this thinking applied to your account? That is our Google Ads management service. Want to know what is wrong with your current build first? Start with a Google Ads audit.
Questions
Straight answers, before you apply.
As many as intent tiers times markets, and not one more. The minimum viable build is three: hire-intent, research, and brand. You grow from data, not from a template.
Yes, as a cheap defensive layer kept in its own campaign. Brand traffic converts easily and cheaply, so mixing it into non-brand campaigns pollutes the data you use to judge everything else.
It has a place, but never as the foundation. Structure first, automation second. PMax without negative controls and without feed discipline is a black box that spends in places you cannot see.
An audit decides. Sometimes a restructure is enough; sometimes the account is so tangled that a clean rebuild costs less than untangling it. We show our work either way before we touch anything.
Both. The build is part of management, not separate from it. A clean architecture is what makes ongoing management possible instead of guesswork.
Ask one question: can your account tell you which campaign deserves more budget and why? If the answer is a blended average across everything, the architecture is the problem, not the bidding.
Structure first.
Spend second.
Apply below. We will map your services by intent, build the account clean, and put every dollar where the data says it belongs.