First, read the ad group status and the review queue
Open the campaign and expand every ad group and ad. Read the status column at each level rather than the campaign's own label, because a campaign can show as active while every ad beneath it waits in review or has been rejected. Rejected ads carry a reason; note it. Where Ads Manager offers diagnostic suggestions for the ad group, read them too, as the TikTok Ads hub recommends treating the help centre's own pages as the reference for anything that changes.
If every ad is approved and live, open the delivery history for the past week: impressions, spend and the learning status by day. Zero impressions from launch with everything approved points at price or audience. A burst of delivery that fades, or an ad group labelled as learning that never progresses, points at the optimisation event or the budget structure. Spend shared unevenly across many ad groups, each with a thin slice, points at fragmentation. Write down which pattern you see; each maps to a different row below.
Diagnosis table
Diagnosis
| Symptom | Check | Likely cause | Fix |
|---|---|---|---|
| Approved ads, almost no impressions, cost cap or bid cap set | The cap against the cost per result the account has actually paid | Bid or cost cap below what the audience costs | Raise the cap to a realistic level, or use lowest cost while learning |
| Value-optimised ad group stops soon after launch | Minimum return target against the return the shop has actually achieved | Return-on-spend target so ambitious that delivery stops | Lower the target to an achieved level, or optimise for purchases first |
| Low impressions with a small estimated audience | Audience estimate; number of interest and behaviour layers; expansion setting | Audience too narrow for the objective with expansion off | Remove layers, turn expansion on, or broaden location and age |
| Campaign active, ads not serving | Ad-level status and rejection reasons | Ads still in review or rejected | Fix the stated reason and resubmit; submit early next time |
| Learning status never completes; delivery erratic | Weekly count of the optimisation event in the ad group | Event too rare to leave learning | Optimise for a more frequent event higher in the funnel |
| Many ad groups each spending a sliver | Number of ad groups per campaign; budget per ad group | Budget split across too many ad groups | Consolidate into fewer ad groups with more budget each |
Fix, verify and prevent each cause
A price ceiling under the going rate
A cost cap asks delivery to keep the average cost per result near a figure, and a bid cap limits what any single auction may cost. Both are useful once a baseline exists; set before one exists, they often sit below the price the audience commands, and the ad group declines nearly every auction.
Fix: compare the cap with the cost per result (CPACost of each conversion.) the account has achieved for this event, and either raise it to that level or run the lowest-cost strategy until there is a baseline, as the campaign structure guide describes for the documented strategies.
Verify: impressions and spend rise within a few days, and achieved cost per result sits near the new cap rather than far below it.
Prevent: set caps only from the account's own results for the same event and market, never from another platform.
A return target the shop cannot reach
Value optimisation can carry a minimum return target. When the target exceeds what the products and margins actually produce, the system finds too few qualifying auctions and delivery stalls. The campaign structure guide notes that an over-ambitious minimum return simply stops delivery rather than finding cheaper sales.
Fix: set the target from the return on ad spend (ROASRevenue generated per unit of ad spend.) the account has achieved, or optimise for purchase count until enough valued purchases exist, following the campaign structure guide.
Verify: spend recovers while achieved return stays near the new target.
Prevent: raise return targets in small steps from proven results, and only when spend is already flowing.
An audience that leaves delivery nowhere to go
TikTok's ranking does much of the audience finding, so an ad group stacked with interests, behaviours, a narrow age band and expansion switched off gives it very little room. The estimate in the ad group shows how small the pool has become.
Fix: remove the narrowest layers, widen location or age where the offer allows, and turn on targeting expansion, as the campaign structure guide suggests for prospecting.
Verify: the audience estimate grows and delivery follows within days, with cost per result holding.
Prevent: keep narrow ad groups for specific hypotheses such as retargeting, and start prospecting broad with sensible exclusions.
Ads waiting in review or turned down
Every ad passes review before it can serve, and a rejection carries a reason in Ads Manager. Common reasons include claims, landing page problems and, for Spark posts, sounds not cleared for commercial use.
Fix: read the rejection reason, correct the ad or the page, and resubmit; the TikTok Ads hub lists the usual causes, and creator posts with sound issues are covered on the Spark Ads code troubleshooting page.
Verify: the ad shows an approved status and records impressions the same day.
Prevent: submit ads ahead of the planned start so review finishes before the budget is meant to run.
An event that happens too rarely to learn from
An ad group learns from conversions of the event it optimises for. TikTok documents a learning phase that ends after a run of conversions in the opening days; if the event is a purchase on a low-traffic shop, the ad group may never collect enough and delivers erratically.
Fix: optimise for a more frequent event, such as adding to cart or initiating checkout, until the deeper event flows, as the campaign structure guide advises. Confirm in the TikTok measurement hub that the event actually arrives.
Verify: the learning status completes and daily delivery steadies.
Prevent: before launch, estimate how often the chosen event occurs at the planned budget, and choose the deepest event that happens often enough.
A budget sliced too thin
Splitting a budget across many ad groups gives each one too little to compete or to learn, and the ad groups also overlap and bid against one another. The symptom is a campaign spending a fraction of its budget spread across many ad groups with tiny delivery each.
Fix: consolidate into a few ad groups, one per hypothesis, and move creative tests into ads within an ad group rather than separate ad groups, as the campaign structure guide recommends.
Verify: the surviving ad groups spend most of their budget and progress through learning.
Prevent: cap the number of ad groups per campaign at what the budget can feed, and add new ones only by retiring old ones.
Prevention: launch rules for budget, event and audience
Most stalled ad groups were built to stall. A few launch rules stop that. Choose an optimisation event that will occur often enough at the planned budget. Give each ad group enough budget to reach the documented learning threshold, and keep the number of ad groups small. Start prospecting broad with expansion available; reserve narrow audiences for retargeting. Leave caps and return targets off until the account has its own baseline. Submit ads early enough to clear review. Then hold edits for the first days, read status, learning and spend daily, and compare the cost of a thousand impressions (CPMCost per 1,000 impressions.) with earlier campaigns to see whether price or volume is the issue. When something does stall, the launch record shows which rule was bent. The TikTok Ads launch QA checklist is a ready-made launch record that also covers the pixel, creative and Spark Ads codes. The TikTok Ads troubleshooting hub lists the related problems once delivery flows, and when an ad group spends freely but each result costs more each week, the diagnosis for a rising TikTok cost per result takes over.
A skincare brand whose ad groups never left learning
A skincare brand launches one website conversions campaign with £300 a day split evenly across 10 ad groups, each targeting a different interest cluster, all optimising for purchases with a cost cap of £18. After a week the campaign has spent £640, every ad group still shows as learning, and the shop has recorded 21 purchases from TikTok across all ten. The brand's purchase cost on other channels averages £26, so the cap sits well below what a sale costs. The team consolidates into 2 broad ad groups with expansion on, switches to the lowest-cost strategy, and optimises for add to cart while purchases are scarce. Over the next week the campaign spends £1,960 of a possible £2,100, records 412 add-to-cart events and 58 purchases, and both ad groups complete learning by day five.
Common mistakes
Raising budget when a cap is the limit. Setting a return target from a hoped-for figure. Splitting one budget into many interest ad groups. Optimising for purchases on a shop that sells a few items a week. Editing daily while the ad group learns.
