The TikTok ad learning phase is a calibration period where TikTok’s algorithm explores audiences, tests delivery, and collects conversion signals before it can optimize efficiently. It lasts up to 7 days, and volatility starts to decline around 25 conversions. The full passing threshold is 50 conversions per ad group, which is the most significant indicator that the algorithm has enough data to deliver consistent, optimized results.
Most TikTok ad campaigns do not fail because of bad creative. They fail because advertisers panic during the learning phase and make changes at exactly the wrong time.
If your CPMs spiked after launch, delivery feels inconsistent, or results look worse than expected in week one, that is the learning phase working as designed.
The problem is that most advertisers do not know what it is, what triggers it, or what they should and should not do while it is active.
Get Up to $6,000 in Free TikTok Ad Credits
TikTok offers ad credit incentives for new advertisers, helping you test campaigns with a lower upfront cost.
This article covers all of it. You will learn what the TikTok ad learning phase is, how long it lasts, why 25 conversions matter, what resets it, and the exact steps to pass it without wasting budget.
Table of Contents
Key Takeaways
- The TikTok ad learning phase lasts up to 7 days or 25 conversions per ad group, whichever comes first
- Learning happens at the ad group level, not the campaign level
- Performance starts to stabilize around 25 conversions, according to TikTok Ads Manager
- TikTok’s official budget guidance recommends a daily budget of at least 20x your target CPA
- Editing the budget, audience, bid, or creative during learning can reset the entire process
- If your ad group fails to learn, fix the creative and targeting first. Do not just increase spend.
What Is the TikTok Ad Learning Phase?
The TikTok ad learning phase is the period after you launch or significantly edit an ad group when TikTok’s algorithm tests your audience, creatives, and bidding to find who is most likely to convert.
During this window, three things are happening simultaneously. The algorithm is exploring a broad audience pool to identify conversion patterns. It is collecting pixel or in-app event signals to model your ideal customer. And it is adjusting delivery timing, bid pacing, and placement mix to move toward efficient performance.
According to the TikTok Ads Manager Help Center, this process applies to every ad group using a conversion-based bidding strategy, including Cost Cap, Bid Cap, and Lowest Cost. Your learning phase status is visible in the delivery column of TikTok Ads Manager.
Until the algorithm has enough data, it cannot predict conversions with confidence. That is why early performance looks unstable. It is not a broken campaign. It is an incomplete model.
How Long Does the TikTok Learning Phase Last?
The TikTok learning phase lasts up to 7 days, but ends earlier if your ad group hits 25 conversions first. According to TikTok’s official documentation, performance volatility typically starts declining after 25 results or 7 days, whichever comes first. The 50-conversion mark is the full completion threshold for reliable delivery optimization.
Learning applies at the ad group level, not the campaign level. Each ad group in a campaign runs its own independent learning process. One ad group can exit learning while another is still in it, which is why campaign-level data can be misleading in the first week.
Note for TikTok Search Ads: According to TikTok’s Search Ads reporting guidance, search campaigns have a shorter calibration period of approximately 5 days, with a target of 20 conversions in the first week rather than 50. Exclude those first 5 days from your performance reporting.
Why Does TikTok Need 25 Conversions to Pass the Learning Phase?
Each conversion tells the algorithm something specific: which content the user watched, what audience segment they belong to, what time they converted, and what path they took. Below 25 conversions, the model carries too much variance to predict future behavior accurately. At 25 or more, patterns stabilize, and the algorithm shifts from broad testing to targeted delivery.
According to the TikTok Learning Phase, hitting 25 conversions is the most significant indicator of passing the learning phase. If your ad group struggles to compete in your target auction, TikTok recommends broadening targeting, increasing your bid, or improving creative to help reach that threshold.
The 25 conversions per week figure also applies to ongoing campaign health after learning ends. Dropping below that consistently can cause delivery to become erratic, even in a previously stable campaign.
Why Your CPM Is Higher During the TikTok Learning Phase
Your CPM is higher during learning because the algorithm is buying broadly. It has not yet identified who converts, so it tests across a wide range of audience signals and placements. That broad testing is inherently less efficient, and the cost shows up in your CPM.
According to TikTok’s learning phase documentation, performance fluctuations during this period are expected. CPMs normalize once the ad group exits learning, and delivery narrows to your best-converting segments.
We have seen advertisers adjust bids or swap audiences the moment CPMs spike, which resets the learning clock and extends the problem. Instead of watching CPM, watch these three signals:
- Conversion volume: Are you on track for 25 in 7 days?
- Impression delivery: Is spend consistent day to day?
- Creative engagement: Are your ads generating clicks, not just views?
If all three are positive, elevated CPM is not a problem. It is the cost of the algorithm doing its job.
What Actions Re-Trigger the TikTok Learning Phase?
Any significant edit to an active ad group can push it back into learning. Budget increases can trigger re-learning depending on scale.
Increasing the daily budget from $100 to $110 lets the system find additional users based on existing settings. Increasing from $100 to $300 forces it to find a large new audience group, which re-triggers learning entirely.
| Action | Re-Triggers Learning? |
| Large budget increase (e.g., $100 to $300+) | Yes |
| Small budget increase (within 10%) | Usually No |
| Decrease the daily budget | Yes |
| Change the bid strategy or the bid amount significantly | Yes |
| Change audience targeting | Yes |
| Change optimization event | Yes |
| Pause and resume (1-2 days) | Usually No |
| Pause and resume (7+ days) | Yes |
| Edit ad copy or caption only | Usually No |
| Change landing page URL | Yes |
| Duplicate an ad group | Yes (starts fresh) |
If you need to make meaningful changes to an underperforming ad group, duplicate it and apply the changes in the new version. Your original ad group keeps its optimization history.
How to Pass the TikTok Ad Learning Phase (Step-by-Step)
- Set the right budget using the formula given below
- Choose the right optimization event: use one that fires at least 25 times per week. If purchase volume is too low, start with Add to Cart or Initiate Checkout
- Launch 3 to 5 creatives per ad group to give the algorithm delivery options
- Keep your audience broad: tight interest targeting restricts exploration. Let TikTok find your best audience first
- Do not edit the ad group until you have at least 25 conversions
- Set your bid with headroom: if your target CPA is $30, test a Cost Cap of around $39 to $42. A bid set exactly at the target is often too tight to spend freely during learning. Note: this is a practitioner benchmark, not an official TikTok figure
- Evaluate performance only after 7 days or 25 conversions, whichever comes first
How to Calculate the Right Budget for the Learning Phase
According to TikTok’s Bidding and Budget guidance, the recommended formula is:
Daily budget = Target CPA x 10 conversions
TikTok also sets a floor of 20x your target CPA as a minimum for efficient learning. At a $20 target CPA: 20 x $20 = $200/day minimum.
If that budget is not viable, optimize for a higher-funnel event like Add to Cart or View Content, where 50 events are achievable at a lower cost per action.
How Many Creatives Do You Need During the Learning Phase?
TikTok recommends 2 to 5 creatives per ad group. According to TikTok’s Smart Lead Generation best practices, add new creatives without deleting existing ones and avoid adding more than 10 at once to prevent budget strain.
Each creative tests a different audience response. One hook might convert 18 to 24-year-olds. Another may perform better with a 30+ segment. One creative gives the algorithm no comparison point. Three to five gives it real data to work with.
What If Your TikTok Ads Fail the Learning Phase?
A failed learning phase means your ad group exited the 7-day window without reaching 50 conversions. The delivery model is incomplete, and performance will stay volatile.
Diagnose before you spend more. According to the TikTok Learning Phase, if an ad group cannot hit 20 conversions in the first 10 days, passing the learning phase is unlikely without structural changes.
Recovery steps:
- Pause the failed ad group after 5 to 7 days of no meaningful conversion activity
- Fix the creative first: new hook, new format, or a direct-response angle you have not tested
- Switch to a higher-funnel event if purchase volume is too low
- Broaden your audience and remove restrictive interest layers
- Duplicate and relaunch with the fixes applied
One important distinction: a failed learning phase is not the same as being stuck in review. “In Review” is a content moderation check, not a learning status. TikTok’s review process typically resolves within 24 hours. If it does not, contact support through the TikTok Business Help Center. The two issues require completely different responses.
What Happens After You Pass the TikTok Learning Phase?
Once your ad group exits learning, three things stabilize: CPA settles into a consistent range, CPMs drop as targeting narrows, and daily spend becomes predictable.
At this point, you can evaluate real performance and consider scaling. According to TikTok’s budget guidance, increase budgets by no more than 30% per adjustment after learning, no more than once every two days. During learning, you can increase by up to 40% per adjustment without re-triggering it.
For scaling without resetting: duplicate your winning ad group and test higher budgets in the new version. Your original keeps its optimization history.
Stable delivery in Ads Manager looks like this: status shows “Active,” CPA variance is under 20% day-over-day for three or more consecutive days, and spend is consistent rather than erratic.
Explore these helpful articles next:
👉 TikTok Ads Credit, Coupon, and Promo Code: How to Get Up to $6,000 in Free Ad Spend
👉 Retargeting with TikTok Ads: Remarketing Strategies That Work
👉 TikTok Ad Types Explained: How to Choose the Right Format for Your Brand
👉 TikTok Ads Lookalike Audiences: Complete Setup Guide
Frequently Asked Questions About the TikTok Ad Learning Phase
Does pausing my TikTok ad restart the learning phase?
No, Short pauses of 1 to 2 days typically do not reset learning if meaningful progress was already made. According to TikTok’s learning phase documentation, pausing is one of the actions to avoid during learning, as it can prevent your ad group from exiting the learning phase status. Pauses of 7 days or longer are more likely to trigger a full reset.
Can a TikTok ad group pass the learning phase without 50 conversions?
Yes, but delivery optimization will be based on incomplete data, and performance will remain less predictable. According to the TikTok Learning Phase FAQ, if an ad group cannot reach 20 conversions in the first 10 days, fully passing learning is unlikely. Switching to a higher-funnel event and relaunching typically produces better results than continuing on a limited model.
Where do I find my learning phase status in TikTok Ads Manager?
It is in the ad group delivery column in TikTok Ads Manager. Labels include “Learning”, “Active”, and “Learning Limited.” Full documentation is available at the TikTok Business Help Center. The status only appears on ad groups using conversion-based bidding strategies.
Why are my TikTok ads performing worse after I made changes?
Edits to the budget, bid, audience, or landing page URL can reset TikTok’s delivery model and force the algorithm to restart audience exploration. You lose the optimization data already collected, and performance reverts to early-phase instability. Follow TikTok’s official limits: no more than 40% budget increase during learning, no more than 30% after learning, and no more frequently than every two days.
How is the TikTok learning phase different from Meta’s learning phase?
Both platforms use an event threshold within a 7-day window before delivery stabilizes. According to Bind Media’s comparison, Meta’s learning phase is generally more stable with clearer in-platform guidance, while TikTok’s algorithm re-triggers more easily and requires more conservative budget editing behavior. Both platforms use a “Learning Limited” label for under-performing ad sets that cannot gather enough data.
What is the best bidding strategy during the TikTok learning phase?
Lowest Cost gives the algorithm the most flexibility to find converting users without a hard bid ceiling restricting delivery. According to TikTok’s bidding guidance, pair it with a daily budget of at least 10x to 20x your target CPA. Once you have passed learning and have stable CPA data, switch to Cost Cap to maintain scale with a cost ceiling.
Conclusion
The TikTok ad learning phase is not a flaw in the system. It is the system working exactly as designed, and most of the frustration advertisers experience comes from editing campaigns too early or launching without enough budget to reach the 25-conversion threshold.
You now have the full picture: what the learning phase is, why 25 conversions matter at the ad group level, why CPMs spike and normalize, which edits reset your progress, the budget formula to give the algorithm what it needs, and how to recover when learning fails. You also know what stable performance actually looks like after you pass it.
Treat the first 7 days as a data collection period, not a performance evaluation. Make your decisions after learning, not during it.