About 30 conversions in 30 days is the practical floor a Google Ads Smart Bidding strategy needs to model bids reliably in 2026, and a strategy stuck below it shows the dreaded 'Learning limited' status. It is one of the most misread alerts in paid search: advertisers either panic and start editing — which makes it worse — or wait passively for a problem that will never resolve on its own. Both reactions cost weeks of underperformance.
This guide explains exactly what 'Learning limited' means, how it differs from the normal learning phase, and the step-by-step path back to a stable 'Eligible' status. To check your account against the most common Smart Bidding signal problems automatically, run our free 5-axis Google Ads audit.
Updated 2026-05-07 with current Smart Bidding learning-phase behavior observed across US, UK and European accounts.
- 'Learning limited' is not 'Learning' — the first is stuck, the second clears on its own. 2. Too few conversions is the top cause — roughly 30 in 30 days for tCPA, 50 for tROAS, per strategy. 3. Frequent edits restart learning — change one thing at a time, then wait a full window. 4. Unrealistic targets starve the algorithm of qualifying auctions — loosen by 15 to 20 percent. 5. Consolidate signal with portfolios and merged conversions before adding more budget.
What does 'Learning limited' mean?
Every Smart Bidding strategy carries a status, and two of them look alike but mean very different things.
Learning — The normal, temporary state right after you create or significantly change a strategy. The algorithm is calibrating against recent data, performance is noisy by design, and the status clears on its own within a few days. This requires patience, nothing more.
Learning limited — An alert that the strategy is stuck in learning because something is blocking it from gathering enough signal. It will not clear on its own. The three usual blockers are too few conversions, too-frequent edits that keep restarting the clock, and a target so strict the algorithm cannot find qualifying auctions.
The single biggest mistake is treating 'Learning limited' like plain 'Learning' and simply waiting. A genuinely stuck strategy can sit in 'Learning limited' for months while you wait for a resolution that never arrives. The status is a prompt to diagnose, not to be patient.
How much conversion volume does Smart Bidding need?
Conversion volume is the fuel for Smart Bidding, and most 'Learning limited' cases trace straight back to a thin tank.
The practical floors — Google's guidance points to roughly 30 conversions in the trailing 30 days for Target CPA, and about 50 conversions for Target ROAS. These are floors, not switches: a strategy sitting right at 30 will learn slowly and noisily, while one at 200 stabilizes fast.
Per strategy, not per campaign — This is the detail most advertisers miss. The conversion count is measured across the whole bid strategy. Three campaigns with 12 conversions each are not three starved strategies; combined into one portfolio they form a single healthy one with 36.
Why thin volume stalls learning — With too few data points, the algorithm cannot tell signal from noise, so it cannot set confident bids. The result is wide performance swings and a status that refuses to leave 'Learning limited'. The fix is almost always more signal per strategy, which is a structural problem, not a budget one. Our guide to reducing CPA covers how cleaner conversion tracking adds usable volume.
How edits and unrealistic targets trigger it
Even an account with enough conversions can get stuck if you sabotage the learning window or set an impossible target.
Frequent edits restart learning — Significant changes reset the clock: switching the bid strategy type, moving the Target CPA or Target ROAS by a large amount, dramatic budget changes, or editing the conversion actions. If you make a big change every few days, the strategy never completes a single clean window. Minor tweaks — small budget nudges, a handful of keywords — generally do not reset it.
Unrealistic targets starve the auction — A Target CPA set far below what your account has ever achieved tells the algorithm to bid only in auctions it can win at that price, and there may be almost none. The strategy then under-delivers and stays 'Learning limited'. The same happens with a Target ROAS set far above your historical return.
The compounding trap — Advertisers who see 'Learning limited' often react by tightening the target or editing again, which makes both problems worse. The discipline is counterintuitive: when stuck, stop editing and, if anything, loosen the target. Use our CPA calculator to set a target your margins and history can actually support.
How to consolidate conversions and structure to exit it
If thin volume is the cause, consolidation is the most reliable cure, because it concentrates the signal you already have.
Merge fragmented campaigns — Several near-identical campaigns each carrying a few conversions starve every one of them. Combining them into fewer, larger campaigns lets a single strategy see the pooled total and cross the conversion floor.
Use portfolio bid strategies — A portfolio strategy pools conversions across multiple campaigns into one learning model. For accounts with many small campaigns, this is the single most effective way to escape 'Learning limited' without adding spend.
Clean up conversion actions — Counting too many low-value or duplicate actions dilutes the signal. Pick one primary conversion action that reflects real business value, and demote micro-conversions to 'secondary'. A consistent, meaningful conversion event gives the algorithm a clearer target to learn against.
Lift conversion volume at the source — Better tracking, Enhanced Conversions, and offline imports each add real conversions the algorithm can use, which is often faster than restructuring alone.
Should you switch targets or bidding type?
When volume cannot be raised quickly, changing the strategy itself is the lever — but it has to be the right change.
Loosen the target first — If your Target CPA or Target ROAS is the bottleneck, relax it by 15 to 20 percent before doing anything drastic. A looser target opens more auctions, restores delivery, and often clears 'Learning limited' on its own within a window.
Switch to Maximize Conversions for thin accounts — Maximize Conversions and Maximize Conversion Value carry no explicit target, so they have the most freedom to enter auctions and gather signal. For a low-volume account, switching to one of these to build a conversion baseline, then layering a target back on later, is a proven path out.
Match the strategy to the goal — The choice between maximize-style and target-style bidding is not just about learning; it shapes your whole account. Our Maximize vs Target CPA guide and our Target ROAS vs Target CPA comparison walk through when each fits. If you are weighing automation against staying manual, our Manual CPC vs Smart Bidding guide helps you decide.
The 12-point learning-limited diagnostic
Work this table top to bottom — it is ordered by how often each cause is the real culprit and how fast it is to confirm.
The instinct when you see 'Learning limited' is to change something — tighten the target, move the budget, swap the strategy. Each significant edit restarts the learning clock, so a habit of weekly tweaks guarantees the strategy never stabilizes. When stuck, freeze the settings for a full window and change one thing at a time. Discipline beats activity here.
How long to wait before acting
Patience and action are both right — at different moments. The trick is knowing which one a given situation calls for.
For a genuinely new strategy — Give it the full learning window: about 7 days and roughly 30 conversions with stable settings. During this window, plain 'Learning' is expected and you should not touch the configuration. Judging performance before the window closes leads to premature, harmful edits.
For persistent 'Learning limited' — If the status still reads 'Learning limited' after two to three weeks of stable settings, time is not your problem. The cause is structural — volume, fragmentation, or an unrealistic target — and waiting another month changes nothing. This is when you consolidate, loosen, or switch.
The decision rule — Wait through exactly one clean learning window for any new or freshly edited strategy. Beyond that, treat persistent 'Learning limited' as a structural diagnosis, not a waiting game.
A clean, well-consolidated account rarely gets stuck in the first place. Run the SteerAds free 5-axis audit to find thin conversion signal, fragmented structure and over-strict targets automatically before they trap a strategy in 'Learning limited'.
Sources
Official sources consulted for this guide:
-
support.google.com — about Smart Bidding
-
support.google.com — Target CPA bidding
-
support.google.com — bid strategy status and learning
-
blog.google — Ads & Commerce updates
FAQ
What does 'Learning limited' mean in Google Ads?
'Learning limited' is a bid-strategy status telling you Smart Bidding cannot collect enough conversion signal to optimize reliably. It differs from the normal 'Learning' status, which is temporary and resolves on its own after a setup change. 'Learning limited' is a warning that the strategy is stuck: usually too few conversions in the rolling window, frequent edits that keep restarting learning, or a target so strict the algorithm cannot find enough qualifying auctions. Until you fix the root cause, performance stays noisy and the strategy never reaches stable 'Eligible'.
How many conversions does Smart Bidding need to exit learning?
Google's guidance points to roughly 30 conversions in the trailing 30 days for Target CPA and about 50 for Target ROAS, measured per bid strategy, not per campaign. These are practical floors, not hard switches: more volume means faster, steadier learning. A strategy hovering near the floor will often show 'Learning limited' because the signal is too thin to model bids confidently. The fastest lever is usually consolidating conversions and campaigns so one strategy sees more data, rather than waiting passively for volume to arrive.
Will 'Learning limited' fix itself over time?
Sometimes, if conversion volume naturally rises and you stop editing the strategy. If a seasonal spike or a new high-volume keyword pushes you past the conversion floor, the status can clear within one to two weeks on its own. But if the cause is structural — a tiny account, an over-strict target, or a habit of frequent edits — waiting changes nothing. Treat 'Learning limited' as a prompt to diagnose volume and stability first, then act, rather than assuming time alone resolves it.
Does editing a campaign reset the Smart Bidding learning phase?
Significant edits do. Changing the bid strategy type, the Target CPA or Target ROAS value by a large amount, the budget dramatically, or the conversion actions all restart learning. Minor tweaks — small budget nudges, adding a few keywords or negatives — generally do not. The practical rule is to make one meaningful change at a time and then wait the full learning window before the next. Stacking several edits in a week is the most common self-inflicted cause of a strategy stuck in 'Learning limited'.
Should I switch from Target CPA to Maximize Conversions when learning limited?
It is often the right move for low-volume accounts. Maximize Conversions (or Maximize Conversion Value) has no explicit target to satisfy, so it has more freedom to enter auctions and gather signal, which helps a thin account exit 'Learning limited'. Start there to build volume, then layer a target back on once you have a stable conversion baseline. If your Target CPA was simply too strict, loosening it by 15 to 20 percent can also clear the status without abandoning target-based bidding entirely.
How long should I wait before acting on 'Learning limited'?
Give a genuinely new strategy the full learning window — about 7 days and roughly 30 conversions — before judging it. If the status reads 'Learning limited' rather than plain 'Learning', and the strategy has had two to three weeks with stable settings yet still cannot exit, the cause is structural and waiting longer will not help. The decision rule: wait through one learning window for new strategies, but for persistent 'Learning limited' after stable settings, act on volume and structure instead of waiting.
What is the difference between 'Learning' and 'Learning limited'?
'Learning' is the normal, temporary state after you create or significantly change a strategy: the algorithm is calibrating and the status clears on its own once it has enough recent data. 'Learning limited' is an alert that the strategy is stuck in learning because something is blocking it — most often too few conversions, frequent restarts from edits, or an unrealistic target. Plain 'Learning' needs patience; 'Learning limited' needs diagnosis. Confusing the two leads advertisers to wait on a problem that will never resolve by itself.