What Actually Resets the Smart Bidding Learning Phase (and What Doesn't)
Google does not publish a complete official list of what restarts the Smart Bidding learning phase, which is exactly why teams keep triggering it by accident. Here is what reliably resets it, what usually does not, and how to make a change without losing a week of performance.
What Actually Resets the Smart Bidding Learning Phase (and What Doesn't)
The Smart Bidding learning phase is the period after a significant change to a Google Ads campaign's bid strategy, budget, or targeting during which the algorithm's performance forecasts are less reliable while it recalibrates on new data - campaigns in this state typically show more volatile costs and results for roughly one to two weeks. The frustrating part for anyone managing an account: Google does not publish one complete, definitive list of every action that triggers this reset, which is exactly why teams keep re-entering it by accident and losing a week of stable performance over a change that seemed minor.
Key takeaways
- Bid strategy changes (switching between Target CPA, Target ROAS, Maximize Conversions, or Maximize Conversion Value) reliably restart the learning phase - this is the most consistently reported trigger.
- Significant budget changes (commonly cited threshold: roughly 20% or more in either direction) can restart learning, though Google has not published an exact, guaranteed percentage.
- Adding or removing a conversion action tied to the bid strategy's optimization target is a documented trigger; adding an unrelated conversion action is less consistently reported as one.
- Pausing and resuming a campaign, minor ad copy edits, and small keyword additions are widely reported by practitioners as NOT reliably triggering a reset, though account-specific results vary.
- Because Google's own documentation doesn't fully enumerate every trigger, the safest practice for any uncertain change is batching it with other planned changes and expecting a short volatility window regardless.
Teams that treat every account edit as risk-free because "it's just a small change" learn the actual trigger list the expensive way - a temporary performance dip right after a change that looked unrelated to bidding at all.
Learning phase: the period following a significant Smart Bidding-related change during which Google's algorithm has less reliable performance data to bid against, typically producing more volatile cost-per-result until it stabilizes - generally one to two weeks, though duration varies by account traffic volume.
Smart Bidding: Google's automated bidding system (Target CPA, Target ROAS, Maximize Conversions, Maximize Conversion Value) that sets bids algorithmically based on the campaign's stated optimization goal and historical conversion data.
Why Google doesn't publish one complete trigger list
The operational pain this creates for anyone trying to plan changes around the learning phase: without an authoritative, exhaustive list, account managers are left piecing together the actual trigger set from community reports, their own account history, and cautious testing - which means two teams can have genuinely different experiences with the same nominal change, since account-specific factors (traffic volume, existing conversion data depth) appear to influence how sensitive a given account is to a particular edit.
The most consistently and widely reported triggers, corroborated across practitioner reports and Google's own general Smart Bidding guidance, are: switching bid strategy type, a large budget change, and adding or removing a conversion action that's part of the bid strategy's optimization target. A budget increase past a certain point stops producing proportional results for a related but distinct reason - worth distinguishing from a learning-phase reset, since the two effects can overlap around the same change. Google's own UI itself signals a "Learning" status label on affected campaigns after a change - useful as a real-time confirmation, but only after the fact, not as an advance warning before making the change.
Stop stitching platform exports together
Every channel in one brief — plus the memory of what each one actually drove.
14 days free · no credit card
Why some changes don't reset learning even though they feel significant
The ICP problem this creates for cautious account managers: overcaution has a real cost too - treating every single edit as learning-phase-risky leads to freezing account optimization entirely, which is its own performance drag separate from the learning phase itself.
Based on widely corroborated practitioner reporting, changes that do NOT reliably trigger a learning phase reset include: pausing and later resuming a campaign (though a long pause can independently cause a slower ramp-up for unrelated reasons - stale signal, not a learning reset), minor ad copy edits or new ad variations within an existing ad group, small keyword additions to an already-established campaign, and routine negative keyword additions. The practical rule: changes that alter the bid strategy's core inputs (its optimization target, its budget constraint, its bidding type) are the higher-risk category; changes that add more creative or targeting variety within an unchanged bidding framework are lower-risk.
How to make a necessary change without losing a full cycle
The ICP problem this creates for teams that need to make a genuinely learning-phase-triggering change (a real bid strategy switch, a real budget shift) on a schedule they don't fully control: the change still has to happen, but the timing of when it happens is often within the team's control even when the change itself isn't avoidable.
The practical approach: batch every planned learning-phase-risky change into a single window rather than spreading them across separate weeks, since each separate trigger restarts a new volatility period - three small changes made a week apart can produce three separate weeks of instability where one combined change would produce one. Schedule the change for a lower-stakes period if the account has a predictable slow season, rather than immediately before a known high-stakes push (a major promotional period, a critical reporting deadline). And set the expectation with stakeholders before the change, not after performance dips - a known, explained volatility window reads very differently to a client or manager than an unexplained one discovered mid-dip.
Prooflytics tracks campaign-level spend and performance trend in the daily briefing, which makes a post-change volatility window visible against the campaign's own recent baseline - useful context for distinguishing an expected learning-phase dip from a genuine performance problem, though the briefing does not currently flag "learning phase active" as a distinct status the way Google Ads' own UI does.
Bottom line
- Bid strategy changes, significant budget shifts, and conversion-action changes tied to the optimization target are the most consistently reported learning-phase triggers - Google has not published one complete official list.
- Minor ad copy edits, small keyword additions, and pause/resume cycles are widely reported as lower-risk, though account-specific results vary.
- Batch planned learning-phase-risky changes into a single window rather than spreading them out - each separate trigger restarts its own volatility period.
- Set stakeholder expectations before a known-risky change, not after a performance dip appears unexplained.
- Book a walkthrough to see how Prooflytics tracks campaign-level spend and performance trend in the daily briefing.
Frequently asked questions
How long does the learning phase actually last?+
Google's general guidance cites roughly one to two weeks for most campaigns, though the actual duration depends heavily on the account's conversion volume - a high-traffic campaign with abundant daily conversion data typically stabilizes faster than a low-volume campaign, since the algorithm has more new data to recalibrate against per day.
Should I avoid making any changes while a campaign is already in the learning phase?+
Generally yes for bidding-related changes specifically - stacking a second bid-strategy-affecting change on top of an already-active learning phase tends to extend the instability rather than run two separate windows in parallel. Non-bidding changes (creative, minor targeting) are lower risk to make during an active learning phase.
Does switching from manual bidding to Smart Bidding trigger this the same way?+
Yes - moving from manual bidding into any Smart Bidding strategy is itself a bid-strategy change and triggers the same learning phase behavior as switching between two different Smart Bidding strategies.
Is there a way to see exactly which of my changes triggered a learning phase reset?+
Google Ads' own change history log, cross-referenced against the campaign's "Learning" status timestamps in the interface, is the most direct way to correlate a specific edit with a subsequent reset - there's no single report that draws this connection automatically, so it takes a manual cross-reference.
You can read independent reviews of Prooflytics on G2 and compare it to other marketing intelligence platforms in the category.
Stop stitching platform exports together
Every channel in one brief — plus the memory of what each one actually drove.
14 days free · no credit card
Continue reading
Google Ads Bid Strategy Testing in 2026 Requires CRM Data, Not Just Campaign Metrics
Google Ads has shifted bid strategy validation toward conversion value by time and first-party CRM data. Testing on campaign metrics alone now produces false positives. Here is the operational setup teams need before running any bid experiment in 2026.
Smart Bidding Model Decay: Why Your Google Ads Performance Drifts Without Warning
Smart Bidding and lookalike audiences are machine learning models trained on your historical data. When the data drifts -- due to seasonality, audience quality changes, or stale conversion signals -- model accuracy degrades silently. Performance falls, but the cause is invisible in standard dashboards. Here is how to detect and fix model decay.
Why Did My CPL Increase? 5 Causes GA4 Will Never Show You
CPL spikes have five systemic causes that no single dashboard surfaces automatically. Here is how to diagnose each one and fix it.
Google Ads Bid Strategy Testing Now Requires CRM Data: What Changed in 2026
Google Ads changed bid strategy validation requirements in 2026, shifting from surface-level metrics like ROAS and CPC toward conversion value by time window and first-party CRM data integration. Testing a new bid strategy without CRM data now risks false positives and budget waste -- the model validates against aggregated campaign metrics that can be disconnected from actual customer lifetime value.