Should we turn on Google's Smart Bidding Exploration for Shopping and Performance Max, at what tolerance, and how do we judge whether the extra queries it buys are profitable?
Our answer
Test Smart Bidding Exploration only on an established, uncapped, campaign-level Target ROAS campaign. Start at 10% tolerance and keep the effective target, less about 15%, above break-even ROAS. Judge after 6 weeks and 50 conversions on profit.
- Confidence
- Medium
- Strongest proof
- 13 Google rules
- 4 practitioners, 6 opinions or claims
- Sources
- 8 sources
- 23 statements
- G
ODWGMA
New here? Take the 1-minute tour
Why this is the answer
3 reasons and our judgmentGoogle sets the rules for Search, Shopping and Performance Max. Start at a 10% tolerance (5% to 30% allowed), keep the budget uncapped and run at least 6 weeks and 50 conversions. Expect ROAS within about 15% of the lowered target.
The measure that matters is traffic diversity, the unique search categories with impressions, clicks and conversions. Google calls it the key measure, read beside total conversions and conversion value. Then ask whether the extra volume was worth the lower ROAS.
Margin should set the tolerance. Dr. Web, a German web trade magazine, argues the effective target with its swing should stay above break-even. Common Thread Collective starts at 3 to 5%, and OneElevate judges exploration on incremental sales, not reported ROAS.
Our judgment
- Turn it on only where the campaign is already steady. That means a Shopping or feed-based Performance Max campaign on campaign-level Target ROAS, with an uncapped budget, a conversion history and a target it hits consistently. Test it as the one change, in the test-first order of Choosing a Google bid strategy and Setting up and reading a Google Ads experiment.
- Start at 10%, or at 5% where the margin is tight. We take Google's suggested 10% start on a 5% to 30% slider, and drop to the lowest setting when the margin leaves little room, closer to Common Thread Collective's 3 to 5%.
- Choose a tolerance only if the effective target, less about 15%, stays above break-even ROAS (1 divided by margin). In Dr. Web's example a 400% target at 20% tolerance can land at 272% to 368%. That clears break-even at a 40% margin (250%) but not at a 25% margin (400%).
- Judge on extra conversion value and profit dollars against the control arm or the weeks before. Use unique search categories with conversions and ROAS within about 15% of the effective target as checks. Confirm with Conversion Lift where the account qualifies (at least 1,000 conversions and a $5,000 study budget). The method holds at every monthly spend, since Google and the agencies work from targets, margin and conversion counts. That $5,000 study budget is the only size floor, and it applies only to the confirming test.
- Turn it off when profit falls or the budget caps it. Do so when profit dollars fall against the control or the weeks before after 6 weeks and 50 conversions, when actual ROAS settles below break-even or when the campaign becomes limited by budget. Pause it through a conversion outage.
Our judgment, built from Google, Common Thread Collective, Dr. Web, Relevant Audience, groas and OneElevate: our reading of these sources, not a finding from any one of them.
What the sources say
23 statements from 8 sourcesGoogle's help page calls Promotion Mode an open beta for standard Search and Performance Max campaigns on campaign-level Target ROAS, with a 5% to 90% tolerance for 3 to 14 days and no portfolio strategies or shared budgets.
“Promotion Mode is currently in an open beta.”
Google says the key success metric for Smart Bidding Exploration is Traffic Diversity, the unique search categories reached by clicks, impressions and conversions. It names no threshold for success.
“A key success metric for Smart Bidding Exploration is Traffic Diversity, which measures clicks, impressions, and conversions driven by unique search categories.”
Google says to enable AI Max first and let it stabilize before Smart Bidding Exploration, and not to use Exploration for keyword mining. Lowering ROAS targets often does not much increase traffic diversity.
“using Smart Bidding Exploration for keyword mining is discouraged.”
Google says to start Smart Bidding Exploration with a 10% ROAS tolerance, that it will not work if campaigns are limited by budget, and that campaign experiments can isolate its impact from seasonal shifts.
“Smart Bidding Exploration will not be effective if campaigns are limited by budget.”
Google's guide to Smart Bidding Exploration (beta) says to judge it by traffic diversity and to turn it on at least 6 weeks before major holiday sales so it can find new traffic before peak demand.
“Traffic diversity should be the key metric to evaluate success for this beta feature.”
Google says Search campaigns using Smart Bidding Exploration averaged 27% more unique converting users and 19% more conversions. Google is grading its own product.
“increases of 27% more unique converting users, and increases of 19% more conversions on average”
Google says a custom experiment shares your original campaign's traffic and budget, and is available for Search, Display, Video and Hotel Ads campaigns but not App or Shopping. Only one experiment can run at a time.
“Custom experiments are only available for Search, Display, Video, and Hotel Ads campaigns.”
Menachem Ani says Google's promotion mode on Search and Performance Max sets a temporary ROAS tolerance of 5% to 90% for a sale window, then reverts. He relays that it also turns on Smart Bidding Exploration.
“promotion mode on Search and PMAX lets you set a temporary ROAS tolerance from 5% to 90% for a sale window.”
PPC Land reports Google's Smart Bidding Exploration is fully launched for Performance Max retail and Shopping campaigns, first reported October 22, 2026. It needs uncapped budgets; Google gives no figures.
“requires uncapped budgets to function properly.”
An explainer relaying Google says Smart Bidding Exploration has no learning period but may need one to two weeks. Google's own figures, with no method, show 19% more conversions and 18% more converting search categories.
“Google says there is no learning period, but also that the system may need one to two weeks to recognise valuable new traffic, partly because of conversion delay.”
Dr. Web's editor in chief says at a 25% margin a retailer loses money below 4 euros of revenue per ad euro, so ROAS tolerance should follow margin: break-even is 400% at 25% margin and 250% at 40%.
“Die Toleranz sollte sich deshalb aus der eigenen Marge ergeben.”
Google's AI Max experiments split traffic and budget inside an existing campaign rather than copying it; campaigns on portfolio bidding, shared budgets or with text customization on can't use this flow.
“If your experiment ends without being applied to your base campaign, all AI Max settings will automatically revert to the original, pre-experiment state.”
Google's Smart Bidding Exploration lowers target ROAS by a tolerance of 5% to 30% so it can bid on high-potential queries; a 200% target at 10% becomes 180%. Campaigns should not be limited by budget.
“Campaigns using Smart Bidding Exploration shouldn't be limited by budget.”
Google says Smart Bidding Exploration raises bids only on selected traffic by accepting a lower ROAS on it, while changing the Target ROAS itself raises bids across all eligible traffic.
“Raises bids selectively on specific traffic slices by accepting a lower Target ROAS tolerance.”
On Google's Ads Decoded podcast, Tal Akavas says Smart Bidding Exploration needs uncapped, demand-led budgets to work well.
“To do so, you do need to have uncapped demand-led budgets”
Google's user-based Conversion Lift needs a Google rep, at least 1,000 observed conversions and a $5,000 minimum budget. Google allows 7-day studies but recommends more than 14 days.
“We've found up to a 17% drop in Absolute Lift in studies with a long conversion lag that run for less than 14 days.”
groas, which sells AI-run Google Ads management, warns a high ROAS target can shrink volume while ROAS looks better; lower it in 10 to 20% steps, two weeks each, and judge profit dollars.
“The cost is easy to miss because the ROAS column improves as volume falls.”
Relevant Audience, an agency selling Google Ads management, says break-even ROAS is 1 divided by margin (a 40% margin breaks even at 2.5) and that break-even is a floor, not a goal.
“a single account-wide ROAS target tends to underfund high-margin items and overfund low-margin ones. Splitting campaigns or product groups by margin band is the usual fix.”
On Google's Ads Decoded podcast, Google says advertisers using smart bidding exploration on Search text ads saw up to a 27% lift in unique user conversions. It is Google's own figure.
“advertisers that use smart bidding exploration on search to drive text ads campaigns, they saw up to 27% lift in unique user conversions”
Google's Ginny Marvin, via PPC Land, says promotion mode runs a set 3 to 14 day promotion and turns itself off, and can run alongside seasonality adjustments; Google doesn't say how the two interact.
“With Promotion mode, there's no need to remember to come back to turn it off,”
Common Thread Collective says Shopping or PMax campaigns that feel volume-constrained while hitting ROAS targets can test Smart Bidding Exploration, starting at a low ROAS tolerance of 3 to 5 percent as an experiment.
“Start with a low ROAS tolerance (3 to 5 percent) and run it as an experiment before applying broadly.”
OneElevate's Alexander Marshall says ROAS tolerance in Smart Bidding Exploration trades efficiency for reach, so its value only shows if advertisers measure the incremental sales exploration delivers.
“ROAS tolerance is an explicit trade of efficiency for reach, and it should be treated as one.”
Google's promotion mode beta for Search and Performance Max lets you schedule temporary ROAS tolerance changes and extra daily budget for demand spikes such as sales and launches. Access is through your account team.
“This feature lets you schedule temporary changes to your ROAS tolerance and add extra daily budget during peaking periods”
In the full answer
For readers who sign in- What to do this week3 steps
- The numbers behind it3 figures from cited sources
- Where each source standsA chart placing each source on the question
- The strongest disagreementThe best case against the answer, from a cited source
- How sure we areWhat its confidence rests on, and what would change the answer