Skip to content
whitead

Fix: Campaign stuck in Smart Bidding Learning status for more than 14 days

finding google ads updated 2026.05.27 8 min read
// contents · 8 sections
  1. Why this matters
  2. If you just launched the audit
  3. How to verify the issue
  4. How to fix it
  5. How to confirm the fix worked
  6. Diagnostic checklist — run within 14-21 days of intervention
  7. Related rules + concepts
  8. Sources

The Smart Bidding badge has been showing "Learning" for two weeks or more. Google's guidance is that the badge "typically" clears in about 7 days; when a campaign overshoots that window, the model is telling you something material changed recently — a budget jolt, a target shift, a tracking edit, a paused conversion action, or a geo expansion — and it is recalibrating against the new reality [1][2]. Until it converges, CPA and ROAS will move in noisy ways that are unsafe to act on.

Why this matters

A Smart Bidding strategy treats every recent change as new information. When you raise a daily budget by 30%, tighten a tCPA by 20%, add or pause a conversion action, expand geo, or swap a primary conversion goal, the model re-enters Learning while it re-fits the auction-time signals (device, location, time, audience, query intent) against the new constraints [1]. Google describes the badge as appearing for "typically up to 7 days"; Optmyzr recommends two to three weeks before judging Smart Bidding after a budget change, and Search Engine Land identifies learning periods beyond three weeks as a red flag for intervention [2][3][4].

The blast radius is decision-making, not delivery. The strategy still bids — Smart Bidding does not stop while in Learning — but the bids are exploratory rather than converged. Comparing CPA against a pre-change baseline in this window will systematically mislead: you read noise as signal, panic-tweak the target, and reset the clock again. This is the failure pattern behind "perpetual learning" — a campaign that never demonstrates its real performance because each weekly review triggers another corrective change [4].

"Learning" and "Learning Limited" are two distinct conditions with different fixes [1]. Learning means the model is re-calibrating after a recent change and conversion volume is sufficient — wait it out. Learning Limited means the campaign sits below the conversion-volume floor the strategy needs (≈30 conv/30d for tCPA and Maximize Conversions, ≈50 conv/30d for tROAS and Maximize Conversion Value) [1]. Waiting does not fix Learning Limited — feed the model more signal (broader audience, more budget, offline conversion imports, value-based event upgrades) or step the strategy back to one that requires less volume.

If you just launched the audit

This rule only fires after the auditor has accumulated enough temporal data to compare campaign-state across time. If your Whitead-tracked account is younger than 14 days from its first conversion, the rule status will be "insufficient data" with the message "Temporal analysis available from the next audit." New campaigns are expected to be in Learning — re-run the audit after two full weeks of active conversion tracking.

How to verify the issue

  1. Open the affected campaign and read the bid strategy column. Confirm the status badge text — Learning, Learning Limited, Eligible, or Misconfigured. If you see Learning Limited, this is the wrong finding — pivot to the conversion-volume root cause instead [1].
  2. Open Tools → Change history. Filter to the affected campaign and the last 30 days. Look for any of the high-impact triggers: budget changes greater than ~20%, target CPA/ROAS changes greater than ~15%, primary conversion-action edits, conversion-action pauses, audience-segment changes, geo expansion or contraction, bid-strategy switches.
  3. Check the bid strategy report (Tools → Shared library → Bid strategies for portfolio strategies, or the campaign settings for standard strategies). Note the "status reason" — Google surfaces a short explanation under the badge ("Recent budget change", "Recent conversion change", etc.) when one is available.
  4. Cross-check conversion volume in Goals → Conversions → Summary. Count campaign-attributed conversions over the trailing 30 days. If the count is below the strategy's volume floor (~30 for tCPA, ~50 for tROAS) the badge will read Learning Limited, not Learning — and the fix path differs [1].
  5. Look at the daily-spend curve over the last 14 days. Erratic daily delivery (one day at 30% of budget, the next at 110%) is corroborating evidence that the model has not converged and is exploring [4].

How to fix it

Total time: 5-15 minutes to diagnose, then 7-14 days of patience. The fix is almost never "change another setting" — most stuck-learning cases resolve once you stop tweaking.

  1. Identify the trigger event. Use Change history (step 2 above) to pin down what changed and when. Most cases trace to one specific event in the last 14-30 days.
  2. Decide: revert, or commit and wait. If the change was speculative ("let's see if a 30% lower tCPA holds CPA"), revert now. If it was intentional and necessary, do not revert — commit and wait the full window.
  3. Freeze the campaign for at least 7 additional days. No budget changes greater than 20%, no target changes greater than 15%, no creative swaps, no structural edits. Treat the campaign as read-only [3].
  4. If the badge is still Learning past 21 days, check whether multiple smaller changes are stacking (ad-group budget edits, negative-list updates, audience modifiers) and stop them [4].
  5. Consider a strategy reset, last resort. If the badge is still Learning after 21 days with no recent changes and healthy conversion volume, step the strategy back to Maximize Conversions (no target) or Maximize Conversion Value (no tROAS target). Run 2-3 weeks, then re-apply tCPA or tROAS with the target anchored to observed CPA or ROAS over that window. Do not re-pick the original target — re-anchor to current reality.
  6. For Learning Limited (the other status), the fix is volume, not waiting: add Offline Conversion Imports via Data Manager to backfill phone-call or pipeline conversions [5], lift Customer Match seed sizes, broaden audience signals, or pool sub-scale campaigns into a Portfolio Bid Strategy.

How to confirm the fix worked

Diagnostic checklist — run within 14-21 days of intervention

  • The Smart Bidding status badge has moved from Learning to Eligible (or Active).
  • Daily spend has settled — no day-over-day swings greater than ~30% in the last 7 days.
  • Trailing-30-day CPA or ROAS is within ±15% of the rolling baseline observed before the original trigger event.
  • Change history shows no operator-initiated changes greater than the 20% budget / 15% target thresholds in the last 14 days.
  • If a strategy reset was used, the new target (re-anchored to observed performance during the Max Conversions / Max Conv Value window) is held flat for 14 more days before any further tightening.
  • If the original status was Learning Limited, trailing-30-day conversion count is now above the strategy's volume floor and the badge has stepped out of Limited.

If the badge has cleared and the metrics are within tolerance, the finding closes.

Sequencing note. Stuck Learning sits at the intersection of three other rules Whitead handles separately: tcpa_premature_switch (tCPA enabled before the campaign cleared the volume floor — that is a Learning Limited root cause, not Learning), troas_without_value_tracking (value-based strategy with no value signal — the bidder cannot converge), and bidding_strategy_goal_alignment (strategy does not match the campaign goal — re-learning will not fix a category-level mismatch). When two or more co-occur, fix the upstream rule first. Diagnostic order: (1) strategy-goal alignment, (2) value signal for value-based strategies, (3) conversion-volume floor for Learning Limited, (4) only then is "wait out the Learning window" the right answer for this rule. Re-judging performance inside the Learning window is the most common operator mistake — and the most common cause of perpetual-learning loops where weekly panic-tweaks reset the clock forever [4].

Sources

  1. Google Ads Help — About Smart Bidding. https://support.google.com/google-ads/answer/7065882 (accessed 2026-05-27)
  2. Google Ads & Commerce Blog — Smart Bidding Exploration: AI-powered bidding. https://blog.google/products/ads-commerce/smart-bidding-exploration-ai/ (accessed 2026-05-27)
  3. Optmyzr — The Impact of PPC Bidding Strategies on Google Ads Performance. https://www.optmyzr.com/blog/impact-of-ppc-bidding-strategies/ (accessed 2026-05-27)
  4. Search Engine Land — When Google's AI bidding breaks – and how to take control. https://searchengineland.com/google-ai-bidding-breaks-take-control-466251 (accessed 2026-05-27)
  5. Google Ads Help — About offline conversion imports. https://support.google.com/google-ads/answer/2998031 (accessed 2026-05-27)
// was this useful?
// anonymous · no personal data stored
← Back to Wiki