Paid Ads & Creative
8 Rules: TikTok Ad Creative Best Practices From Research

Joshua Jackai
20 min

Published 17 November 2026 · Last updated and fact-checked November 2026
Short answer: The TikTok ad creative best practices that hold up are eight rules: show what you sell in 3 seconds, open on a rising emotion, put entertainment after the brand, shoot native on a phone, cast a creator like your buyer, write readable captions, test 3 to 5 different creatives, and refresh before wearout.
Here is a pattern I see often, shown as an example. A DTC skincare brand has a $15,000 monthly TikTok budget and twelve ads that all start the same way: a logo, a slow pan across the bottle, music, and the product name at second eight. Spend goes out, views come in, and almost nobody stays past the third second. The ads are not ugly. They are built for a TV slot, not a feed where the thumb decides in a moment.
I make ad creative for DTC brands and B2B teams, and this page is how I brief TikTok ads before anyone films. TikTok's own creative best practices page gives the platform rules. I add the research that explains why those rules work, so you can judge a creative instead of copying a checklist. Each of the eight rules below has a shot list, the study behind it and a visual you can hand to a creator.
Key takeaways
Elsen, Pieters and Wedel (2016) found in the Journal of Marketing Research that ads which show what they promote at once were rated well at exposures as short as 100 milliseconds; mystery ads were not.
Teixeira, Wedel and Pieters (2012) found the speed at which joy rises, more than its level, kept viewers from zapping away.
Teixeira, Picard and el Kaliouby (2014) tested 82 TV ads on 178 consumers: entertainment after the brand linked to purchase intent; entertainment before it did not.
Hautz, Füller, Hutter and Thürridl (2014) found user videos beat agency videos on trust only when both were low in technical quality.
TikTok's help center (vendor source, last updated June 2025) recommends 9:16, at least 720P, the content proposition in the first 3 seconds and 3 to 5 creatives per ad group.
Rule | What it fixes | Key number | Source type |
|---|---|---|---|
1. Say what you sell in the first 3 seconds | Viewers who never learn what the ad is for | Ads judged at 100 milliseconds | Peer reviewed |
2. Open on a rising emotion | Early scroll-aways | Speed of joy drove retention | Peer reviewed |
3. Put the entertainment after the brand | Fun ads that sell nothing | 82 ads, 178 consumers | Peer reviewed |
4. Shoot native on a phone | Polished ads that read as ads | 2 by 2 experiment, user vs agency | Peer reviewed |
5. Cast a creator who looks like your buyer | Creators who do not move your audience | Similarity linked to trust in branded posts | Peer reviewed |
6. Write captions people can read | Sound-off viewers who miss the offer | 5 to 10 words per second | Vendor (TikTok) |
7. Test 3 to 5 very different creatives | Tests that only compare hook tweaks | 3 to 5 creatives per ad group | Vendor (TikTok) |
8. Refresh before wearout | Ads that slowly stop working | Wearout from repetition | Peer reviewed |
1. Say what you sell in the first 3 seconds: best fix for ads nobody understands
The first rule is to show the product and the problem it solves before the viewer can scroll. Elsen, Pieters and Wedel (2016) ran three experiments with exposures from 100 milliseconds to 30 seconds and found upfront ads, which instantly convey what they promote, were evaluated well at both brief and long looks.
When: frame 1 to second 3
What appears: the product, in use or in hand
What is said: the problem, in the buyer's words
What to cut: logo intros, slow pans, "wait for it" reveals
TikTok's rule: "Introduce your content proposition in the first 3 seconds" (help center, June 2025)
The finding that changes briefs is what happened to the other ad types. Mystery ads, which hold back what they promote, were rated poorly after brief exposures and only recovered after long ones. On TikTok almost every exposure is brief, because the viewer controls the scroll. A mystery ad only works for the few people who stay, and most do not.
The researchers also tested false front ads, which open as one thing and turn out to be another. Those did well at first glance and then lost ground the longer people watched. That is the pattern of a clickbait hook: a strong open, then a feeling of being tricked. I avoid hooks that promise something the ad does not deliver.
In practice I write the first three seconds before anything else. If a creator cannot say what the product does and show it in that window, the concept is not ready to shoot.
"Upfront ads, which instantly convey what they promote, are evaluated positively after brief but also after longer exposure durations," Elsen, Pieters and Wedel reported in the Journal of Marketing Research.

2. Open on a rising emotion: best fix for early scroll-aways
The hook should climb, not plateau. Teixeira, Wedel and Pieters (2012) tracked faces, eye movements and zapping for internet video ads and found that surprise concentrates attention and joy keeps viewers watching, with the speed of rising joy mattering most for retention.
My brief, second 0 to 1: a surprise (an odd visual, a result shown first, a bold claim you can prove)
Second 1 to 6: joy that climbs (a reaction, a reveal, a small win)
Avoid: a flat open that stays at one level
Measure: 2-second and 6-second view rates per creative
TikTok's rule: "Prioritize your hook in the first 6 seconds" (help center, June 2025)
The detail that matters for a hook is the split between the two emotions. The level of surprise drove attention most, while the velocity of joy, how fast it rose, drove whether people kept watching. So a surprising first frame gets eyes on the ad, and a quick climb in good feeling keeps them there.
The study also found joy is asymmetric. Gains helped more than drops hurt. That gives a creator room to show the problem honestly before the fix, as long as the ad climbs once the product appears.
Belanche, Flavián and Pérez-Rueda (2017) found something similar for skippable video ads: high-arousal ads were watched for longer and worked better when they fit the content around them. My reading: on TikTok that context is other creators' videos, which is one more reason the open should look like one. An ad that opens like a TV spot breaks the fit before the hook has a chance to work.
"High-arousal ads are watched for longer time and are more effective in congruent contexts," Belanche, Flavián and Pérez-Rueda reported in the Journal of Interactive Marketing.
"The velocity rather than the level of joy affects viewer retention most," Teixeira, Wedel and Pieters wrote in the Journal of Marketing Research.

3. Put the entertainment after the brand: best fix for fun ads that sell nothing
Entertainment works on TikTok, but where it sits relative to the brand decides whether it sells. Teixeira, Picard and el Kaliouby (2014) measured smiles, viewing interest and purchase intent for 82 ads shown to 178 consumers and found entertainment after the brand appeared was linked to purchase intent, while entertainment before it was not.
Brand moment: product or name on screen by second 3
Entertainment: the joke, trend or reaction after that point
Dose: enough to keep watching, not a sketch with a product cameo
Check: can a viewer name the brand at second 5?
Format link: pairs with rule 1 (upfront) and rule 2 (rising joy)
The same study of TV ads found an inverted U-shape between entertainment and purchase intent. Some entertainment helped, and past a point more of it went with lower purchase intent. The abstract does not say why. For a DTC founder, the pattern to watch in trend-led TikTok ads is high views, high likes and a cost per purchase that never comes down.
My rule when I brief a trend format is that the product has to be the reason the trend works in this video. If you could swap in any other brand and the joke still lands, the entertainment is sitting before the brand in practice, even if the logo appears early.
This is also why I put the brand in the creator's hands rather than as an end card. In my experience a product used inside the funny moment gets the credit, and an end card after it often does not. The study compared entertainment before and after the brand in TV ads, not this specific case. When I review a cut, I pause at second 5 and ask whether a stranger could name the product. If not, the edit goes back.
"We separate entertainment into that which comes before the brand versus that which comes after, and find that the latter is positively associated with purchase intent while the former is not," Teixeira, Picard and el Kaliouby wrote in Marketing Science.

4. Shoot native on a phone: best fix for polished ads that read as ads
TikTok ads should look like TikToks. Hautz, Füller, Hutter and Thürridl (2014) ran a 2 by 2 experiment of video source (user vs agency) by technical quality and found that at low quality, user videos were rated more trustworthy and expert than agency videos. At high quality the trust edge disappeared.
Camera: a phone, held or propped, vertical 9:16
Resolution: at least 720P (TikTok help center, June 2025)
Light: window light or a ring light, not a studio set
Sound: a real voice to camera or natural sound, plus captions
Safe zone: keep text and faces clear of the TikTok interface
The result is evidence that polish does not buy trust by itself. The user video held its trust advantage only at low technical quality. At high quality the trust gap disappeared, although users were still rated higher on expertise, by a smaller margin. The test used videos promoting the Tyrol region in Austria, not TikTok ads, so I treat it as a reason to test a native cut, not as proof.
That does not mean careless footage. Sound has to be clear and the product has to be in focus. It means the framing, pace and talking style match what people already watch in the For You feed. I brief creators to film the way they film their own videos and keep studio lights out of the room. If a creator offers to reshoot with better gear, I usually say no and ask for a second take of the opening instead.
The numbers that matter here come from TikTok's own spec guidance: 9:16, at least 720P and content inside the UI safe zone, so captions and the product are not hidden behind the like button or the caption bar.
"There is no general superiority of UGVs over AGVs," Hautz and colleagues reported in the Journal of Interactive Marketing, where UGVs are user-generated videos and AGVs agency-generated videos; the effect depended on technical quality.

5. Cast a creator who looks like your buyer: best for Spark Ads that scale
Cast for how your buyer sees the creator, not for audience size. Lou and Yuan (2019) surveyed users who followed at least one influencer and found informative content, trustworthiness, attractiveness and similarity to followers were each linked to more trust in branded posts, which was linked to awareness and purchase intentions.
Cast for: age, problem and setting that match your buyer
Run as: a Spark Ad from the creator's handle, or in-feed from yours
Permission: written usage rights and the in-app authorization code
Disclosure: TikTok's paid partnership label plus the creator's own words
Start with: the concept already proven from your brand handle
Spark Ads vs in-feed ads is a common question. A Spark Ad boosts an existing organic post, so it runs from the creator's handle with their profile and comments. A standard in-feed ad runs from your brand's account. I test a concept from the brand handle first, then recast it with a similar creator and run it as a Spark Ad once it has earned the spend.
Disclosure is not optional. The FTC's Endorsement Guides expect a paid creator's material connection to the brand to be disclosed clearly. Boerman, Willemsen and Van Der Aa (2017) found a "Sponsored" label changed how people processed a post only when a celebrity posted it, not the brand. So disclose every time. In that Facebook experiment, the label led readers to recognize the post as advertising, distrust it more and say they were less likely to share it, and only when a celebrity posted. The study did not test smaller creators or TikTok. I cast creators who look like the buyer because of the Lou and Yuan findings, not because of this study.
"Influencer's trustworthiness, attractiveness, and similarity to the followers positively affect followers' trust in influencers' branded posts," Lou and Yuan reported in the Journal of Interactive Advertising.

6. Write captions people can read: best fix for sound-off viewers
On-screen text carries the offer for anyone watching without sound, and it can be read too fast. TikTok's help center (vendor source, June 2025) recommends captions or text overlays for context and "displaying 5-10 words per second when using text."
Rate: 5 to 10 words per second, at most
Size: large enough to read on a phone at arm's length
Position: inside the safe zone, above the caption bar
Job: the problem, the proof and the offer, in that order
Sound: keep voice and music; captions back them up
There is research on why text size matters. Pieters and Wedel (2004) analyzed 1,363 print ads tested with eye tracking on more than 3,600 consumers. The picture captured attention whatever its size, the brand moved attention to the other elements, and text captured attention in direct proportion to its surface size. That study is print, not video, so I treat it as a reason to set captions big rather than as a TikTok benchmark.
In a 15-second ad at 5 to 10 words per second, that is room for about 75 to 150 words at the most. I aim far lower. Three short lines in the first six seconds usually do more than a paragraph that runs the whole ad. I write the captions from the script, then cut every word a viewer could lose without missing the offer. Each line stays on screen long enough to read twice, and the offer line gets the largest type in the ad.
"The text element best captures attention in direct proportion to its surface size," Pieters and Wedel reported in the Journal of Marketing.

7. Test 3 to 5 very different creatives: best fix for tests that learn nothing
TikTok recommends 3 to 5 creatives per ad group and says it is better to use creatives with big differences, especially when testing (help center, June 2025, vendor source). I read that as: test different concepts, not five versions of the same hook.
Per ad group: 3 to 5 creatives
Per campaign: 3 to 5 diversified ad groups (TikTok's guidance)
What differs: the concept, creator, opening or format
What stays: the offer and landing page during the test
Measure: cost per purchase per creative, plus a holdout for lift
Platform numbers tell you which creative won inside the platform. They do not tell you whether the ads caused sales that would not have happened anyway. Johnson, Lewis and Nubbemeyer (2017) built a "ghost ads" method to compare exposed buyers with a true control group, and in one retailer's retargeting campaign the ads lifted website visits by 17.2% and purchases by 10.5%. That number is one campaign's display retargeting, not a TikTok benchmark. The point is the method: when the budget allows, run a holdout or a conversion lift study before you scale a winner.
My test sheet has one row per creative and four columns: the concept in five words, the 2-second view rate, the 6-second view rate and cost per purchase. A creative that wins on view rate and loses on cost per purchase is a good hook on the wrong offer. A creative that loses on view rate and wins on cost per purchase is a good offer with a weak open. Both get a second version.
"Ghost ads can reduce the cost of experimentation, improve measurement precision, deliver the relevant strategic baseline, and work with modern ad platforms," Johnson, Lewis and Nubbemeyer wrote in the Journal of Marketing Research.

8. Refresh before wearout: best fix for ads that slowly stop working
Research on repeated exposure finds that ads wear out, so plan the next TikTok creative before the current one fades. Calder and Sternthal (1980) found that repeating a set of TV commercials caused wearout in how viewers rated both the ads and the products, and that tactics designed to raise attention did not stop it.
Trigger: "a consistently declining trend" in delivery results (TikTok help center)
Method: add new creatives to the existing ad group
Cadence: a new concept batch every two weeks (my practice)
Signals: rising cost per purchase, falling 6-second view rate
Weekly check: keep, rest or kill each creative
Calder and Sternthal found that attention tricks did not prevent wearout, so I plan each refresh around a new concept rather than a small edit (my practice, not a finding of that study). Lehnert, Till and Carlson (2013) compared creative and control TV commercials. They found creative ads were recalled better, were more liked, wore in faster and were less susceptible to wearout, though repeat exposure narrowed their recall edge. That points toward original concepts lasting longer, but it is TV evidence, not TikTok data.
TikTok's own guidance is to refresh when delivery shows a consistently declining trend or when daily new users are low, and to add the new creatives to an existing ad group instead of creating a new one. It gives no fixed number of days. I set a two-week rhythm for new concepts so there is always something ready when a winner starts to fade. For more on reading the signals, see my guide to creative fatigue on Meta ads; the same signals apply on TikTok.
"Wearout was not forestalled by strategies designed to enhance attention," Calder and Sternthal reported in the Journal of Marketing Research.

TikTok ad creative rules compared: matrix by problem
Use this matrix to pick the rule that fixes the problem you see in your numbers first. Find the symptom in your TikTok Ads Manager report, make the first change listed, and leave the other seven rules alone until that change has run for a full week of spend.
Rule | Symptom in your numbers | First change to make | Source |
|---|---|---|---|
Say what you sell in 3 seconds | Low 2-second view rate | Product and problem in frame 1 | Elsen et al., 2016 |
Open on a rising emotion | Drop between second 2 and 6 | Surprise first, then a quick climb | Teixeira et al., 2012 |
Entertainment after the brand | High views, weak purchases | Brand by second 3, joke after | Teixeira et al., 2014 |
Shoot native on a phone | Low engagement on polished ads | Creator-shot phone version | Hautz et al., 2014 |
Creator who looks like your buyer | Spark Ad reach without sales | Recast with a similar creator | Lou and Yuan, 2019 |
Readable captions | Weak results on sound-off views | 5 to 10 words per second, large type | TikTok help center, 2025 |
Test 3 to 5 different creatives | Tests with no clear winner | Different concepts, not hook tweaks; a holdout to measure lift | TikTok help center, 2025; Johnson et al., 2017 |
Refresh before wearout | Cost per purchase climbing weekly | New concepts in the same ad group | Calder and Sternthal, 1980 |
Calculator: how many TikTok creatives your budget can feed
The calculator splits your budget across the creatives in one ad group so you can see how many purchases each one gets a week. With $15,000 a month, a $40 target cost per purchase and 5 creatives, each creative gets about 17 purchases a week.
How to use it with your own numbers:
Monthly budget: your TikTok spend; $15,000 is a placeholder
Target cost per purchase: your current average; $40 is a placeholder
Creatives per ad group: TikTok suggests 3 to 5
Weekly purchases per creative: budget divided by cost per purchase, by 4.33 weeks, by creatives
Read: if each creative gets only a handful of purchases a week, test fewer creatives at once
Worked example: $15,000 divided by $40 is 375 purchases a month, or about 87 a week. Split across 5 creatives, each gets about 17 a week. Split across 3, each gets about 29. With fewer purchases per creative, a winner takes longer to show, so I start smaller budgets at 3 creatives.
What are TikTok ad creative best practices?
TikTok ad creative best practices are the rules for making ads that work in the For You feed: vertical 9:16 video, the offer in the first seconds, a native look, readable captions, sound, a clear call to action, and steady testing and refresh. TikTok publishes its own version in its ads help center.
The platform rules are a starting point. TikTok's page says its recommendations are general practices that have worked for advertisers. The research above explains why they work, which helps when your numbers say a rule is not working for your product.
How long should a TikTok ad be?
A TikTok ad should be as long as it takes to show the problem, the product and the offer, and no longer. TikTok's help page sets no duration rule. It asks for the content proposition in the first 3 seconds and the hook in the first 6, so the length has to fit that structure.
I cut most concepts at 15 to 25 seconds and make a shorter cut of the same concept (my practice, not a platform rule). The Elsen study found upfront ads were rated well at both brief and long exposures, while mystery ads recovered only with longer ones. Since most TikTok viewers give a brief look, I state the product at the start and use the remaining seconds for proof.
How much do TikTok ads cost to make?
TikTok ad creative costs the creator's fee, the usage rights for paid ads, editing and any product shipped. Rates vary widely by creator and by how long you want to run the video, so ask for a written quote that lists all four lines.
I have no source I trust for a market average, so I will not quote one. What I can state is my own pricing. Demand Designer makes TikTok and Meta ad creative as part of the Conversion Creatives plan, my monthly ad creative subscription, at a fixed $999 or $2,449 a month (USD), with briefs, casting notes and edits included.
Frequently asked questions
What is the best hook for a TikTok ad?
The best hook shows the product and the problem in the first 3 seconds and opens on a surprise. Research found surprise holds attention and fast-rising joy keeps viewers watching. Test 3 to 5 openings that differ in concept, and keep the best 6-second view rate.
Are Spark Ads better than in-feed ads?
Neither wins by default. Spark Ads run from a creator's handle and borrow their profile and comments; in-feed ads run from your account. I prove a concept from the brand handle first, then recast it with a similar creator as a Spark Ad.
Do TikTok ads need sound?
Yes, sound is part of TikTok's own creative basics, but captions should carry the offer too. TikTok recommends captions or text overlays for context and 5 to 10 words per second on screen. Keep the voice or music, and make the captions large and inside the safe zone.
How often should I refresh TikTok ad creative?
Refresh when results show a steady decline, TikTok's own trigger, and add new creatives to the same ad group. Research found repetition wears ads down and attention tricks do not stop it. I add a new concept batch every two weeks.
Should TikTok ads use influencers or customers?
Use whoever looks and sounds like your buyer. A survey study found similarity to followers, trustworthiness and useful information raised trust in branded posts. Real customers carry that naturally; paid creators need careful casting and a clear paid partnership disclosure every time.
How I make TikTok ads for DTC brands
Demand Designer makes TikTok ads as one monthly creative system. Each batch has at least three different concepts, each written to the eight rules above, with the first three seconds scripted before anything else, creators cast to match the buyer and written permission and disclosure handled before filming.
Every concept is cut in a TikTok-native version and a shorter version, captioned inside the safe zone, then sent through my ad creative testing steps. For the formats I brief most often, see my UGC ads examples, and for the choice between static and video on Meta, see static vs video ads. The wider plan sits in my Meta ads creative strategy rules, which use the same research.
If you want a quick weekly way to decide which TikTok ads to keep, rest or kill, start with my free Kill Sheet: five signals, three verdicts and ten minutes every Monday.
Sources
Peer-reviewed research
Belanche, D., Flavián, C., and Pérez-Rueda, A. (2017). Understanding interactive online advertising: Congruence and product involvement in highly and lowly arousing, skippable video ads. Journal of Interactive Marketing, 37, 75 to 88.
Boerman, S. C., Willemsen, L. M., and Van Der Aa, E. P. (2017). "This post is sponsored": Effects of sponsorship disclosure on persuasion knowledge and electronic word of mouth in the context of Facebook. Journal of Interactive Marketing, 38, 82 to 92.
Calder, B. J., and Sternthal, B. (1980). Television commercial wearout: An information processing view. Journal of Marketing Research, 17(2), 173 to 186.
Elsen, M., Pieters, R., and Wedel, M. (2016). Thin slice impressions: How advertising evaluation depends on exposure duration. Journal of Marketing Research, 53(4), 563 to 579.
Hautz, J., Füller, J., Hutter, K., and Thürridl, C. (2014). Let users generate your video ads? The impact of video source and quality on consumers' perceptions and intended behaviors. Journal of Interactive Marketing, 28(1), 1 to 15.
Johnson, G. A., Lewis, R. A., and Nubbemeyer, E. I. (2017). Ghost ads: Improving the economics of measuring online ad effectiveness. Journal of Marketing Research, 54(6), 867 to 884.
Lehnert, K., Till, B. D., and Carlson, B. D. (2013). Advertising creativity and repetition: Recall, wearout and wearin effects. International Journal of Advertising, 32(2), 211 to 231.
Lou, C., and Yuan, S. (2019). Influencer marketing: How message value and credibility affect consumer trust of branded content on social media. Journal of Interactive Advertising, 19(1), 58 to 73.
Pieters, R., and Wedel, M. (2004). Attention capture and transfer in advertising: Brand, pictorial, and text-size effects. Journal of Marketing, 68(2), 36 to 50.
Teixeira, T., Picard, R., and el Kaliouby, R. (2014). Why, when, and how much to entertain consumers in advertisements? A web-based facial tracking field study. Marketing Science, 33(6), 809 to 827.
Teixeira, T., Wedel, M., and Pieters, R. (2012). Emotion-induced engagement in internet video advertisements. Journal of Marketing Research, 49(2), 144 to 159.
Government, platform and industry sources (not peer reviewed)
TikTok Ads Help Center. Creative best practices for performance ads. Last updated June 2025; read 17 November 2026. Vendor source.
Federal Trade Commission. FTC's Endorsement Guides: What People Are Asking. Read 17 November 2026. Government source.
Free workbook
The Kill Sheet
Know when to kill an ad and when to let it run, with thresholds for the five signals that matter.
Five signals, three verdicts, ten minutes every Monday. Free.







