How does the calculator work?
It turns a daily budget into a weekly count. Seven days of budget, divided by your cost per result, is the number of results each ad set can expect. Reach 50 and the ad set has the budget for the point at which Meta says learning usually ends. Fall short and the calculator works out the budget that would get there, how many ad sets your total can fund at that level, and what to change.
- Cost per result is the cost of your optimization event: the outcome the ad set is optimized for, which Meta notes can differ from the campaign objective. Optimize for purchases and it is your cost per purchase.
- Each ad set spends its full budget every day. Meta treats a daily budget as an average over the week and says a week won’t spend more than 7 times it.
- All ad sets share the cost per result you enter. If yours differ, run each ad set on its own.
- Advantage+ campaign budget: choose One total budget. Meta distributes the budget across ad sets, so the per-ad-set figure is an average.
- Shops ads follow a different rule, at least 17 purchases through your website and 5 through Meta, so the 50 here does not apply to them.
Sizing the total from a revenue goal first? Start with How much should you spend on Meta ads?
What is the formula?
- Results a week, per ad set
- daily budget per ad set × 7 ÷ cost per result
- Daily budget needed, per ad set
- 50 × cost per result ÷ 7
- Ad sets a total budget can fund
- total daily budget ÷ daily budget needed
- Highest cost per result that still reaches 50
- daily budget per ad set × 7 ÷ 50
The 7 is Meta’s window, the week after the ad set’s last significant edit, and 50 is Meta’s “about 50 results”. The first line rearranges Meta’s definition of cost per optimization event: the total spent divided by the number of optimization events. Rounding never flatters the answer: amounts count in whole cents, a total split across ad sets rounds down to the cent, results and ad sets round down, the budget needed rounds up to the cent, and the highest cost per result rounds down.
Meta also says “some ad sets stabilize earlier”, and its Marketing API now reports a per-ad-set “dynamic learning phase” with its own conversions threshold and days to exit. Read 50 in a week as the typical case, a planning target rather than a fixed rule.
Will these ad sets exit learning?
Five example setups, run through the same code as the calculator. The inputs are illustrations, not benchmarks.
| Setup | Budget per ad set | Cost per result | Results a week | Reaches 50? | Budget needed |
|---|---|---|---|---|---|
| $300 a day split across 3 ad sets, optimizing for purchases | $100 | $30 | 23 | No | $214.29 |
| The same $300 a day in 1 ad set | $300 | $30 | 70 | Yes | $214.29 |
| 3 ad sets of $100, optimizing for add to cart | $100 | $6 | 116 | Yes | $42.86 |
| A $50-a-day test ad set, optimizing for purchases | $50 | $40 | 8 | No | $285.72 |
| A higher-priced product, optimizing for purchases | $500 | $60 | 58 | Yes, with little room | $428.58 |
The first two rows spend the same $300 a day. Split three ways, no ad set reaches 50; in one ad set it gets 70. The third row is Meta’s own suggestion, a cheaper and more frequent event. Whether add to cart is a fair stand-in for your sales is a separate question, answered in the FAQ below.
What can you do if an ad set can’t reach 50 a week?
Meta’s fixes for Learning limited are to combine ad sets, expand the audience, raise the budget, raise a low bid or cost control, or optimize for a more frequent event. Here they are as steps, in the order we would try them:
- Check the real count. In Ads Manager, open Columns, click Customize columns, tick Optimization events and add Last significant edit. Meta counts the week from that edit.
- Combine ad sets that share an objective and creative until each can fund 50 a week. Meta says combining gets you results faster, and Account Overview may suggest which to merge.
- Raise the budget to the calculator’s figure in the ad set’s editing pane (the campaign’s, with Advantage+ campaign budget). Make one change: Meta says frequent or large budget changes can restart learning.
- Switch to a more frequent event in the ad set’s Conversion section, such as purchases to add to cart, Meta’s own example. That is a significant edit, so learning starts again.
- Expand a narrow audience in the ad set’s editing pane. Meta says the larger the audience, the more people there are to complete your optimization event.
- Raise a bid cap or cost per result goal set below your real cost (Conversion section). Hover over Learning limited: Meta shows a tooltip if a low bid or cost control is the cause.
- Wait while the Delivery column reads Learning. Meta advises holding edits until the ad set exits, because a significant edit resets learning; make one change and let it run.
For the six conditions behind the status, read What causes a Meta ad set to be marked learning limited? The learning check is also one of the 30 in our Meta ads audit checklist.
Learning phase FAQ
How many results does an ad set need to exit learning?
About 50 in the week after its last significant edit, says Meta, adding that some ad sets stabilize earlier. Meta counts your optimization event, the outcome the ad set is optimized for, so it means purchases if you optimize for purchases. Shops ads have their own rule: at least 17 purchases through your website and 5 through Meta.
Source: Meta, About the learning phase (checked 1 Oct 2026)
What does Learning limited mean?
Meta shows Learning limited in the Delivery column when an ad set is unlikely to get about 50 optimization events in the week after its last significant edit. Meta says it isn’t a penalty: the budget isn’t being spent effectively because delivery can’t optimize with the current setup.
Source: Meta, About learning limited (checked 1 Oct 2026)
Read next: What causes a Meta ad set to be marked learning limited?
Which edits restart the learning phase?
Meta lists any change to targeting, creative or optimization event, adding a new ad, pausing for 7 days or longer, and changing bid strategy. Budget, bid and spending-limit amounts may or may not count, depending on size: Meta’s example is that $100 to $101 is unlikely to, while $100 to $1,000 may.
Source: Meta, Significant edits and learning phase (checked 1 Oct 2026)
Does Advantage+ campaign budget change the math?
Not the threshold. Meta says ad sets under Advantage+ campaign budget take the same time to learn as ad sets with their own budgets, and that moving budget between them does not restart learning. Use the total-budget option here: because Meta distributes the budget, the per-ad-set figure is an average.
Source: Meta, Significant edits and learning phase (checked 1 Oct 2026)
Should I optimize for add to cart instead of purchases?
Only if purchases can’t reach 50 a week at a budget you can afford. Meta suggests the switch, but the optimization event is what its system bids on, so it then looks for people likely to add to cart, not necessarily to buy. Compare cost per purchase before and after.
Source: Meta, Optimization events (checked 1 Oct 2026)
Sources: Meta Business Help Center, About the learning phase, About learning limited, Significant edits and learning phase, Optimization events, Cost per optimization event, Combine ad sets and campaigns and Troubleshoot ad delivery; Meta Marketing API v26.0, Ad Campaign Learning Stage Info (all checked 1 Oct 2026). Drafted with AI; facts checked against these Meta pages on 1 Oct 2026.
Who made this calculator?
Bach.ai is our product. It connects to your Meta ad account and audits it; among its checks, it flags ad sets in learning or Learning limited and accounts where no ad set reaches 50 results a week, then proposes a fix. It applies a change only after you approve it. Free is a 7-day full trial with no card, then 1 audit a month. Start the free trial.