Creative controls your TikTok hook rate, completion rate and click-through rate. It does not control your conversion rate, which is set almost entirely by your offer, your price and your landing page. If cost per acquisition is bad and click-through is fine, testing another twelve hooks will burn a week and change nothing, because the problem is downstream of the video.
We write that flatly because the mistake is expensive and extremely common. Creative is the fun part of the account and the part a tool can produce quickly, so it absorbs effort that belongs somewhere else.
Which metrics does TikTok creative actually control?
Split the funnel at the click. Everything above it is a video problem, everything below it is a business problem, and the two need different people looking at them.
| Stage | Metric | Owned by | Typical TikTok reference point |
|---|---|---|---|
| Impression to 3s view | Hook rate | Creative | About 33 percent median |
| 3s view to end | Completion rate | Creative | Pacing dependent, read per cut |
| View to click | Video CTR | Creative plus offer framing | About 0.84 percent median |
| Click to landing page view | Page load, tracking | Site and tech | Should be above 90 percent of clicks |
| Landing page view to add to cart | Page, price, proof | Offer and page | Vertical dependent |
| Add to cart to purchase | Checkout, shipping, payment | Checkout | Vertical dependent |
Those medians come from the platform figures in our 2026 fatigue benchmark, a synthesis of published platform and agency sources rather than our own measurement. Note what is not in the table: a median conversion rate. We deliberately do not quote one, because a number that averages a nine pound impulse product and a four hundred pound mattress across paid social is not a benchmark, it is noise. Your own last-90-day conversion rate from any channel is a far better baseline than anything published.
The one way creative does touch conversion rate
Creative changes who clicks, and that changes conversion rate without touching anything on the page. A curiosity hook that withholds what is being sold will lift CTR and dilute intent, so more clicks arrive from people who did not know what they were about to see. A hook that names the product and the price does the opposite: fewer clicks, higher conversion rate, often a better CPA.
So creative sets traffic composition, not page performance. Which means the honest version of the rule is: creative moves the quantity and qualification of clicks, the offer and page move what happens to them. If you change a hook and conversion rate moves, you have changed the audience, not improved the funnel.
How do I tell where the drop is?
Read four numbers in order and stop at the first one that is out of line. Do this before you brief anything.
- Hook rate. Compare against roughly 33 percent as the TikTok median. Below about 20 percent, the first second is the problem and nothing else matters yet.
- Completion rate against hook rate. If hook rate is healthy and completion collapses, you have a pacing sag, usually between second three and second seven. This is a mid-cut problem, not a hook problem.
- CTR against completion. People watched and did not click. That is an offer problem stated in the video: the ask is unclear, the price is absent, or the value did not land. Fix the script's last three seconds and the CTA copy before you reshoot anything.
- Click to landing page view. If you lose more than about 10 percent of clicks here, you have a technical problem, not a marketing one. Slow mobile page, redirect chain, broken pixel, consent banner eating the event.
Only after those four does the page itself come into scope. And at that point the useful test is not creative at all: it is price presentation, proof above the fold, shipping cost visibility, and the number of steps in checkout.
| Symptom | Almost certainly | Test this | Do not test this |
|---|---|---|---|
| Low hook rate, everything else fine | First second | 5 to 8 new openings, same body | Landing page |
| Good hook, poor completion | Middle of the cut | Shorter cut, earlier product reveal | New hooks |
| Good completion, low CTR | The ask | CTA wording, price on screen, last 3 seconds | Hooks, page layout |
| Good CTR, poor conversion | Offer or page | Price framing, proof, checkout steps | Anything about the video |
| Everything decent, CPA still rising | Fatigue | Fresh variations, check frequency | Nothing, this is a volume problem |
That last row is worth separating out. A rising CPA with stable per-impression metrics is usually frequency, not quality. Our benchmark report puts CTR decline at 15 to 20 percent in a creative's first two weeks, with week three at 45 to 70 percent below launch baseline, and TikTok's frequency ceiling at about 3.0. If frequency has crossed 3 and CPA is drifting, the fix is more variations in rotation, not better ones.
What to test on each side of the click
Above the click, keep the unit small. Test one variable, hold the rest, and run enough variations that a winner is visible: 8 to 20 live per active campaign is what we plan for on TikTok, refreshed weekly. Hooks are the cheapest and highest-leverage variable, product reveal timing is second, length is third, voiceover and caption style are a distant fourth.
Below the click, the tests are slower and worth more. Price framing beats page design most of the time. Adding the shipping cost to the product page beats optimising the hero image. Cutting one step from checkout beats a new headline. None of these are creative work, and none of them can be fixed by a video tool, ours included.
The sequencing rule we use: never run a creative test and a landing page test in the same week on the same campaign. You will get one clean answer for two weeks of spend, or two dirty answers for one.
Where a creative tool helps, and where it does not
Genyad is our product, so read this with that in mind. It turns footage you already own into publishable variations: you upload once, it transcribes and tags the clips, and each variation is a fresh script, shot selection, voiceover and caption set rather than a re-cut of one timeline. That makes the top half of the funnel cheap to iterate on, which matters when the diagnosis lands on hook rate or completion. The TikTok ad maker page covers how that works, and it is credit-based with one credit per variation, so a diagnostic round of eight hook variants is a small spend.
What it cannot do is anything below the click. It does not touch your landing page, your price, your checkout or your shipping table. It also has no predicted performance score, no AI avatars, no static banner formats and no direct publishing to TikTok, so you export the files and set up the campaign in Ads Manager. If your diagnosis says the drop is between add to cart and purchase, buying more creative is the wrong purchase, and we would rather say so than sell you credits.
Frequently asked questions
What is a good conversion rate for TikTok ads?
There is no useful cross-vertical median, and we would treat any single published figure with suspicion given how much price point and product category move it. Use your own conversion rate from other paid channels over the last 90 days as the baseline and expect TikTok traffic to convert somewhat lower at the same price, because intent is lower than search. What matters is whether CPA clears your target, not whether CVR clears someone else's average.
Why is my TikTok CTR fine but conversion rate terrible?
Usually because the hook withheld what you sell. Curiosity-led openings pull clicks from people who did not know what they were about to be sold, which lifts CTR and dilutes intent. Try a version that names the product and the price in the first three seconds and compare CPA, not CTR.
Should I fix creative or the landing page first?
Diagnose first, then fix one at a time. If hook rate is under about 20 percent against a 33 percent TikTok median, start with creative because nothing downstream gets a fair test until traffic exists. If click-through is healthy and conversion is not, the page and the offer are where the money is.
Does TikTok's algorithm penalise low conversion rate?
Delivery responds to the events you optimise for, so a campaign optimising for purchases with few purchases will find it hard to exit the learning phase and costs tend to drift up. That is a data volume problem rather than a penalty. If purchase volume is thin, optimising for a mid-funnel event and watching CPA manually is often more stable.