Read creative test results in this order: hook rate, completion rate, CTR, then cost per result. Each metric answers one question, and reading them out of order is the fastest way to kill a variation that would have won if you had fixed the right layer first.
Hook rate asks whether the first three seconds stopped the scroll. Completion rate asks whether the video held attention after the hook. CTR asks whether the creative made the click feel worth taking. Cost per result asks whether the click converted efficiently. The mistake most teams make is reading CTR first, which mixes all three upstream layers into one number that cannot tell you where the creative broke.
Read the four metrics in this order
The sequence works because each metric is downstream of the one before it. A low CTR can come from a weak hook, a weak hold or a weak CTA. If you rewrite the hook first when the real problem is completion, CTR will not move and you will kill a script that never had a fair run.
| Order | Metric | Question it answers | What it cannot tell you |
|---|---|---|---|
| 1 | Hook rate | Did the first three seconds stop the scroll? | Whether the body will hold. |
| 2 | Completion rate | Did the video hold after the hook? | Whether the offer was clear enough to click. |
| 3 | CTR | Did the creative make them click? | Whether the click will convert. |
| 4 | Cost per result | Did the click convert efficiently? | Which upstream layer broke. |
When you read in this order, you isolate the first broken layer. The fix can then target that layer instead of disturbing the layers that already work.
High hook rate, low completion
High hook rate and low completion is a retention problem, not a hook problem. The first three seconds stopped the scroll and the next ten lost the viewer. This is one of the most common false negatives I see in creative tests.
On Meta Reels, the published composite median hook rate is 28 percent. If a variation is at or above that and completion is below your account baseline, the audience is interested but the video is not paying off the promise. Keep the opening shot or line, then change the body. Cut the setup, move the product reveal earlier, or replace the middle proof beat.
Do not kill this variation for low CTR before the body is fixed. If viewers leave before the CTA, the click never got a fair chance.
Good completion, low CTR
Good completion and low CTR means the video holds attention but never makes the click feel worth taking. This is a click layer problem, not a retention layer problem. It is rarer, but it isolates the fix cleanly.
Keep the script, shots and pacing. Test the offer framing, the on-screen CTA or the product demonstration. If the video withholds a clear reason to click until the final frame, CTR is already capped. Move the CTA earlier or add a visible text prompt through the middle third.
This is also the point to separate a click problem from a conversion problem. CTR tells you only whether they clicked. Cost per result tells you whether the click was worth buying.
When frequency invalidates the comparison
There is one case where the numbers are not measuring the creative. If one variation ran at high frequency and another ran at low frequency, the comparison is invalid before you read anything else.
On Meta prospecting, decline starts around a weekly frequency of 2.5. Meta internal research shows CTR drops 45 percent after a fourth exposure to the same creative. Our 2026 fatigue benchmark puts the week three drop at 45 to 70 percent below the launch baseline.
So if one ad ran at 2.8 weekly frequency and the other at 1.2, the higher-frequency ad was already decaying. Pause or duplicate the ad set to reset frequency, or compare first-exposure metrics where the platform exposes them. Do not rewrite the creative because a tired audience stopped responding.
Use the order before you kill a variation
Before killing a variation, write down the four numbers and find the first layer that failed. If hook rate failed, kill or iterate only the opening. If completion failed, keep the hook and edit the body. If CTR failed, keep the video and change the CTA or offer. If cost per result failed, check the audience or the landing page before blaming the creative.
If frequency is above about 3 on TikTok or 2.5 on Meta, reset the comparison first. The fatigue benchmark also reports brands shipping 15 to 50 creative variants a month see 3 to 5 times longer campaign lifespan than quarterly refreshers. Most active campaigns need 8 to 20 live variations in rotation.
That throughput is not just an editing problem. Genyad builds each variation from a source library, so a new version is a fresh script, shot selection, voiceover, caption set and export rather than a re-cut of the same timeline. One standard variation costs 1 credit, credits never expire, and the free plan includes five variations with no card. Genyad does not have AI avatars, predicted performance scores or direct publishing to Meta or TikTok. If an AI avatar is the actual creative idea, Creatify or Arcads is the better tool.
Frequently asked questions
What order should I read creative test metrics in?
Read hook rate first, then completion rate, then CTR, then cost per result. Hook rate shows whether the first three seconds stopped the scroll, completion shows whether the video held, CTR shows whether the creative made them click, and cost per result shows whether the click converted efficiently. Reading CTR first hides which upstream layer broke.
Why does high hook rate and low completion happen?
It means the opening promised something the body did not deliver. The first three seconds worked, then the video lost the viewer before the CTA. Keep the hook and change the body, the middle proof beat or the pacing.
Should I kill a creative with good completion but low CTR?
No. Good completion means the video holds attention. Low CTR means the click layer is weak, so change the offer framing, the on-screen CTA or the product demonstration. Rewriting the hook in that situation risks breaking a layer that already works.
When does frequency make a creative test invalid?
Frequency invalidates the comparison when one variation has been seen far more often by the same audience. On Meta prospecting, decline starts around a weekly frequency of 2.5, and CTR drops 45 percent after a fourth exposure. Before comparing results, check the ad set frequency and reset or pause if one ad is far above the other.