Google PMax in 2026: Asset Groups, Signals and Profit Control
TL;DR
A practical performance max strategy 2026 guide: asset groups, audience signals, AI Max and profit controls that actually move ROAS for D2C and B2B.
Most Performance Max accounts I audit in 2026 share the same problem: the campaign is technically running, spend is going out the door, and nobody on the team can actually explain why it allocates budget the way it does. PMax has eaten a larger share of Google Ads budgets every year since it launched, and with the rollout of AI Max controls and agentic optimisation, the temptation to "set it and let Google figure it out" is stronger than ever. That is exactly how brands end up paying premium CPCs to re-buy customers who would have converted anyway.
Here is the one-sentence version for anyone skimming: a winning performance max strategy 2026 treats PMax as a profit-allocation engine you steer with first-party signals and creative inputs, not a black box you feed money into and hope. The levers that move the number are now audience signals, asset group structure, and the conversion value you teach the algorithm to chase. Everything below is how we actually run it for D2C and B2B accounts spending anywhere from a few lakh to several crore a month.
Why PMax behaves differently in 2026
Two shifts changed the game. First, Google has folded its broader AI advertising features (marketed as AI Max for Search and increasingly extended into PMax inventory) into the bidding and matching layer, meaning the system now decides not just bids but query interpretation, asset combinations, and placement mix with far less manual override. Second, the death of third-party cookies and the maturing of Google's Privacy Sandbox mean the model leans heavily on your first-party data and its own logged-in signals to find buyers.
The practical consequence: the quality of your inputs now determines the quality of your output more than at any point before. Google itself has consistently reported that advertisers feeding strong first-party data into Smart Bidding see materially better performance, and that gap has only widened. If you hand PMax weak conversion signals and one generic asset group, agentic optimisation will confidently optimise toward the wrong thing.
Asset groups: structure by intent, not by SKU
The single most common structural mistake is one giant asset group stuffed with every product. PMax uses asset groups as themes for matching creative to audiences, so collapsing everything removes the algorithm's ability to differentiate. But the opposite extreme, one asset group per SKU, starves each of conversion volume and slows learning.
Structure asset groups around buyer intent and margin tiers instead:
- Hero / high-margin products - your best contribution-margin items get their own group with dedicated creative and the most generous targets.
- Category clusters - group products a customer shops interchangeably (e.g. all protein powders for a D2C nutrition brand) so PMax can cross-sell within a coherent theme.
- New / launch SKUs - isolated so you can monitor whether they are getting served at all, since PMax tends to starve new products in favour of proven winners.
- Prospecting vs. retention - for B2B/SaaS, separate groups for cold demand-gen assets versus content for warm pipeline.
Each asset group needs the full asset load: 5 headlines minimum (15 ideal), multiple long headlines and descriptions, square and landscape images, logos, and at least one video. If you do not supply video, Google auto-generates one, and the auto-generated versions are usually mediocre. Treat that as a hard requirement, not an optional extra.
Audience signals are hints, not targeting
This is the concept most teams get wrong. Audience signals do not restrict who PMax reaches; they tell the model where to start looking. The algorithm uses them to seed, then expands. So the goal is to give it the strongest possible seed.
The hierarchy of signal strength we use, best first:
- Your highest-LTV customer list (Customer Match) - actual purchasers, ideally segmented by value.
- Recent converters and cart/checkout abandoners from your own site data.
- High-intent in-market and custom segments built around competitor and category search behaviour.
- Broad affinity/demographic signals - only as a fallback.
For Indian D2C brands, the practical move is to push enriched first-party data, phone-number and email-based Customer Match, into PMax and keep it fresh. Post-cookie, that list is your durable advantage. A static list uploaded once and forgotten decays fast; we refresh value-segmented lists at least monthly.
Profit control: stop optimising to revenue
PMax will happily maximise conversion value, and if you feed it order revenue, it will chase your lowest-margin, highest-discount, easiest-to-sell orders. That is how accounts post a "great" 6x ROAS while the P&L bleeds. The fix is to teach the algorithm what a customer is actually worth to you.
Use profit, not revenue, as the value signal
Where your setup allows, pass contribution margin (or a margin-adjusted conversion value) instead of gross revenue. Even a simple approximation, revenue minus COGS minus shipping, dramatically changes which orders PMax prioritises. For brands with wide margin spread across SKUs, this is the highest-leverage change you can make all year.
Layer new-customer acquisition value
Google's new-customer acquisition goal lets you assign extra conversion value to first-time buyers. For D2C brands where the first order is a loss-leader and profit lives in repeat purchase, this aligns PMax with LTV rather than one-off transactions. Klaviyo and other retention platforms have long shown that repeat buyers drive a disproportionate share of e-commerce revenue, so paying to acquire them is the point.
How PMax fits the wider 2026 channel mix
PMax does not run in a vacuum. With Meta CPMs climbing amid signal loss and its Andromeda retrieval system reshaping delivery, and with AI search surfaces, Google's AI Overviews now appear on a large share of queries per Ahrefs and similar studies, plus ChatGPT, Perplexity and Gemini becoming genuine discovery channels, the allocation question matters as much as the in-platform tactics. Here is how we frame the trade-offs.
| Lever | What it controls | Best for | 2026 watch-out |
|---|---|---|---|
| Asset group structure | Creative-to-audience matching by theme | Multi-category catalogues | New SKUs get starved without isolation |
| Audience signals | Where the model seeds its search | First-party-data-rich brands | Stale lists decay post-cookie |
| Value rules / profit signal | Which conversions get prioritised | Wide-margin catalogues | Revenue-only feeds chase low-margin orders |
| New-customer goal | Acquisition vs. repeat weighting | D2C with strong LTV | Misfires if retention data is weak |
| Search themes / AI Max | Query interpretation and expansion | Capturing emerging intent | Less manual control, needs monitoring |
The reporting and guardrails that keep it honest
Agentic optimisation removes manual levers, so your discipline has to move to inputs and monitoring. The non-negotiables we set up on every account:
- Brand traffic exclusion - use brand exclusion lists so PMax cannot claim credit for branded searches that would convert organically. This alone often reveals the true incremental ROAS.
- Channel-level transparency - pull the asset group and channel performance via scripts or the API so you can see Search vs. Shopping vs. Display vs. video allocation, which the default UI still partially obscures.
- Search term and asset insights review - weekly, to catch irrelevant query expansion and to retire fatigued creative.
- Geo and device value adjustments - for India specifically, tier-1 vs. tier-2/3 buyers convert and repeat very differently; reflect that in value rules.
Creative remains the primary lever. When bidding and targeting are increasingly automated, the input you most directly control is the asset itself. The accounts that win in 2026 are the ones shipping more, sharper creative variations into well-structured asset groups. If you want a partner to architect and run this end to end, our Google Ads management team builds exactly this kind of profit-first PMax structure.
FAQ
Should I still use Performance Max in 2026?
Yes, for most e-commerce and lead-gen accounts PMax is now the core of Google Ads delivery. The question is not whether to use it but how to steer it. With profit-based value signals, well-structured asset groups, and fresh first-party audience signals, PMax performs strongly. Run without those inputs and it quietly optimises toward low-margin, low-incrementality conversions.
How many asset groups should one PMax campaign have?
There is no fixed number, but structure by buyer intent and margin tier rather than by individual SKU. Most mid-sized D2C catalogues run well with three to eight asset groups: hero products, category clusters, new launches, and separate prospecting versus retention themes. Too few removes differentiation; too many starves each group of the conversion volume it needs to learn.
Do audience signals limit who PMax reaches?
No. Audience signals are seeds, not targeting restrictions. PMax uses them to decide where to start searching for buyers, then expands beyond them. That is why feeding your strongest first-party data, value-segmented Customer Match lists and recent converters, matters so much. A weak or stale signal sends the algorithm looking in the wrong place from day one.
How do I make PMax optimise for profit instead of revenue?
Pass contribution margin or a margin-adjusted conversion value to Google instead of gross order revenue, even a simple revenue-minus-COGS-minus-shipping figure works. Layer in the new-customer acquisition goal so first-time buyers carry extra value where LTV justifies it. This shifts PMax away from chasing easy, discounted, low-margin orders toward the customers your P&L actually wants.
Build a PMax setup that defends your margin
PMax in 2026 rewards the brands that treat it as a steerable profit engine and punishes the ones that treat it as autopilot. The work is upfront, in the structure, the signals, and the value you teach it, and ongoing, in the guardrails and creative supply. If your current account is posting healthy-looking ROAS while your margins tell a different story, that gap is fixable. Book a call with Balistro and we will audit your PMax structure, signals, and profit controls, then rebuild it to chase the customers worth acquiring.
