Est. reading time: 6 minutes
An ad set to spend money that isn’t producing views has one of two problems, and they need different doctors. Either TikTok isn’t delivering the ad at all, which is mechanical and usually fixable in an afternoon, or TikTok is delivering it and the auction keeps deciding your ad isn’t worth showing, which is a creative and structural problem. Most advertisers jump straight to rewriting hooks when the actual issue is a payment method or a bid cap. Run the triage first.
First, rule out the mechanical failures
Open Ads Manager and check the delivery column before touching anything else. An ad still in review hasn’t started. A rejected ad never will, and that’s its own diagnosis with its own fixes. Beyond status, the quiet killers are a payment method that failed silently, a schedule with dates that haven’t started or already ended, a Cost Cap bid set below what the auction actually clears (the ad group simply won’t spend), and an audience that got stacked and excluded down to a size too small to deliver against. Learning Limited status means the ad group can’t accumulate enough conversion volume to stabilize, which points at budget or optimization event rather than creative.
If everything checks out mechanically and impressions are still crawling, the problem is that you’re losing auctions, and that’s worth understanding before you fix it.
Why TikTok ads don’t get views: you’re losing an auction you can’t see
TikTok doesn’t hand impressions to the highest bidder. It ranks competing ads by effective CPM, which blends your bid with the system’s prediction of how users will respond to your ad. Predicted watch time and engagement are part of the price. An ad the system expects people to skip has to pay more for every impression, and below a certain threshold it barely delivers at all, regardless of budget.
This is the mechanism that makes creative a delivery lever and not just a conversion lever. “Not getting views” and “creative looks like an ad” are usually the same problem stated twice, and it’s why the fixes below start with the video rather than the settings.
Make TikToks, not ads
The feed skips anything that announces itself as advertising, so the first job is removing the ad signals. Handheld energy over studio polish, a human face leading, bold captions for sound-off viewing, subtle branding rather than a corporate open, and music from TikTok’s Commercial Music Library so the track is both licensed and native. Nine to twenty seconds is the working range, with a structure that runs hook in the first three seconds, problem or desire by six, demonstration and proof through fifteen, and a soft close. Put the product in hand within the first two seconds, and open like a creator (“I stopped using X for 7 days and…” or “3 things I wish I knew before buying Y”) rather than like a brand.
Then manage creative by numbers instead of taste. Our working baselines are a 2-second hold rate above 30 percent, a 6-second view rate above 15 percent, and CTR above 1 percent, with three to five variants per ad group and a weekly refresh cadence on accounts with real spend. Retire an ad when its hold rate or CTR drops roughly 20 percent from its own baseline, because fatigue on TikTok arrives fast and compounds, a curve we mapped in keeping your ads fresh without burning out. The underlying principle, that engagement quality buys you cheaper delivery, isn’t TikTok-specific either, and we’ve written about the Meta version of the same lever.
Spark Ads deserve a specific mention for view problems, because promoting an organic post (yours or a creator’s, with authorization) carries its real comments, shares, and watch history into the auction, and inherited engagement is exactly the signal that predicted-performance ranking rewards.
Targeting that lets delivery breathe
Delivery that crawls often traces to an audience that’s either too tight or assembled at random. Start broad, meaning your core geo and language, a wide age range unless the product genuinely constrains it, and TikTok-only placements, then let the optimization goal do the sorting while creative proves itself. Keep estimated audience sizes in the several-million range or larger, run single-theme interest ad groups rather than ten stacked interests, and leave targeting expansion on so the system can step outside your definition when it finds better performance there.
Layer in your own data as it accumulates. With the Pixel and Events API feeding events, build custom audiences from visitors, add-to-carts, and purchasers on 30-to-180-day windows, cut 1 to 3 percent lookalikes from your best-value customers, and exclude recent buyers so the budget stops re-buying people who already converted. The full setup lives in Mastering TikTok Ads: Strategies for Maximum Reach and Conversions. One caution as retargeting pools grow, which is that a slice of every pool will never respond no matter the frequency, and recognizing that segment beats hammering it, the subject of why some visitors never respond to retargeting.
Check the optimization event against your volume, because this one masquerades as a views problem constantly. Purchase optimization with fewer than roughly 50 conversions a week per ad group leaves the system without enough signal to learn, and delivery throttles. The fix is optimizing temporarily for a higher-funnel event like add-to-cart while purchase volume builds, then stepping back down. When comparing broad against interest against lookalike, use the A/B test tool rather than eyeballing parallel ad groups, so the comparison doesn’t contaminate the learning it’s measuring.
Budget and bids that can actually win
Underfunded ad groups starve quietly. Our working floor is a daily budget of 20 to 50 times target CPA per ad group, and never below TikTok’s practical minimums of $20 to $50 for testing, because a budget that can’t buy enough events to exit learning isn’t a lean budget, it’s a stalled one. Scale in 20 to 30 percent steps every 48 hours rather than doubling overnight, and batch your changes, since every significant edit to budgets, bids, or creative destabilizes what the system has learned.
On bidding, start with Maximum Delivery (the strategy formerly labeled Lowest Cost) to buy data quickly and establish a real CPA baseline. If costs spike once volume exists, test Cost Cap set 10 to 20 percent above your true target so delivery keeps breathing while costs come under control, and if delivery sticks under a cap, raise the bid, broaden the audience, or drop back to Maximum Delivery to re-prime spend. Structure stays boring throughout, one to three ad groups per campaign with three to five creatives each, testing horizontally with new hooks and angles before scaling vertically, then consolidating proven winners under Campaign Budget Optimization so spend flows to the best performers on its own.
Work the sequence in order and the views problem resolves into its parts. Mechanical checks first, then creative the auction wants to show, then audiences and events the system can learn from, then budgets and bids that can win. And once the impressions return, make sure they land somewhere worth the trip, because a destination that sends visitors straight back is the next leak in line, one we’ve diagnosed in the design mistake that’s sending visitors back to Google.
