Smart Bidding Exploration reaches Shopping and Performance Max

Google's Ads help centre now carries a page describing how Smart Bidding Exploration works across Search, Performance Max and Shopping campaigns, including a single toggle that applies exploration to both Search and Shopping traffic in a Performance Max campaign with a Merchant Center feed. The feature was announced for Search campaigns at Google Marketing Live in May 2025, and Google's June announcement this year extended it to all Performance Max campaigns without a product feed and opened a beta for the retail versions.

The mechanics are the same on every campaign type. You pick a Target ROAS tolerance between 5% and 30%, and Google lowers your effective target by up to that amount so the bidder can pay more for searches it thinks could convert but has not had room to test. For an Australian retailer heading into Black Friday, that is a margin decision, and the number to check first is break-even ROAS.

Exploration raises bids but leaves broad match keywords alone

Google's page is clear about what exploration does and does not touch. It changes bidding only. The help page states that the feature "doesn't expand your broad match keywords or target outer limits." Google's own analogy is a fishing pond: your keywords, dynamic search ads and URL expansion set the size of the pond, and exploration is better gear for catching fish you were already allowed to catch but missed because the ROAS limit was too tight.

Coverage differs by campaign type. Search campaigns get exploration on search traffic. Performance Max campaigns without a product feed get it on Search traffic only. Performance Max campaigns with a Merchant Center feed get one toggle that turns exploration on for both Search and Shopping traffic, and standard Shopping campaigns are on the eligible list as well. Google also says the feature works alongside AI Max on Search and Shopping campaigns, with AI Max widening the queries a product can appear for and exploration bidding more freely on them.

Google's Target ROAS tolerance example turns 200% into 180% effective target

Google's worked example uses a 200% Target ROAS and a 10% tolerance. Opting in means accepting a return 10% below the target, so the effective target becomes 180%. Google then says you should expect to land within roughly 15% of that 180% when you judge performance over one to two conversion cycles.

The tolerance is the discount you agree to, and the 15% is the variance Google puts around the discounted figure. On the example above, a campaign could settle anywhere from about 153% to 207% and still be performing as documented. At the top of the range, a 30% tolerance on a 200% target gives an effective 140%, and 15% under that is 119%.

Google's published results for the feature so far come from Search. Its Accelerate page says Search campaigns using exploration saw on average 27% more unique converting users, based on global internal data for Search text ads from 1 January to 20 February 2026. The June announcement repeated the figures from the original 2025 launch: an 18% lift in unique search query categories with conversions and 19% more conversions overall, from internal data covering 11 March to 11 April 2025. Those are Google's averages, not a forecast for any one account. Google has not published Shopping results.

Three rules gate exploration eligibility

The eligibility rules narrow the field more than the headline suggests.

• Campaigns must use Target ROAS bidding. A lead-gen account on Target CPA, or a store on Maximise conversion value without a target, has no tolerance to adjust.
• The campaign should not be limited by budget. Google says a budget cap stops the algorithm exploring new traffic, and if a campaign hits its limit the system may pause exploration for a while.
• Portfolio bid strategies, in Google Ads or Search Ads 360, only work with exploration on Search campaigns. Shopping and Performance Max campaigns need a campaign-level bid strategy, and moving one of them into a portfolio switches exploration off automatically.

That last rule is the one most likely to catch an agency-managed account, because a shared portfolio target is a common way to run several retail campaigns against one ROAS goal. Exploration is a setting that only means something inside Target ROAS, in much the same way that device bid adjustments under Target CPA shift the target rather than the bid.

Ginny Marvin confirms Shopping exploration fully launched despite Accelerate lag

The new help page is silent on one point: whether every retail advertiser can use it yet, or only those admitted to the beta. Google's June announcement extended exploration to all Performance Max campaigns without a product feed, and opened a separate beta for standard Shopping campaigns and Performance Max campaigns with a product feed, with advertisers told to contact their Google account team to join.

Google's Accelerate page for the feature, last updated in September, still describes the Shopping and feed-based Performance Max rollout in the future tense. The help page describes the retail versions as if they are simply part of the product. Google has since settled it. On 16 October its Ads Product Liaison, Ginny Marvin, wrote in the Google Ads newsletter Ads Decoded that Smart Bidding Exploration is now fully launched for Performance Max campaigns with a product feed and for Shopping campaigns, and that it requires uncapped budgets to function properly. The Accelerate page has not caught up. Launches can take time to reach every account, so the practical check is the campaign settings of a Shopping or feed-based Performance Max campaign that already bids on Target ROAS.

A 350% target carries 16% but 280% carries none

The decision is arithmetic, and it starts with your gross margin. Break-even ROAS is 1 divided by the margin you keep after product cost, shipping and payment fees. A store that keeps 40% of each order breaks even at 250% ROAS. A store that keeps 25% breaks even at 400%.

Because Google's 15% band sits under the discounted target, the test is whether target x (1 - tolerance) x 0.85 still clears break-even. Rearranged, the most tolerance a campaign can carry is 1 - break-even / (0.85 x target). In a hypothetical store, not a client result, with a 40% margin (a 250% break-even) and a 350% Target ROAS, that works out to about 16%. Anything up to that keeps the bottom of Google's range above break-even, and 20% or 30% does not.

Against the same 250% break-even, a 600% target could carry the full 30%. A 280% target has no headroom at any setting, which is a pricing and margin problem, not a bidding one.

Smart bidding also needs conversion volume to learn from. The constraint we found when comparing single keyword ad groups with smart bidding applies here too: a campaign that produces a handful of orders a month gives the bidder very little to explore with, and one to two conversion cycles on a slow-moving product can run into weeks.

Which campaign gets exploration first?

Black Friday falls on 27 November, about five weeks away. Google's own holiday guidance for Shopping and Performance Max says to set budgets at around three times expected spend so campaigns have headroom, to make target changes a few days before a key promotion day, to move targets gradually in steps of around 30% with one to two conversion cycles between changes, and to use seasonality adjustments for short sales where conversion rate will jump. It also notes campaigns can spend up to twice the average daily budget when demand rises.

Two of those holiday rules connect directly to the exploration page. Exploration may pause when a campaign hits its budget limit, and Google's holiday advice already assumes far more budget headroom than usual, so a Shopping or Performance Max campaign showing Limited by budget is not ready for the setting. Google's advice to move targets in steps, with one to two conversion cycles between changes, also argues for a single exploration test rather than several: one Target ROAS campaign with steady sales, a tolerance chosen with the formula above, and no other target changes until those cycles have passed. The extra bids go to queries the campaign has little history on, so feed accuracy matters more than usual while the test runs. Google's change to its Merchant Center image rules means poor images no longer block products, with a 500 by 500 pixel minimum arriving on 31 January 2027.

If the setting has not reached your account yet, give the rollout a few days or ask your Google rep. Retail campaigns need to stay on campaign-level bid strategies, because Google switches exploration off for any Shopping or Performance Max campaign moved into a portfolio.