Google Ads Budget Allocation: The 70/20/10 Spreadsheet That Survives PMax

Most Google Ads accounts split budget by campaign type instead of by what's actually working. The popular 70/20/10 framework (70% to proven performers, 20% to optimization, 10% to experiments) only holds up if the proven 70% isn't being cannibalized by Performance Max winning auctions a Search campaign should have. Track each campaign's CPA against its share of spend weekly, then re-split every four weeks.

If you want the broader strategy this fits inside, start with our Google Ads strategy guide. This piece is the zoom-in on the spreadsheet most accounts skip and end up reverse-engineering after their next QBR.

Why most budget splits fall apart by week three

The split a rep walks you through in onboarding is almost always one of two shapes. Either it's a flat percentage across campaign types (40 Search, 40 PMax, 20 Shopping is a common one), or it's whatever Google's budget recommendations nudge you toward when you log in, which tend to reward whichever campaign already spent the most.

Both fail for the same reason. Neither one reflects what's actually closing.

By week three, three things have usually happened. PMax has crept above its allocation because it can spend across Search, Shopping, Display, YouTube, and Gmail under one budget and Google routes spend to whichever placement looks cheapest that day. Branded search has eaten more share than it should because brand traffic is artificially efficient. And the experimental 10% has either been reabsorbed quietly or it spent its money on one campaign that produced zero conversions and now nobody wants to touch it.

The fix is not a better split. It's a spreadsheet that forces you to look at spend share against contribution share every week, instead of letting the platform decide for you.

The five buckets every allocation spreadsheet should track

I'd argue any working budget allocation spreadsheet needs five columns, not three. The 70/20/10 framing is a starting position, not a structure. Here are the buckets that actually matter:

1. Brand search. Separate line. Always. Brand search converts at 5 to 10x non-brand and skews every blended number you'll look at later. If you don't pull it out, your "high-performing Search campaign" is probably your brand campaign with a couple of generic keywords stuck on for cover.

2. Non-brand high-intent Search. The keywords people type when they have a wallet open. Bottom-of-funnel queries, comparison terms, "near me" variants if you're local. This is the bucket Google's automation will quietly starve in favor of PMax if you let it.

3. Performance Max. Single line, even though the budget covers six surfaces underneath. You don't get to allocate inside PMax, so don't pretend you do. The new channel-level reporting that rolled out earlier this year finally lets you see what PMax is actually doing under the hood, but you still can't steer it.

4. Standard Shopping (e-commerce only). If you run both Standard Shopping and PMax with product feeds, this needs to be split out. Smarter Ecommerce's PMax guidance recommends putting the majority of your Shopping budget on Standard Shopping with a lower target ROAS so it wins the auctions, then letting PMax run lower priority for query discovery. That only works if you're tracking both lines separately.

5. Experiments. One line, capped. Demand Gen, a new ad group, a fresh keyword theme, whatever you're testing this month. The cap is the point. Without a cap, this bucket either disappears or eats the whole account.

Every line has the same five fields next to it: budget, spend-to-date, conversions, CPA, and share of total spend. That last one is the column most people skip and it's the one that catches drift.

How to actually fill in the numbers

Start with your last 60 days of conversion data. Not 30. Sixty. PMax especially needs the longer window because each PMax campaign needs roughly 20 to 30 conversions per month to exit the learning phase, so a 30-day view is too volatile to plan against.

For each of the five buckets, calculate the contribution share: percentage of total conversions that bucket produced over those 60 days. Then calculate the spend share. The gap between the two tells you where to move money first.

Concrete benchmark: if a bucket's spend share is more than 1.5x its contribution share, it's overfunded. If contribution share is 1.5x or more above spend share, it's starved. The 1.5x floor sounds arbitrary but it's the threshold I keep landing on after watching reallocations under 1.5x produce essentially zero CPA movement, mostly because account-level noise eats the change.

Apply 70/20/10 as a guardrail, not a target. The rough shape most accounts I'd trust look like:

• 50 to 60% in non-brand Search and Shopping combined (the proven workhorses)
• 15 to 25% in PMax
• 10 to 15% in brand (capped, never uncapped)
• 5 to 10% in experiments

If your account is bigger or you're running heavy seasonal pushes, Google's campaign total budgets feature is now available for Search, Standard Shopping, Performance Max, Demand Gen, and YouTube. It lets you set a fixed total spend across 3 to 90 days with no daily cap, which is the cleanest way to run a flighted promo without babysitting daily budgets. PPC Land's coverage noted that existing campaigns can't switch to this, only new ones can use it, so plan for that constraint.

The reallocation rule: when to move money, when to leave it alone

The hardest discipline in budget allocation is doing nothing for the first two weeks of any change. Smart Bidding rebuilds learning data based on the new budget signal, and a campaign that looks underperforming on day 8 often looks normal again on day 18. Sarah Stemen's analysis of campaign total budget pacing walks through how the algorithm front-loads spend in flighted campaigns, which means the first week of any reallocation is essentially noise.

The rule I keep coming back to: never reallocate more than 25% of any bucket in a single week, and never touch the same bucket two weeks in a row.

This sounds slow. It is. But every time I've seen an account thrashed by weekly aggressive reallocations, the CPA gets worse before it gets better, and most teams pull the plug before the new equilibrium settles. Patience is roughly the line between accounts that compound and accounts that re-split every quarter and wonder why nothing improves.

One exception: if a bucket's CPA is more than 2x your account target for two consecutive weeks, cut it 50% immediately and don't wait for the math. That's not reallocation, that's bleeding control.

What the spreadsheet won't catch

A spreadsheet only sees what your campaigns report. It won't catch the two biggest leaks in most Google accounts.

The first is PMax cannibalization. PMax operates at higher campaign priority than standard Search, which means when a query matches both, PMax wins. Savvy Revenue's brand splitting framework argues you should pull brand terms out of PMax and Shopping if you're a large brand, have a high brand conversion rate skewing your data, or need clean new-customer acquisition metrics. You shouldn't split if you have very few brand searches (under 10% of spend) or a small account where splitting starves the algorithm. That nuance matters because the wrong call here makes your spreadsheet look right while your incremental conversions tank.

The fix most accounts settle on: account-level negative keywords for branded terms, plus a separate brand-only Search campaign. That keeps PMax from absorbing brand traffic and inflating its own apparent ROAS.

The second leak is automated bidding running on conversions you don't actually want. If your conversion action is "all conversions" and that includes newsletter signups, downloads, low-value form fills, your bidding is optimizing toward those because they're cheap and plentiful. The spreadsheet will show low CPA across the board and the revenue line will quietly underperform. Audit your conversion action settings before you trust any allocation math, and if you're seeing reps push aggressive Target CPA on accounts without enough conversions, our piece on the 30-conversion floor for tCPA is worth reading first.

From what I've seen, accounts that pair this five-bucket spreadsheet with a quarterly conversion action audit catch about 80% of the budget waste that would otherwise get blamed on creative or seasonality.

Two questions I get asked about this every quarter

Should I run a shared budget across multiple campaigns or keep them separate?

Keep them separate unless you have a very specific reason. Shared budgets simplify the management view but they hide the spend-share drift this whole spreadsheet is designed to catch. If campaign A and campaign B share a budget and A is quietly eating B's allocation, your spreadsheet will show one combined line and you'll miss it. The only place I'd use a shared budget is for a small experimental cluster where you genuinely want Google to pick the winner, capped tightly.

How do I decide between Target CPA and Maximize Conversions when I'm reallocating budget?

If a campaign has fewer than 30 conversions in the last 30 days, leave it on Maximize Conversions. Target CPA needs volume to learn against, and pushing it on a thin campaign just produces erratic delivery. Once you cross 30, switch to Target CPA at roughly 1.1x your current actual CPA, then ratchet down 10% every two weeks if the CVR holds. Don't drop the target by more than 15% in any single change or Smart Bidding pulls back on impressions hard and you lose the volume you were trying to optimize.

The honest closer

Budget allocation isn't a Q1 exercise you do once and revisit at planning. It's a weekly check-in with five numbers and a 25% reallocation cap. The accounts that compound aren't the ones with the prettiest splits. They're the ones where someone actually opens the spreadsheet on a Friday afternoon, looks at where contribution share and spend share have drifted apart, and moves a small amount of money. Then closes the laptop.

The hard part isn't the math. It's resisting the urge to redo everything when one week looks bad.

Notice Me Senpai Editorial