Around 2020, Google’s message to advertisers became increasingly clear: consolidate your campaigns. Fewer campaigns meant more data flowing into each one, which meant the algorithm could make smarter decisions. Split your campaigns too finely and you were starving the AI of the signals it needed.
It was compelling logic. And a lot of e-commerce businesses followed it.
Now, a few years on, it’s worth being honest about what actually happened.
The Theory Behind Consolidation
The argument was straightforward. Google’s Smart Bidding systems — Target ROAS, Maximise Conversion Value, and so on — learn from conversion data. They need a certain volume of conversions to make accurate predictions. If you split a product range across eight campaigns, each one might only see 15-20 conversions a month. That’s not enough for the algorithm to work reliably.
Consolidate those eight campaigns into two or three, and suddenly each campaign is seeing 60-80 conversions a month. The AI has enough signal to make good decisions, bid more intelligently, and find the right customers at the right price.
In principle, this is correct. Data pooling works. The algorithm is genuinely better when it has more signal.
Where Consolidation Delivered
To be clear: consolidation did help in the right circumstances.
The clearest wins came from tidying up campaigns that had been fragmented for no good reason. If you had five separate campaigns for variations of the same product at similar price points — say, five different single-origin coffees all priced between £10 and £15 — consolidating them into one was almost always an improvement. The algorithm could pool the data across all five, bid more accurately, and you’d typically see better ROAS within a few weeks.
Brand campaigns were another obvious case. Splitting branded traffic across multiple campaigns — one for exact match, one for phrase, one for different product categories — was common practice in the manual bidding era, when you needed that control. Under Smart Bidding, it was unnecessary fragmentation. Consolidating brand into a single campaign with good negative keyword structure worked well.
The general principle held: if campaigns are targeting similar products at similar margins, pooling them usually helps.
Where It Went Wrong
The problems emerged when businesses applied consolidation logic indiscriminately — merging campaigns that had genuinely different characteristics into single buckets and expecting the algorithm to sort it out.
The most common mistake was mixing products with different profit margins. Imagine an e-commerce business selling both budget furniture (30% margin) and premium furniture (55% margin). Consolidating these into a single campaign and setting one ROAS target creates an impossible situation. To hit 300% ROAS on the budget range you might break even. To hit 300% ROAS on the premium range you’re leaving substantial profit on the table. The algorithm optimises towards the target you’ve set, which means it will naturally favour whichever products are easiest to hit that target — often the cheaper, lower-margin ones, because there are more searches and more conversions.
The result: budget floods into products that perform fine on the chosen metric but don’t actually generate the profit the business needs.
Brand and non-brand consolidation caused similar problems. Brand searches convert at a fundamentally different rate to non-brand searches. A user searching your brand name is much closer to purchasing than one searching a generic category term. Mixing the two in a single campaign means the algorithm’s conversion rate expectations get skewed by the high-intent brand traffic, causing it to underbid on non-brand terms where the rates are naturally lower.
The Nuance That Emerged
What the consolidation experiment taught us is that the original principle was right but incompletely stated.
The correct version is: consolidate campaigns that are targeting the same customer intent at the same margin tier. Don’t consolidate across meaningful differences.
The relevant segmentation lines aren’t arbitrary. They’re based on:
Margin tier. Products with substantially different margins need different ROAS targets. If you group them, you’re setting the wrong target for at least one group. Keep high-margin and low-margin product ranges in separate campaigns with appropriate ROAS targets for each.
Brand vs non-brand. These are different customer journeys. Brand searches should generally stay in their own campaign where you can set appropriate bids without contaminating the algorithm’s view of non-brand performance.
Product intent. Seasonal or gift products behave differently to everyday staples. Consolidating them can cause the algorithm to misread demand signals — especially around peak periods when one product type spikes and the other doesn’t.
What to Do Now
If you’ve already consolidated and things are working well, don’t touch it. The algorithm doesn’t benefit from disruption without good reason.
If you’ve consolidated and you’re seeing the algorithm consistently prioritising certain products while ignoring others, look at whether margin differences are driving that. The algorithm isn’t making a mistake — it’s doing exactly what you’ve told it to do, which may be optimising towards the wrong goal.
For anyone building campaign structure from scratch: consolidate at the same category and margin tier. Don’t consolidate across meaningfully different product groups. Brand stays separate. And check your ROAS targets — if you’re running campaigns with very different margin profiles under the same target, that’s where most of the structural problems I see in Google Ads accounts are hiding.
The fewer-campaigns argument was never wrong. It was just applied too broadly, to situations where the differences between product groups mattered more than the benefits of data pooling.
If your campaign structure has grown complicated over the years and you’re not sure whether it’s working for or against you, a Shopping campaign review from Roksys tends to surface exactly these kinds of structural issues.