Flexible ROAS Targets: Google's Newest Bidding Lever and How to Test It

If your best-performing Search campaigns have plateaued and you've already squeezed out every efficiency gain you can find, there's a reason the ceiling feels permanent. Standard tROAS bidding favors higher-volume, reliable search queries and restricts exposure to lower-volume or unproven queries due to uncertainty about their performance. Over time, this creates a self-fulfilling cycle, as the system optimizes within that narrow framework. The algorithm becomes an expert at catching the same fish in the same pond - and it quietly ignores every adjacent pond that could be full of buyers. Google's answer is Smart Bidding Exploration, built around a mechanism called flexible ROAS targets. Google calls it "our biggest update to bidding in over a decade." The feature, first spotted in late September 2024 and officially unveiled at Google Marketing Live 2025 as part of the broader AI Max initiative , is now globally available. For advertisers running tROAS Search campaigns, it represents the first time you can give the algorithm explicit permission to prospect outside your proven query categories - without dismantling your campaign structure or lowering your ROAS target across the board. This guide breaks down what flexible ROAS targets actually change inside the bidding system, who should enable Smart Bidding Exploration, and exactly how to design a test that produces trustworthy results.

What Flexible ROAS Targets Actually Change Under the Hood

Most advertisers already understand standard target ROAS: Target ROAS is a Smart Bidding strategy in Google Ads that lets you set a specific goal for the revenue you want to generate from your ad spend. Smart Bidding uses auction-time bidding to predict the value of a potential conversion every time a user searches, then adjusts your bids for these searches to maximize your return. For example, if you set your goal at 500%, you're telling Google that for every $1 you spend, you want to generate at least $5 in revenue, and Google adjusts CPC bids to deliver that return.

The critical limitation is what happens at the margins. When you use tROAS, Google's bidding algorithm gets really good at knowing which search terms and audience segments will hit your goals. This is great, but it also creates a ceiling that limits campaign performance. Your campaign gets comfortable bidding on certain search queries it knows will deliver your target ROAS, but there's a whole universe of valuable traffic sitting just outside that comfort zone.

Smart Bidding Exploration introduces a controlled way to push past that ceiling. When you enable it, you're not just lowering your ROAS target across the board. Instead, you're giving Google's smart bidding a dual mandate: keep hitting your original ROAS targets on the keywords and search categories that historically perform well, while using a percentage of ROAS target tolerance (10–30%) to explore new traffic sources.

Why This Isn't the Same as Manually Lowering Your tROAS

This distinction matters enormously and trips up even experienced practitioners. Lowering ROAS targets will generally increase bids uniformly across all matched queries. This often doesn't significantly increase traffic diversity. Enabling Bidding Exploration will inform the Smart Bidding algorithms to bid higher on specific traffic segments, with the goal of capturing additional valuable performance.

One practitioner put it bluntly: when you lower your ROAS target, you're essentially telling Google to increase all your bids to get more volume. It's a blunt instrument. Smart Bidding is just chasing a new, lower efficiency target across the board. The problem is that this often just means paying more for the traffic you're already getting.

Smart Bidding Exploration changes the dynamic. With this feature, you tell Google: "Don't increase the bids for our existing auctions. Instead, take 10% (or 20%, or 30%) of my ROAS headroom and go find completely new queries." Instead of overpaying for what already works, you're allocating headroom specifically to discovery.

The Math Behind the Tolerance Slider

The setup includes a tolerance setting you control. Smart Bidding Exploration will lower your effective target ROAS. For example, if your target ROAS is 200% and you choose a 10% ROAS tolerance, the effective average target ROAS will be 180%. So if you're running a campaign at a 500% tROAS and set exploration tolerance to 10%, your effective floor drops to 450%. That headroom is where the algorithm experiments with queries it previously wouldn't have bid on.

You have the option to select a target ROAS tolerance between 5–30%. Start conservative. A 10% tolerance is enough for the algorithm to find meaningful new territory without creating stakeholder panic.

Who Should (and Shouldn't) Enable Smart Bidding Exploration

Not every account is a fit. The feature has clear prerequisites, and ignoring them will waste budget. Enable it when: - Your campaign is already performing well. This isn't for brand new campaigns with limited data. You need historical data and consistent conversion tracking for Google's systems to know what "good" looks like.

• Your campaign is not limited by budget. If your campaign is budget-constrained, just increase the budget - that'll naturally let you enter more auctions without needing Smart Bidding Exploration.
• You have room to grow. If you're already capturing super-high impression share, there's probably not much left to explore.

Avoid it when: - Your ROAS targets are already low - if so, Smart Bidding is likely in a constant state of exploration anyway. The feature becomes redundant.

• You have no more budget to spend. If you are budget-capped, exploration will find new queries that you can't afford to capitalize on. It's a complete waste.
• Demand is seasonally low. Trying to find new customers for swimwear in September is throwing money away.

Wordstream's assessment is candid: this feature "really should only be used by a select few advertisers who are seeing excellent returns on their campaigns with high conversion volume and can afford to see an incrementally lower ROAS to gain new customers." That's a narrower audience than Google's marketing suggests, and being honest about it will save you from a messy experiment.

Prerequisites: What Your Account Needs Before You Start

Before enabling the toggle, audit these foundations: Conversion tracking accuracy. Make sure your account records "Purchases" only as a primary conversion action, with all other actions including "Add To Cart" and "Page view" counted as secondary conversions. Make sure there is only one Purchase action set to Primary. Double-counted conversions will teach the algorithm to optimize for noise. Sufficient conversion volume. To use Target ROAS bidding, most campaign types need at least 15 conversions in the past 30 days. That's Google's minimum for standard tROAS. For Smart Bidding Exploration to have enough signal, you'll want substantially more - practitioners report best results with well-established campaigns that consistently hit conversion targets. Unconstrained budgets. Smart Bidding Exploration will not be effective if campaigns are limited by budget. If a campaign is limited by budget when Smart Bidding Exploration is enabled, an alert will appear to notify you. Resolve budget constraints before enabling this feature. Exploration without spending room is exploration without action. Broad targeting compatibility. It's recommended that the majority of the campaign traffic is opted into broad targeting - broad match keywords, Dynamic Search Ads, or AI Max for Search. These targeting types will give the best performance outcome along with Bidding Exploration. The feature is designed to explore longer-tail queries, which means exact match campaigns won't see much benefit. Value-based bidding maturity. Google recommends a minimum 4-week ramp-up period before activating tROAS bidding. During this period, conversion values should be uploaded consistently across at least three complete conversion cycles, giving the algorithm sufficient data to build reliable value predictions. If you haven't completed this foundational step, do that first.

Step-by-Step: How to Enable Flexible ROAS Targets

The actual setup is straightforward, but each decision point matters. 1. Navigate to campaign settings. Select the eligible campaign you want to edit. Click Settings in the page menu for this campaign. Open Bidding exploration, then select the checkbox next to "Allow Smart Bidding to explore new traffic."

2. Set your tolerance. Choose a percentage between 5–30%. Start at 10% for your first test. The feature calculates your new effective ROAS floor automatically. 3. Portfolio-level option. You can also set up Smart Bidding Exploration at the portfolio level, under portfolio bid strategy settings. If you enable Exploration, it applies to the entire portfolio - all campaigns within a portfolio using a single portfolio bid strategy will be enabled. For initial testing, keep it at the individual campaign level until you understand its behavior. 4. Don't simultaneously change your tROAS target. It's not suggested to increase your ROAS target to offset Exploration, because this will limit the ability to drive increased traffic diversity. Leave your existing target alone and let the tolerance slider do its job. 5. Save and monitor. Use the Bid Strategy report to monitor your target ROAS performance. Average target ROAS will factor in the lowered ROAS tolerance for Smart Bidding Exploration along with additional elements like ad group level targets or specific bid adjustments.

Designing a Reliable Test: The Experiment Framework

Enabling the feature directly on a campaign is one approach. A better one is using Google Ads Experiments to isolate the variable.

Setting Up the Experiment

Use campaign experiments. Split your traffic 50/50 between your base campaign and an experimental version with exploration enabled. This data-driven approach lets you monitor performance and see the actual impact before fully committing.

Follow Google's own best practices for experiment design:

• Ramp-up period.

Prior to enabling Smart Bidding Exploration in the trial arm, wait 2–3 conversion cycles or 7 days (whichever comes first) to ensure that base and trial arms have comparable spend and performance. Exclude that time period from performance evaluation.

• Minimum duration.

Enable Exploration in the trial phase and test for 6+ weeks, noting that it might take 1–2 additional weeks for Exploration to fully ramp.

• No major changes during the test.

After you opt in, don't adjust the Target ROAS. Change one variable at a time. If you're simultaneously changing creatives or audiences, you won't be able to attribute results to exploration. - Test in established campaigns only. It isn't recommended to test Smart Bidding Exploration in a new campaign. For optimal performance, test in a well-established campaign.

What Metrics to Watch

Standard ROAS isn't the primary success metric here - traffic diversity is. Google introduced a Traffic Diversity Metric that counts unique search categories that produced impressions, clicks, and conversions. It helps quantify how exploration expanded the range of searches driving results.

Measure success by looking at the increase in "Unique search categories with clicks," not just top-line performance metrics. Track these alongside conversion volume, conversion value, and actual ROAS. One agency documented real results: when they enabled exploration before peak season, the number of unique search categories with clicks jumped from around 840 to over 1,400. It did exactly what it was supposed to do - it went out and found new pockets of traffic. When they turned it off, categories dropped back to baseline - confirming the causal relationship. At the account level, campaigns using Smart Bidding Exploration see, on average, an 18% increase in unique search query categories with conversions and a 19% increase in conversions, according to Google internal data from March–April 2025.

Tactical Playbook: When and How to Use Exploration Strategically

Practitioners with months of testing experience have converged on a clear pattern: this is not a set-it-and-forget-it feature. It should be used strategically throughout the year.

Three High-Impact Use Cases

Peak season ramp-up. This is the number one use case. Right before or as your high season begins, turn on exploration at 10% or 20%. Even 10% can have a big effect. You're giving the algorithm license to find new buyer segments precisely when purchase intent is highest across the market. During major promotions. Big sales are another perfect time to find new customers who might be searching differently due to the promotion. Promotion-driven search behavior creates new query patterns the algorithm hasn't seen before. Algorithm re-education. If you're constantly adding new product categories or have been running at a very high ROAS target for a long time, the algorithm can get stuck in a rut. Enabling exploration for a focused sprint can force the system to re-learn its surroundings.

Sprint Duration and Cadence

The sweet spot for exploration sprints sits between 2 and 6 weeks. Think sprint, not marathon. Run it for 2–3 weeks minimum to give it time to work, but don't leave it on for more than 6 weeks. Align these bursts with a peak season, a sale, or a product launch.

After each sprint, analyze what the algorithm found. Mine your search terms report for the new query categories that converted. Promote the winners into their own ad groups with dedicated creatives. This compounds the value of each exploration cycle into permanent campaign improvements.

Common Mistakes That Undermine Exploration Results

Even well-intentioned tests fail when practitioners miss these nuances. Raising the tROAS to "compensate." Some advertisers increase their ROAS target before enabling exploration, thinking they'll net out at their original floor. Google explicitly advises against this because it will limit the ability to drive increased traffic diversity. You're defeating the purpose. Setting CPC limits. Setting maximum CPCs can restrict the AI's ability to bid competitively on high-value auctions, limiting conversions and overall performance. If you've historically used bid caps as guardrails, remove them before testing exploration - or the algorithm can't actually explore. Testing in budget-constrained campaigns. Max Conversion Value is intended for customers who have a finite daily budget. Smart Bidding Exploration is intended for advertisers with uncapped budgets who are comfortable to accept a lower ROAS in order to scale performance. Budget caps and exploration are fundamentally incompatible. Fix the budget first. Ignoring the stakeholder conversation. Accepting that the reported ROAS metric may decline from its historical peak while overall revenue and profit grow is a counterintuitive outcome that finance teams need to be prepared for before the strategy is activated. Brief your CFO or client before the test starts. Show them the math: a 10% ROAS decline paired with a 19% conversion increase often means more total profit. Running exploration alongside AI Max testing. The current recommendation is to test these features separately, as both aim to expand an advertiser's scope, allowing for a clearer view of what is truly effective. Stack too many AI-driven changes at once and you lose the ability to isolate what's working.

Reading the Results: How to Decide What Happens Next

After 6–8 weeks of testing, you'll have enough data to make a clear call. Structure your evaluation around three questions: Did traffic diversity increase meaningfully? Compare the Traffic Diversity metrics in the bid strategy report between your control and trial arms. To compare Traffic Diversity metrics, compare the respective Bid Strategy reports for the control and trial arm. A healthy exploration test should show a visible jump in unique search categories with conversions. Did total conversion value grow? Your actual ROAS will likely sit below your original target - that's by design. The question is whether the total conversion value in the exploration arm exceeds the control. More revenue at a slightly lower return rate is a net win. Are the new query categories worth keeping? Review the search terms report for the new categories. Some will be genuinely valuable expansion territory. Others will be fringe queries with poor downstream quality. Promote the strong categories, add negatives for the weak ones, and carry those learnings into your next sprint. If results are positive, roll exploration out to other campaigns. Prioritize your highest-budget campaigns first, as they'll see the most absolute benefit from expansion.

If results are inconclusive or negative, check whether your prerequisites were actually met. Budget constraints, thin conversion data, and narrow match types are the most common reasons for flat results. --- Smart Bidding Exploration isn't a universal growth lever, and anyone who presents it that way is oversimplifying. It's a precision tool for a specific stage of campaign maturity - the stage where your fundamentals are sound, your tracking is clean, and your campaigns have genuinely run out of room to grow within their existing query universe.

This feature is mainly for campaigns that are already working, not those that need fixing. Get your fundamentals right first, then use Smart Bidding Exploration to push beyond your current limits. Used in focused sprints at the right moments, flexible ROAS targets solve the real problem that has frustrated high-performing advertisers for years: how to scale profitable campaigns that have already been optimized to their apparent ceiling. The algorithm was always capable of finding more buyers - it just needed your permission to look.