Google Ads Campaign Structure: SKAG vs Alpha/Beta
Eight years back, in my first year of Google Ads, I made a dumb mistake: whatever keywords came to mind for a single product all went into one ad group. Every dollar of that thirty-buck daily budget leaked away: impressions arrived, clicks arrived, orders never did. Two weeks of bid tweaking and five versions of the ad copy got me exactly nowhere.
The realization took a while, but it landed: neither bids nor copy were the problem. Structure was.
Nobody finds structure exciting. It's the least glamorous piece of search advertising. Bid strategies and ad copy get the attention; structure quietly decides where the money goes. Set it up badly and everything you optimize later is decorative paint on a leaking bucket. Two search campaign structures have carried my accounts over the years, and I'll walk through both, plus which of the two I'd genuinely pick in 2026.
Before you touch anything: set up conversion tracking
Skip this and nothing else you do will matter. Launching with no conversion tracking is driving blind: you can see the spend leaving, you can't see which keyword or ad group produced the orders. Come optimization time, there's no evidence to work from.
The line I hold: no ad gets enabled until I've verified tracking, by walking through a test purchase personally and then checking that it registers correctly. Don't skip it.
During testing, don't hoard keywords
New advertisers love dumping in hundreds of keywords on the theory that wider coverage has to be better. The reverse is true. Testing one product on a tight budget, my total keyword count stays between 10 and 30.
Keeping the list short isn't laziness. Judging any keyword requires a pile of clicks first. Put $30 a day against 200 keywords and you're looking at two or three clicks each per week, three months of running, still no usable verdict. Narrow and focused is what makes data show up at all.
Group ad groups by theme, split what doesn't belong
Grouping comes after keyword selection, and the rule is short: similar intent earns a shared ad group; a different theme earns its own campaign.
Take an example. Suppose the product is a portable espresso maker. Three clusters fall out:
- Portable → Ad group 1: portable espresso maker, portable espresso machine
- Travel/camping → Ad group 2: travel espresso maker, camping espresso machine
- Handheld/manual → Ad group 3: handheld espresso maker, manual espresso maker
All three share enough intent that a single campaign can hold them; "portable espresso maker" is the umbrella. Coffee grinders, though, would mean another product and another search intent. That gets its own campaign, not a slot crammed in beside them.
A clean split lets you write copy and build landing pages per ad group. Relevance improves, and Quality Score plus conversion rate follow.
Add these negatives from day one
The level you attach a negative keyword to is easy to overlook. Terms the entire product line should block belong at campaign level, so every ad group beneath inherits them. Terms that matter for one ad group only belong at ad group level.
Day one of any new campaign, in goes the brand-name list: amazon, ebay, etsy, walmart, target, home depot, plus the rest of the major US marketplaces. The thinking is crude and correct: anybody typing "xxx amazon" has already chosen Amazon, so your ad is money burned. Geography works the same way (US-only? keep out canada, uk, near me, and similar).
(Digging negatives up systematically and sorting them by match type deserves its own post — another day.)
Small budget: SKAG (Single Keyword Ad Group)
Time for the first structure. SKAG stands for Single Keyword Ad Group: 3 to 5 ad groups inside one campaign, each holding exactly one keyword.
On a small budget, a beginner used to get one recommendation from me nearly every time: start with SKAG. Money is what it saves — waste stays minimal. The gains are genuine:
- A single keyword per group lets the ad copy be as tight as copy gets — CTR and relevance both benefit
- That keyword can have a landing page of its own, lifting Quality Score alongside conversion rate
- A keyword that isn't performing shows up immediately, which keeps optimization tidy
The costs, to be even-handed:
- Ad group count balloons, and building them is grunt work
- Locking keywords down that hard can hide valuable related searches
- Because everything is so constrained, traffic and impressions stay thin
Bigger budget: the Alpha/Beta structure
Loosen the budget and Alpha/Beta becomes my preference. What draws me in: Google gets room to exercise automatic keyword discovery, while your precise traffic stays under your control. Two roles do the work:
The Alpha ad group is exact match, aimed at the most precise traffic you've already validated. Think SKAG with a looser ceiling, several exact keywords can share one group.
The Beta ad group runs broad match and explores, surfacing converting search terms that would never have occurred to you. The critical move sits here: every exact-match term in Alpha must be blocked inside Beta with a negative keyword.
Why? Skip that step and one search term puts Alpha and Beta in a fight with each other. Whichever group wins takes the impression, and the data underneath turns to mush.
Aside: one keyword quietly competing with itself in two ad groups, the "same keyword fighting itself across two ad groups" situation, is exactly what my Cannibalization rule was built for. It lives inside ADM, the AI agent I wrote so Google Ads optimizes itself. The rule looks for keywords duplicated between ad groups of one campaign, ranking each on Quality Score, ROAS and conversions, then pauses whichever is weaker and hands its traffic to the winner. That kind of watch, done manually, is a chore, and a worse chore each time an ad group is added.
Let Beta run and it will throw off fresh terms that convert. Which is the core mechanic of Alpha/Beta: whatever converted in Beta gets promoted into Alpha as an exact-match keyword. Beta can be shut down once it stops surfacing anything useful.
That loop, broad match scouting, then promoting whatever converts to exact, used to be my weekly ritual, done religiously by hand. Eventually tired of it, I built it into ADM as the Keyword_Harvesting rule. It reads converted search terms and promotes the high-intent ones into exact-match keywords, applying a quality filter in between so obvious flukes never get harvested and luck doesn't pass for skill.
When Alpha/Beta is working, scaling is the next move: switch search remarketing on, widen targeting, and move to Maximize Conversions as the bid strategy, or go with a Target CPA so volume can build.
In 2026, what would I pick?
Straight answer: both of these are old strategies. In recent years Google has pushed broad match and Smart Bidding relentlessly, and it openly says "close variants" of exact match will keep loosening, pick an exact keyword and it will stretch to whatever Google judges close in intent. So purist SKAG, "one keyword per group, locked down tight", has lost the edge it once had.
What hasn't changed: "think the structure through before you build" applies to every campaign I launch, without exception. My accounts today run an Alpha/Beta variant — Smart Bidding and broad match hunting fresh traffic, exact match holding the validated terms, negatives and harvesting shuttling words back and forth between the two. Two cases still send me to SKAG: a tiny budget ($10–20 a day, small enough that broad match would run wild), or one keyword so valuable it deserves a dedicated landing page.
The bottom line
Structure done right doesn't make savings disappear; those dollars move on their own toward the keywords that convert. That is what structure is for, not a tidy dashboard, but every dollar sitting where it can produce a sale and nudge ROAS upward.
Stuck on structuring your own account, or running a setup unlike it? Tell me. I genuinely want to know how other people approach this.