Google Ads Campaign Structure: SKAG vs Alpha/Beta

When I started running Google Ads eight years ago, I did something dumb: I dumped every keyword I could think of for a single product into one ad group. Thirty bucks a day, straight down the drain — impressions, clicks, and zero conversions. I spent two weeks tweaking bids and rewrote my ad copy five times. Nothing moved.

Eventually it clicked: the problem wasn't the bidding, and it wasn't the copy. It was the structure.

Structure is the least glamorous part of search advertising. It doesn't grab attention the way bid strategies or ad copy do, but it decides where your budget flows. Get it wrong and every optimization you make afterward is just painting over a leaky bucket. So let me walk through the two search campaign structures I've run over the years — and which one I'd actually pick in 2026.

Before you touch anything: set up conversion tracking

This one's a prerequisite, but too many people skip it. Launching without conversion tracking is like driving with your eyes closed — you know how much you're burning, but not which keyword or which ad group is actually bringing in orders. When it's time to optimize, you've got nothing to go on.

My rule: I don't turn on a single ad until conversion tracking is verified — I place a test order myself and confirm it fires back correctly. Don't cut this corner.

During testing, don't hoard keywords

Beginners love to dump in hundreds of keywords, figuring "wider coverage is better." It's the opposite. When budget is tight, for testing a single product I keep the total keyword count between 10 and 30.

Fewer isn't laziness — it's that every keyword needs enough clicks before you can judge it. Spread $30 a day across 200 keywords and each one might get two or three clicks a week; you could run for three months and still not have enough data to make a call. Few and focused is how the data actually comes through.

Group ad groups by theme, split what doesn't belong

Once you've picked your keywords, the next step is grouping. The rule is simple: keywords with similar intent go in one ad group; different themes get split into different campaigns.

Here's an example. Say I sell portable espresso makers. The keywords roughly split into three groups:

Ad group 1 (portable): portable espresso maker, portable espresso machine
Ad group 2 (travel/camping): travel espresso maker, camping espresso machine
Ad group 3 (handheld/manual): handheld espresso maker, manual espresso maker

All three live under the "portable espresso maker" umbrella with similar intent, so they can sit in the same campaign. But if I also sold coffee grinders, that's a different product and a different search intent — I'd spin up a separate campaign rather than cram it in.

When the split is clean, you can write dedicated copy and build dedicated landing pages for each ad group. Relevance goes up, and Quality Score and conversion rate climb with it.

Add these negatives from day one

One easily overlooked part of structure is the level you place negative keywords at. Universal terms — the ones your whole product line should block — go at the campaign level, where they apply to every ad group underneath. Ones specific to a single ad group go at the ad group level.

The set I add to every new campaign on day one is the brand names of the big US marketplaces: amazon, ebay, etsy, walmart, target, home depot, and so on. The logic is blunt — someone searching "xxx amazon" has already made up their mind to buy on Amazon, so showing them your ad is mostly money wasted. Same goes for geography (running US? exclude canada, uk, near me, and the like).

(How to systematically dig up negatives and split them by match type is a whole separate post — I'll leave it for another day.)

Small budget: SKAG (Single Keyword Ad Group)

With that groundwork laid, let's get into it. The first structure is SKAG — Single Keyword Ad Group: one campaign with 3 to 5 ad groups, and each ad group holds exactly one keyword.

For beginners on a small budget, I used to almost always suggest starting with SKAG, because its biggest win is saving money — it keeps waste to a minimum. The upsides are real:

One keyword per group means you can write the tightest possible ad copy for it — high CTR and relevance
You can build a dedicated landing page for that keyword, pushing conversion rate and Quality Score up
When a keyword isn't working, you spot it instantly, and optimization stays clean

The downsides, to be fair:

You end up with a lot of ad groups, and building them out is grunt work
Keywords are locked down so tight you can miss valuable related searches
Traffic and impressions run low, precisely because everything's so restricted

Bigger budget: the Alpha/Beta structure

With a little more budget, I lean toward Alpha/Beta. The appeal is that it lets Google flex its automatic keyword-discovery muscle while you still hold onto your precise traffic. Two roles:

The Alpha ad group uses exact match to capture your most precise, already-validated traffic. It's basically SKAG, except one group can hold several exact keywords.

The Beta ad group uses broad match to go exploring — to surface converting search terms you'd never have thought of. And here's the critical move: the Beta group has to use negative keywords to block all of Alpha's exact-match terms.

Why? Because if you don't, the same search term pits your Alpha and Beta groups against each other — only one group ends up getting the impression, and your data turns to mush.

A quick aside: that exact "same keyword fighting itself across two ad groups" situation is what I later built into the Cannibalization rule in my own tool, ADM (an AI agent that optimizes Google Ads automatically). It scans for duplicate keywords across ad groups inside the same campaign, ranks them by ROAS, conversions, and Quality Score, and automatically pauses the weaker one — handing the traffic to the group that's winning. Watching for this by hand is a chore, and it only gets worse as your ad group count grows.

Run Beta for a while and it'll throw off a batch of new, converting search terms. That leads to the heart of Alpha/Beta: take the terms that converted in Beta and add them to Alpha as exact-match keywords. Once Beta stops surfacing useful new terms, you can shut it down.

This back-and-forth — broad match scouts, then the converting terms get promoted to exact match — was something I used to do by hand, religiously, every week. Eventually I got tired of it and turned it into ADM's Keyword_Harvesting rule: it automatically scans the search terms that have already converted, adds the high-intent ones as exact-match keywords, and runs a layer of quality filtering in between — it won't harvest the obvious flukes, so you don't mistake luck for skill.

If Alpha/Beta is working, the next move is to scale: turn on search remarketing, expand your targeting, and switch your bid strategy to Maximize Conversions or set a Target CPA to let volume build.

In 2026, what would I pick?

Let me be straight with you. SKAG and Alpha/Beta are both strategies that have been around for years. Lately Google has been pushing broad match plus Smart Bidding hard, and it's openly said exact match "close variants" will keep getting looser — meaning the exact match you set, it'll happily stretch to terms it decides are close in intent. So the pure-SKAG approach of "one keyword per group, locked down tight" just doesn't have the edge it used to.

But "think the structure through before you build" is something I still do every single time I launch a new campaign. My own accounts these days run a variation of Alpha/Beta: exact match guards the validated terms, broad match plus Smart Bidding goes hunting for new traffic, and negatives and harvesting shuttle terms back and forth in between. I only still reach for SKAG in two situations — a really small budget ($10–20 a day, where you can't afford broad match running wild), or a single keyword valuable enough to deserve its own landing page.

The bottom line

Get your structure right and the budget you save doesn't vanish — it flows automatically toward the keywords that actually convert. That's the whole point of structure: not to make your dashboard look tidy, but to put every dollar where it can drive a sale and push your ROAS up, bit by bit.

If you're wrestling with how to structure your own account — or you've got an approach that's different from mine — let me know. I'd genuinely like to hear how other people do it.