Google Ads campaign structure that actually scales

Account structure is where most Google Ads performance is won or lost, and it's the thing least likely to get attention because it isn't as satisfying as writing new ads.

Here's the approach I use.

Structure follows two things: intent and margin

Not product categories. Not what your website navigation looks like.

Intent determines how someone should be bid on. A person searching "buy running shoes size 9" is worth more than someone searching "are running shoes good for flat feet". Same category, completely different value.

Margin determines what you can afford to pay. If one product line runs 60% margin and another runs 18%, they cannot share a target CPA. Yet they routinely sit in the same campaign.

Split on those two axes first.

The over-segmentation trap

The old single-keyword-ad-group approach made sense when bidding was manual. With Smart Bidding it actively hurts you: every campaign needs conversion volume to learn, and splitting into twenty campaigns means twenty campaigns that never gather enough.

Rule of thumb: a campaign should be able to generate roughly 30+ conversions a month. If it can't, it should be merged with something.

The under-segmentation trap

The opposite failure is one campaign holding everything, where you can't set different targets for things with genuinely different economics, and budget flows to whatever converts cheapest rather than whatever is most profitable.

Cheapest and most profitable are not the same thing. High-margin products usually have more competition and higher CPCs. A consolidated campaign will quietly starve them.

A structure that works for most accounts

Brand campaign, separate, always. Your own name converts at a rate that flatters every average it's mixed into. Keep it isolated so you can see true non-brand performance. Watch for Performance Max eating this traffic and claiming credit.

Non-brand Search, split by margin tier. Usually two or three campaigns: high-margin, standard, and clearance or low-margin if relevant.

Performance Max, if e-commerce. Genuinely effective, with two caveats: exclude your brand terms, and use asset group structure that mirrors your margin tiers rather than dumping the whole catalogue in one group.

Remarketing, separate. Different intent, different messaging, different economics.

Match types, practically

Broad match with Smart Bidding and a clean conversion signal works far better than it used to — but only if your conversion signal is accurate. On a broken signal, broad match will find you enormous volumes of the wrong thing.

My default: phrase and exact for core terms where you know what works, broad match in a separate campaign for discovery, with search term reviews weekly and negatives applied ruthlessly.

The weekly maintenance that actually matters

1. Search terms review. Every week. This is where wasted spend lives.
2. Negative keyword list updates. Shared lists across campaigns.
3. Check for cross-campaign overlap. Two campaigns bidding on the same query is you competing against yourself.
4. Asset performance. Replace the worst-performing headlines and images.

None of this is exciting. All of it compounds.

When to rebuild vs when to fix

Rebuild if the tracking was wrong when the account was built — everything learned since is built on bad data.

Fix in place if the structure is merely untidy but the data is sound. Rebuilding resets the learning, and that cost is real.

For the tracking layer underneath all of this, read why your conversion tracking is probably wrong.