PMax Budget:
Fund It Right.

Why PMax budget is different (auto-allocation across asset groups)

In a traditional Google Search campaign, you control the flow of capital with surgical precision. If your "Leather Boots" ad group is performing well, you increase its budget. If "Sandals" are tanking in October, you cut them. You are the conductor of the orchestra.

Performance Max is a different beast. You set a budget at the campaign level, but Google's AI becomes the conductor. It decides how that budget is distributed across asset groups and, critically, across placements. PMax is an "all-or-nothing" play that spans Search, Shopping, YouTube, Display, Discover, Gmail, and Maps.

This means your budget isn't just buying clicks; it is buying probabilistic data across the entire Google ecosystem. When you set a £200/day budget, you are telling Google: "Find me buyers wherever they are." If the AI finds that Shopping is hitting your ROAS target but YouTube is slightly under, it will only spend on YouTube if there is excess budget that isn't needed to capture the Shopping demand.

Because you cannot manually shift budget between Asset Groups, your only structural lever is campaign segmentation. This is why the architecture of your account is the most important decision you make. You must decide which products "share" a budget and which products deserve their own dedicated capital. This is a core pillar of our Google Ads eCommerce guide.

Minimum viable PMax spend

There is a technical "floor" to Performance Max that many advertisers ignore. Because it relies on Smart Bidding, it requires conversion density to stay stable.

Google's documentation suggests 30 conversions in 30 days. In our experience with UK retail, that is the bare minimum for survival, not success. For a campaign to be truly stable-meaning it doesn't have "wild" swings in ROAS from one week to the next-you need closer to 50 conversions per month.

Let's look at the math for a mid-market UK brand:

This is why we set our Minimum Viable Spend at £2,000 - £3,000 per month per campaign for standard retail. Below this, the machine is in the "Starvation Zone."

In the Starvation Zone, the machine lacks enough signal to differentiate between a lucky conversion and a repeatable trend. It starts chasing "ghosts"-it sees one conversion from a random Display ad and shifts all your budget there for 48 hours, getting zero results, before shifting back. This cycle of "re-learning" is what kills your profit margin.

If you have a total budget of £1,000/month, you shouldn't be using PMax. You will get much better results with Standard Shopping, where you can use negative keywords and manual bids to protect your spend.

Budget per campaign - split vs consolidate

The "Order Mistake" is the most common reason for PMax failure. Advertisers want their account to look neat, so they create 10 campaigns for 10 product categories.

Data is the currency of AI. By splitting your budget, you are devaluing your currency. One campaign with 150 conversions per month is infinitely "smarter" than five campaigns with 30 conversions each. The machine with 150 conversions has enough data to understand placement quality, audience overlap, and time-of-day trends. The machines with 30 conversions are just guessing.

Our "Consolidation Bias" means we always start with the smallest number of campaigns possible. We only split when there is a commercial necessity, such as radically different profit margins or different seasonal targets.

When to actually split your budget:

• Target ROAS variance: If your Electronics have a 4x ROAS target and your Accessories have a 10x target, they must be in separate campaigns.
• Regional budgets: If you have strict budget caps for UK vs. US vs. EU.
• Brand protection: If your brand search terms are eating 80% of your generic PMax budget and you want to force spend into "Cold" audiences.

Remember: Consolidated signals beat structured beauty. Use Asset Groups for your creative themes, but keep the budget together for the signals.

Budgeting for High-AOV vs Low-AOV

The average order value (AOV) of your products should directly dictate your minimum budget. A campaign selling £1,000 sofas needs a much larger budget to "learn" than a campaign selling £20 t-shirts. Why? Because the path to conversion for a high-ticket item involves more touchpoints, more research, and more time.

In the PMax world, this translates to longer attribution windows and slower signal gathering. For high-AOV brands, we recommend doubling the standard minimum viable spend to £5,000/month per campaign.

When you have a low budget and high AOV, the algorithm often gets "impatient." It might see no conversions for three days and drastically cut bids, just as your potential customer was about to return and buy. This creates a cycle of failed conversions. If you cannot afford the £5k/month floor for a high-AOV product, consider using "Maximise Conversions" without a target ROAS initially to force the machine to buy enough data to find those rare but valuable buyers.

Low-AOV items, on the other hand, can often function on smaller budgets because the volume of transactions is higher, providing the machine with more frequent feedback loops. However, even with low AOV, the £2k monthly floor remains the point where we see the most significant jump in ROAS stability.

ROAS-driven budget rebalancing

Once your campaigns are out of the 30-day learning phase, you enter the "Performance" cycle. This is where you rebalance capital from losers to winners.

The rule at Visionary Marketing is the 15% Capital Migration. We never "slash" budgets. We "migrate" them. If Campaign A is at 6.0x ROAS and Campaign B is at 3.0x, we move 15% of B's budget to A every 30 days.

Daily vs monthly behaviour

Google operates on a 30.4-day average. This is the most common point of frustration for UK finance managers. Your "£100/day" campaign will spend £200 on Black Friday and £40 on a random Sunday.

The AI is smarter than you at day-parting. It knows when your customers are likely to buy. If you try to manually "throttle" the budget on slow days, you are simply denying the machine the data it needs to know when not to bid.

Budget shifts within the learning phase

The "Day 1 Full Spend" myth is dangerous. When you launch a new PMax campaign, the algorithm is in "Discovery Mode." It doesn't know what works yet, so it tests everything.

If you give it £10,000 on Day 1, and you have a technical error-like a broken checkout or a Merchant Center feed disapproval-you can lose £2,000 in 24 hours.

The Visionary Soft Launch: We recommend starting at 50% of your target budget for the first 7 days. Once you see the first 10 sales come through and you verify the attribution data, only then do you ramp up to 100%. This "Soft Launch" approach protects your capital while the machine is at its most volatile.

Our internal tracking shows that accounts that "ramp" their budget over 14 days reach their target ROAS 25% faster than those that start at full throttle.

The Portfolio Bid Strategy Lever

One advanced tactic for managing PMax budgets is the use of Portfolio Bid Strategies. Normally, each PMax campaign has its own budget and its own target ROAS. However, you can group multiple PMax campaigns into a single Portfolio Strategy.

This allows campaigns to "share" a target ROAS, which can help stabilise performance if one campaign has a low-volume week. While they don't share the budget itself (budget remains at the campaign level), the shared target ROAS prevents the machine from over-optimising for a single campaign's noisy data.

We typically implement Portfolio Strategies when an account has more than 5 PMax campaigns. It acts as a "governor" on the system, ensuring that the budget rebalancing we discussed earlier doesn't lead to wild swings in overall account efficiency. It is the bridge between granular control and machine learning stability. For larger UK retailers, this is often the "secret sauce" for maintaining a 400%+ ROAS at scale.

When to increase budget (and by how much)

Scaling a winning PMax campaign is an art. We use the 20/30/50 Rule:

Before you scale, check your "Impression Share Lost to Budget." If it's 0%, increasing your budget will not increase your sales. It will only force the machine to bid on "junk" traffic to spend your money. If you are at 0% IS loss and you want to scale, you need to broaden your audience signals or add new product categories.

When to decrease budget (or pause)

When a campaign is failing, the human instinct is to "dial it back." In PMax, this is often the kiss of death.

Reducing the budget on a failing campaign makes the data set even smaller, which makes the machine even "dumber." It enters a downward spiral.

The Rule: If a campaign has been below target for 60 days, and you've already tried refreshing the assets and adjusting the ROAS targets: Pause it. Don't starve it. Take that budget and consolidate it into your winners, or launch a completely fresh campaign with a new asset structure. Starting from zero is often better than trying to "fix" a polluted learning history.

Common budget mistakes

For a deeper dive into PMax management beyond budget, see our full Performance Max agency guide.