Product Value Optimization in PMax and Shopping 2026: Bidding Toward Margin

Your feed holds 4,000 products. Fifty of them carry a 45% margin, three hundred carry 6%, the rest sit somewhere in between. To Smart Bidding they are identical: a €100 order is a €100 order, whether you keep €45 or €4 of it. The algorithm dutifully pushes budget toward whatever converts most easily — and that is routinely your thinnest-margin catalogue. In mid-September 2026 Google opened the Product Value Optimization beta: product-level value multipliers that steer bidding toward the SKUs that actually pay you, without restructuring a single campaign.

This guide covers what the tool does, how it stacks with tROAS and conversion value rules, how to prepare margin data your feed can carry, a five-step setup, which multipliers to start with, how to verify the result, and when you should not use it at all.

The problem it solves

The standard chain looks correct: the feed sends product price, the tag sends order revenue, a tROAS strategy optimises against that revenue. Every link is right, and the outcome is still wrong, because revenue and profit are distributed differently across a catalogue.

The retail pattern is familiar. Accessories and consumables convert easily, small basket, fat margin. Big-ticket electronics convert slowly, large basket, razor-thin margin. A 400% tROAS target is equally happy with both — and thin-margin, high-ticket sales produce excellent ROAS alongside mediocre profit.

Until now there were three workarounds, each with a cost:

• Split products into campaigns with different tROAS targets. Works, but fragments your data: each campaign learns from its own slice of conversions, and small campaigns starve for signal.
• Send profit instead of revenue in the tag. Mathematically honest, but it breaks finance reconciliation and revenue reporting.
• Conversion value rules. They operate on audience, device and location — never on the individual product.

Product Value Optimization fills the missing fourth option: campaign structure stays put, tag revenue stays put, and you simply tell the system that a given group of products is worth more to the business than its price suggests.

What the tool does — and what it does not

The mechanics: you define rules that adjust conversion value in real time for specific product segments. Automated bidding then optimises against the adjusted value and shifts impressions toward the priority SKUs.

• Supported in Performance Max and Shopping campaigns.
• Segments are defined by product attributes — brand, category, custom labels.
• It does not replace tROAS. Campaign tROAS sets the average return you want; Product Value Optimization says which products inside that campaign deserve more than average.
• It does not change actual revenue in your books. It changes the signal the algorithm bids on.
• It does not fix bad data. If your feed lies or conversion values arrive incomplete, multipliers simply multiply the error.

That last point deserves emphasis. Before layering multipliers on top, confirm that base conversion values arrive correctly and completely — the diagnostic sits in our guide to supplemental conversion data sources.

How Product Value Optimization stacks with the other value tools

Tool	Level	What it controls
Campaign tROAS	Campaign	Average target return across the campaign
Conversion value rules	Audience, location, device	A multiplier for who converted and from where
Product Value Optimization	Product / feed segment	A multiplier for what they bought
Sending profit in the tag	Tag	Replaces the optimisation base entirely

These compound rather than compete, which is exactly where the risk lives: a 1.4× value rule plus a 1.5× product multiplier equals 2.1×, and the algorithm now sees a sale as worth more than double reality. If value rules are already running in your account, calculate the combined coefficient before launch, not after — the setup is covered in conversion value rules and value-based bidding.

Data prep: without margin data the tool is decoration

The most underestimated work happens outside Google Ads, in your accounting system. You need margin by product, or at minimum by product segment — calculated, not estimated from memory.

What belongs in the margin calculation

• Cost of goods or production cost.
• Shipping cost, where you absorb it.
• Payment processor and marketplace fees.
• Return rate by category. In apparel and footwear this dominates everything: a 50% margin product with 40% returns nets less than a 25% margin product with 5% returns.

A practical starting point: if SKU-level margin does not exist, build four or five margin tiers by category. Tiers are less precise, but a maintained tier system beats a perfect SKU model nobody updates.

Getting margin into the feed

Rule segments are built from product attributes, and custom labels are the most flexible carrier. A structure that holds up in practice:

• custom_label_0 — margin tier: margin_high, margin_mid, margin_low.
• custom_label_1 — product role: core, accessory, clearance.
• custom_label_2 — stock turn or seasonality, where it drives priority.

Labels must refresh automatically when cost prices move — weekly at minimum. The full mechanics are in custom labels in your product feed.

And before any bidding experiment, audit feed health: disapproved products never enter the auction and no multiplier fixes that. The checklist is in Merchant Center product disapprovals.

Setup in five steps

1. Request beta access. The feature rolls out in waves — through your Google representative, or from the account interface if the beta card is surfaced to you.
2. Calculate margin by segment. Category, brand, tier — at whatever granularity you can genuinely maintain.
3. Label products in Merchant Center. Apply the custom-label structure above, then verify the labels reached the feed and products landed in the expected buckets.
4. Set multipliers. Baseline 1.0; priority segments 1.2–1.5; a reducing multiplier on thin-margin lines. Start narrow — one or two rules, not twelve.
5. Observe for 14 days. Watch how impression share redistributes across product segments before expanding the rule set.

Which multipliers to start with

A directional rule, not a formula: the multiplier should roughly reflect the ratio of segment margin to catalogue average margin. If your average margin is 20% and a segment runs 30%, something around 1.3–1.5 is a defensible start — and in the first 30 days take the low end of the range.

What to avoid:

• Setting 3.0 “to be sure”. Extreme multipliers during learning wreck pacing: budget lurches to one segment and parts of the campaign go dark.
• Editing multipliers every three days. Every change is a new learning signal — the same logic that governs the Smart Bidding learning period.
• Covering 100% of the catalogue with rules on day one. If everything is multiplied, nothing is prioritised.

How to tell it worked

This is where self-deception is easy: campaign ROAS may drop after implementation, and that can be a success. The goal was never ROAS — it was profit.

1. Impression redistribution

Did impression share and clicks rise for priority segments? That is the first proof the rules are being applied at all, usually visible within 7–14 days. Segment-level reporting is your check here, alongside Performance Max channel reporting to confirm the budget did not simply migrate between surfaces. If AI dashboards are live in your account, that segment cut is one prompt away — just verify the conclusion in a standard report.

2. Gross profit, not revenue

Calculated outside Google Ads: export paid orders for the period, multiply by SKU margin, subtract spend. Profit up at flat spend means the tool is working, even with ROAS down.

3. Basket composition

Check whether low-margin traffic drivers that recruit new customers have been squeezed out. That is a genuine risk: optimising the margin of a single transaction can cut off a stream of high-LTV customers. If you have repeat-purchase data, fold it into the calculation — the method is in POAS versus ROAS.

A worked example

Abstract “1.2–1.5 multipliers” make more sense against a real catalogue. Take a retailer spending €20,000 a month at a 22% average catalogue margin.

Segment	Margin	Share of spend	Share of gross profit	Multiplier
Accessories (core, high margin)	42%	12%	28%	1.5
Main catalogue	24%	38%	44%	1.1
Large appliances	8%	34%	21%	0.8
Clearance / end-of-line	5%	16%	7%	0.6

Read the two middle columns against each other. Large appliances absorb 34% of spend and return 21% of profit — overvalued by the algorithm, because a high ticket produces flattering ROAS. Accessories return 28% of profit on 12% of spend — undervalued, because a small basket looks unimpressive in revenue terms.

These multipliers are not derived from a formula. They are chosen to shift spend toward profit without zeroing any segment. Note the 0.6 on clearance: you cannot switch it off, because end-of-line stock still has to move, but paying market rate for it makes little sense.

What to expect: roughly 10–20% of spend redistributing between segments over two to three weeks — a directional benchmark for a mid-sized account, not a promise. If nothing moves at all, check that your labels actually reached the feed and that segments carry enough conversion volume to matter.

Seasonality, promotions and stock

Margin is not the only priority signal. There are periods when turnover or clearing warehouse space matters more, and the multiplier scheme should follow the calendar.

• Before a seasonal peak. Raise multipliers only on products you can actually supply in volume. Driving demand into something that sells out in a week leads straight to disapproved items and wasted spend.
• During clearance. Priority shifts from margin to sell-through: last season’s stock may deserve a temporarily elevated multiplier despite thin margin.
• After the peak. Restore the baseline scheme. Forgotten holiday multipliers are a classic cause of bizarre January spend distribution.
• On stock shortages. A reducing multiplier is gentler than excluding the product entirely: impressions fall but the product stays in the learning pool.

The working rule: review the multiplier scheme quarterly, plus targeted edits before major promotions. More often and the algorithm never settles; less often and your multipliers drift away from current cost prices.

Common mistakes

1. Multipliers built on stale margin. Cost prices moved, labels did not. Two months later you are aggressively promoting what stopped being profitable.
2. Double counting with value rules. Compute the product of all active multipliers, not each in isolation.
3. Rules on segments with no volume. A segment with three conversions a month teaches nothing; consolidate thin segments.
4. Ignoring returns. High nominal margin plus high return rate is a trap, especially in fashion.
5. Changing tROAS at the same time. Then you cannot attribute the result. One variable at a time.
6. Reducing multipliers on brand products. Brand queries are cheap and convert well; devalue them and you may hand your own traffic to competitors.
7. No reconciliation with finance. Adjusted value lives in the ad account; real money lives in your ledger. Reconcile monthly.

What happens inside the auction

The mechanics are worth understanding, because they explain why the effect is gradual and why extreme multipliers backfire.

A value-based strategy solves one problem: for every available auction, estimate the probability of a conversion and its expected value, then bid on the product of the two. Product Value Optimization touches only the second term — the expected value of that specific product. Conversion probability stays whatever the model infers from history.

Three consequences follow.

• A multiplier does not make a product convert. If a SKU converts poorly, inflating its value simply makes the system pay more for the same conversion volume. Check segment conversion rate alongside margin.
• The effect compounds slowly. The system needs data under the new value distribution; early-day spend swings should not be read as results.
• Budget is still the binding constraint. If the campaign is budget-capped, raising priority-product value displaces other products rather than adding reach. Sometimes that is exactly what you want — but it should be a decision, not a surprise.

One more nuance specific to Performance Max: redistribution can happen across serving surfaces as well as across products — budget migrating from Shopping into Display, for example. Keep a channel-level view next to your segment reports, or you will mistake a surface migration for a successful margin shift.

Who does not need this

• Flat margin across the catalogue. If category spread is a few percentage points, the effort does not pay back.
• Fewer than 50 products. Manual campaign structure is simpler and just as effective.
• No reliable cost data. Without it you are scaling a guess.
• Manual CPC or Maximise Clicks. The tool works through conversion value, so it assumes a value-based strategy.
• Lead generation with no product feed. This is retail tooling; for services the same job is done by value rules and lead-type segmentation.

If you do not have beta access

The workaround is old and it works: split the catalogue into margin tiers across separate campaigns or ad groups with different tROAS targets. The cost is data fragmentation, and it is real — a campaign with few conversions learns slowly.

The compromise that usually wins at mid volume is two campaigns rather than six: “high margin” with a looser tROAS and “everything else” with a tighter one. Fragmentation stays moderate and you capture most of the effect.

Meanwhile, prepare the data: segment margin, current custom labels, automated refresh. When the beta reaches your account, rollout becomes a day of work instead of a month.

An algorithm optimises precisely the number you hand it. Feed it revenue and it will maximise revenue — including the revenue that leaves you four euros a sale.

FAQ

What is Product Value Optimization in plain terms?

It lets you tell Google Ads that some products in your feed are worth more to the business than others, and get bidding redistributed toward them — without restructuring campaigns or replacing revenue in the tag.

Which campaign types support it?

Performance Max and Shopping campaigns. Search campaigns without a product feed are out of scope.

Does it replace target ROAS?

No. tROAS still sets the average return target for the campaign; Product Value Optimization operates on top, reprioritising products within it.

How are product segments defined?

By feed attributes — brand, category, custom labels. The most flexible approach is to pre-label products by margin tier using custom labels.

Which multipliers should I start with?

Baseline 1.0, priority segments 1.2–1.5. Treat that as directional: the multiplier should reflect the real ratio of segment margin to average margin. Avoid extreme values in the first 30 days.

Will this change revenue figures in my reports?

Conversion value in Google Ads will be adjusted, so it can no longer be reconciled directly against finance. Calculate true revenue and profit in your accounting system.

How long until I see an effect?

Impression redistribution across segments is usually visible in 7–14 days. Profit impact should not be judged before a month has passed and returns have settled.

What if conversion value rules are already active?

Multipliers compound. Calculate the combined coefficient for each scenario in advance, or you will easily inflate value two- or threefold.

Should I adjust tROAS after implementing?

Not simultaneously. Give the strategy two to three weeks, review the actual distribution, then adjust the target if it is still warranted.

Can I reduce the value of low-margin products?

Yes, reducing multipliers are allowed. Be careful with traffic drivers: they can lose money on the first sale and earn it back across the customer lifetime.

What should I do without beta access?

Split the catalogue into two margin-tier campaigns with different tROAS targets, and get margin data plus custom labels in order in parallel. Migration then becomes immediate.

How do I prove it improved things?

Measure gross profit from paid traffic over the period, not ROAS. Falling ROAS with rising profit is the intended outcome.

Where to start this week

Export margin for your top 100 products and compare it against where spend actually goes — the top-by-spend and top-by-profit lists usually overlap far less than anyone expects. Then: apply custom labels, automate their refresh, request beta access, and in the meantime test a two-campaign margin split.

If the constraint is account access and stability rather than data, PPC Rebels runs Google Ads agency accounts for high-volume advertisers; the rest of the service list is on the PPC Rebels homepage.