Smart Bidding Exploration now covers Shopping and Performance Max. How much ROAS tolerance can your margin pay for?

KEY TAKEAWAYS

1. A tolerance is a price. The campaign has to find enough extra value to pay for the lower target on all its traffic, not just the new searches.
2. Google's 19% and 27% headline averages do not state a baseline. Against a similar target cut, its own figure is 10% more conversions.
3. The less your target sits above break-even, the faster the required lift climbs. Below a headroom of about 1.3, even 10% rarely pays.
4. Google advises six weeks of exploration before a holiday peak. For Black Friday 2026 that window has closed, so test in a quieter month.

Smart Bidding Exploration reached standard Shopping and Performance Max with a product feed in mid October. Most advice about it ends with the same line: check your break-even ROAS first. That is right, but it stops one step short.

The real question is how much extra conversion value a given tolerance has to find before it adds profit. That number depends on your margin, and it climbs faster than most people expect. This piece works it out, weighs Google's figures against the one comparison that matters, and sets out a test.

What changed in October?

Exploration is an opt-in setting for Target ROAS campaigns. You pick a tolerance and Google lowers the effective target by that share. The cut is relative, not in points: a 200% target with a 10% tolerance becomes 180%.

It launched for Search in 2025. On 15 October 2026 Google's Ads Liaison, Ginny Marvin, said it was rolling out to feed-based Performance Max, and on 16 October that it was fully launched for standard Shopping and feed-based Performance Max, according to PPC Land. PPC News Feed reported two new help pages on 22 October.

The retail rules, as reported from Google's help page by PPC Land and TechWyse:

• Tolerance runs from 5% to 30%.
• Performance Max without a feed explores Search traffic only.
• Shopping and retail Performance Max need a campaign-level strategy. Moving one into a portfolio turns exploration off.
• Campaigns must not be limited by budget. Google's FAQ says exploration is disabled in the background when a budget constrains it.

Google has published no results for Shopping or Performance Max. Every figure so far comes from Search.

Does exploration beat simply lowering the target?

Google says yes, by design. Its FAQ says a plain target cut raises bids evenly across matched queries and adds little new traffic, while exploration bids higher on selected segments. It is not a match type and does not widen targeting.

The size of the gain depends on what you compare it with. Google's headline figures are 27% more unique converting users and 19% more conversions for Search campaigns using exploration (Google internal data, global, Search text ads, 1 January to 20 February 2026). The page does not state the baseline, which reads as campaigns without exploration. The same help page gives a second comparison, against a similar target cut: 13% more converting search categories and 10% more conversions, for campaigns allowed a 10% lower ROAS. No sample size or method is published for either.

The second figure is the fair one. The alternative most advertisers have is not doing nothing. It is lowering the target, which also buys volume. And both figures count conversions, not conversion value or profit, and neither is a holdout test.

Practitioners are more cautious. Andrew Lolk of Savvy Revenue welcomed the finer control in October but said "the jury is still out" against a plain target cut, which he has not fully tested. His objection to target cuts is that they also raise bids on auctions you already win.

How much extra value does a tolerance need?

Treat the tolerance as a price. The lower target is an average across all the campaign's spend, so the extra value has to cover the cheaper return on everything, not just on the new searches.

Start with headroom: your target ROAS multiplied by your pre-ad margin. A 400% target at a 40% margin gives a headroom of 1.6. Break-even ROAS is 1 divided by the margin, the same formula behind the margin-based Target ROAS calculator Google was spotted testing on 6 October.

If the campaign hits its target before and its effective target after, profit rises only if conversion value grows by at least:

required lift = tolerance ÷ (headroom × (1 - tolerance) - 1)

That is our arithmetic. If headroom × (1 - tolerance) is 1 or less, the effective target sits at or below break-even and no lift pays.

Read it against Google's like-for-like 10%. Only campaigns with a headroom of about 2.5 or more clear that bar at a 10% tolerance. At 20% or 30%, few retail margins come close.

What does that look like for one store?

A hypothetical example, with made-up figures.

A Shopping campaign spends $20,000 a month at its 400% target, so it books $80,000 in conversion value. The store's pre-ad margin is 40%, so break-even ROAS is 250% and headroom is 1.6. Contribution after ad cost is $32,000 minus $20,000, or $12,000.

It switches on a 10% tolerance. The effective target becomes 360%. By the formula, it needs about 23% more conversion value to stand still.

All figures are our arithmetic. Even a lift the size of Google's headline 19% leaves this store $364 a month worse off, because the extra spend returns 2.4 against a break-even of 2.5. The averages hide this. Reported ROAS of 360% still looks healthy.

Will ROAS actually fall by the full tolerance?

Not always, and two sources pull in opposite directions.

Google says actual ROAS should land within about 15% of the effective target over one to two conversion cycles. It does not say whether that band is symmetrical. On the downside, a 360% effective target could run near 306%. That still clears 250%. A 500% target at a 25% margin is different: break-even is 400%, the effective target at 10% is 450%, and the low end of the band is about 383%, below break-even (our arithmetic).

On the other side, ADW Service studied about 9,900 Target ROAS changes across 201 accounts over roughly 15 months in 2025 and 2026, comparing medians before and after each change. When targets were lowered in Performance Max, volume rose 11%, cost per click rose 6% and actual ROAS fell only 3%. In Shopping, volume fell 17%. The author flags regression to the mean, and all of it predates Google's 17 August change for budget-limited campaigns.

Both can be true. A campaign that already beats its target may see realised ROAS barely move when the target drops. Plan with the effective target, then measure the drop you actually get. If realised ROAS falls 3% instead of 10%, swap that figure into the formula. At a headroom of 1.6, the required lift drops from about 23% to about 5% (our arithmetic).

Should you switch it on before Black Friday?

Probably not now. Google advises turning exploration on at least six weeks before major holiday sales, and running it six weeks before judging it. Six weeks from the 16 October launch ends on 27 November 2026, which is Black Friday. Starting today, six weeks runs to 12 December.

Peak weeks also break the test. Budgets cap out, and a capped campaign has exploration switched off. Conversion rates move for reasons unrelated to bidding. BW Digital Marketing suggests January for a clean baseline, and that fits Google's own timing advice.

Which tolerance should you choose?

Common Thread Collective suggested 3% to 5% in June. The floor reported from Google's October help page is 5%.

How do you test it?

1. Compute headroom with a mix-weighted margin. Use margin after product cost, shipping and fees, weighted by what the campaign actually sells.
2. Check eligibility. Target ROAS, no budget cap in the last 30 days, campaign-level strategy for retail, and enough volume for the 50 conversions Google wants before you judge, ideally in each arm.
3. Pick the tolerance from the table using a lift you can defend. Google's like-for-like 10% is a vendor average, so treat it as optimistic.
4. Run a campaign experiment. Google says experiments are supported for exploration in Search and Performance Max. Its exploration pages do not mention standard Shopping, so use a matched period and a control campaign there and treat the result as weaker.
5. Hold everything else still for six weeks. Exclude the first one to two conversion cycles. Do not change feeds, budgets or targets mid-test.
6. Judge on contribution, not ROAS. Per arm, compare conversion value, cost and contribution after ads. Put the realised ROAS drop into the formula.
7. Check incrementality before you scale. Conversion Lift now lists Search and Performance Max, with at least 1,000 observed conversions and a US$5,000 budget. Smaller accounts can use a geo test.

The same margin logic applies to Performance Max channel sliders and to new customer value multipliers. Each one asks you to accept a weaker average return for volume you cannot see separately.

The short version

Exploration trades a lower ROAS target for searches the campaign used to skip. Google's best like-for-like figure is 10% more conversions, measured by Google. Whether that pays depends on your headroom, so compute the lift each tolerance needs before you set it. Most retail campaigns should start at 5% or 10%, test in a quiet month, and judge the result on contribution after ad cost.

Questions

What is Smart Bidding Exploration in Google Ads?

It is an opt-in setting for Target ROAS campaigns. You choose a ROAS tolerance and Google lowers the effective target by that share, so a 400% target with a 10% tolerance becomes 360%. Google says it bids higher on selected traffic it would otherwise skip, inside your existing targeting. It does not add keywords or widen match types.

Is Smart Bidding Exploration available for Shopping and Performance Max?

Yes. Google's Ads Liaison said it was fully launched for standard Shopping and Performance Max with a product feed on 16 October 2026. Performance Max without a feed explores Search traffic only. Retail campaigns need a campaign-level Target ROAS strategy, and moving them into a portfolio strategy turns exploration off.

Is exploration better than just lowering my ROAS target?

Google's own comparison says modestly. Against a similar target cut, campaigns with exploration averaged 10% more conversions and 13% more converting search categories. That is Google internal data with no published method. Independent tests against a plain target cut have not been published, so run your own experiment.

What ROAS tolerance should I start with?

Work it out from your margin. Multiply your target ROAS by your pre-ad margin to get your headroom. At a headroom of 2, a 10% tolerance needs about 12% more conversion value to break even. At 1.5 it needs about 29%. Google suggests 10% as a start, which suits only campaigns with real headroom.

Sources

1. Google Ads Help, About Smart Bidding Exploration support.google.com
2. Google Ads Help, About Smart Bidding Exploration (FAQ) support.google.com
3. Google Ads Help, About Smart Bidding Exploration (best practices) support.google.com
4. Google Ads Help, Measure and report performance with Smart Bidding Exploration support.google.com
5. Google Ads Help, Set up Smart Bidding Exploration support.google.com
6. Google Ads and Commerce Blog, New AI-powered bidding and budgeting innovations in Search and Shopping blog.google
7. Google Ads Help, Set up Conversion Lift based on users support.google.com
8. PPC Land, Google lets Shopping and PMax campaigns lower ROAS targets by up to 30% ppc.land
9. TechWyse, Google Ads lets Shopping and Performance Max pursue more sales with flexible ROAS targets techwyse.com
10. PPC News Feed, Smart Bidding Exploration expands to Performance Max and Shopping ppcnewsfeed.com
11. Savvy Revenue, Smart Bidding rules. 5 new ones for Google Ads savvyrevenue.com
12. ADW Service, tROAS in Performance Max. The lever that doesn't work the way the market thinks adwservice.com.ua
13. TechWyse, Google Ads tests a margin-based starting point for Target ROAS techwyse.com
14. BW Digital Marketing, Smart Bidding Exploration for Performance Max and Shopping. Should you turn it on? bwdigitalmarketing.co.uk
15. Common Thread Collective, Google Ads Promotion Mode is live. Smart Bidding expansion and the August 17 change commonthreadco.com

NEXT STEP

See where your brand stands.

Your Digital Authority Score, who AI recommends for your target questions, and the placements that close each gap. Your first report is free.

Get your free report →