Kill an ad when a specific metric has enough data to be trusted and it misses the threshold by enough that recovery is unlikely. A fixed day count is less useful than a decision table built on impressions, clicks, hook rate and frequency. In most Meta accounts that means judging hook rate at around 2,500 impressions and CTR at around 5,000 impressions, then checking frequency as an override.
Kill rules by metric and data volume
The reason most ad accounts keep bad creative too long is not loyalty, it is noise. A 0.4 percent CTR on 900 impressions predicts almost nothing. A 0.4 percent CTR on 8,000 impressions is a signal. The table below is a starting point for Meta feed and Reels. TikTok needs a lower CTR bar because the median video CTR there is 0.84 percent, not Meta's 1.62 to 1.78 percent.
| Signal | Minimum data volume | Kill now | Hold and re-read |
|---|---|---|---|
| Hook rate | 2,500 impressions, 48+ hours | Below 15% on Meta feed/Reels or below 20% on TikTok | Above the kill line but below account median |
| CTR | 5,000 impressions, 3+ days | Below 0.8% on Meta with weak hook rate | Below 0.8% with completion rate above account median |
| Completion rate | 3,000 impressions on a 30s+ ad | Below account median and CTR weak | Above account median while CTR lags |
| Frequency | Not volume dependent | Weekly frequency above 2.5 Meta prospecting or above 3.0 TikTok | Frequency low, creative still learning |
The volume minimums matter more than the calendar. A slow campaign can take five days to reach 5,000 impressions. A fast one gets there in six hours. Both should be read the same way.
The 48 hour hook rate cut
At 48 hours, make the first kill decision on hook rate only. Hook rate is the share of impressions that watch at least three seconds. It is the earliest reliable interest signal and it does not need conversion data.
On Meta feed or Reels, cut an ad if it has at least 2,500 impressions and a hook rate below 15 percent. On TikTok, cut below 20 percent with at least 3,000 impressions. Do not wait for CPA or ROAS to catch up. A hook that low will not recover, and every extra euro spent is buying more proof of the same answer.
Why 48 hours? It is enough for most broad targeting to clear the minimum impression threshold without being so late that you have already paid for the week three CTR cliff. If the hook rate is between the kill line and your account median, hold. If it is above your account median, hold even if CTR is still weak.
Why a low-CTR ad with strong completion deserves a second week
Low CTR usually means the ad is not stopping thumbs. High completion means it is holding the people who do start watching. When those two signals conflict, trust completion rate for a second week before you trust CTR for the kill.
Suppose a Meta feed ad has a CTR of 0.95 percent, below the 1.62 percent median in our 2026 fatigue benchmark, but its completion rate is above the account median. That ad is not necessarily broken. It may be attracting cheap or broad impressions, or the platform may still be exploring placements. Kill it now and you lose an ad that could work once the algorithm finds the audience. Hold it into the second week if completion remains strong.
This pattern is even more common on YouTube in-stream, where the median CTR is 0.42 percent. A 0.4 percent in-stream CTR with above-median completion is often a placement reality, not a creative failure.
Frequency is an override for every kill rule
Frequency overrides the metric rules in both directions. Use frequency as the final check after CTR, hook rate and completion.
Our fatigue benchmark puts the Meta prospecting decline at a weekly frequency of 2.5, with CTR dropping 45 percent after a fourth exposure to the same creative. That means an ad can still look healthy on CTR while frequency is already high. Kill or pause it at 2.5 weekly on Meta prospecting and around 3.0 on TikTok, even if the metrics are fine. The alternative is to watch it fall off the week three cliff.
Frequency also explains weak metrics. If frequency is above 3 and CTR is falling, the creative may be tired rather than bad. Kill the current placement, not the concept. Ship a variation from the same winning script with a new hook or cut.
Build kill decisions into creative throughput
Fast kill decisions only work if you have a next variation ready. The benchmark concludes that throughput, not talent, is the bottleneck. Brands shipping 15 to 50 creative variants a month see 3 to 5 times longer campaign lifespan than quarterly refreshers. A typical active campaign needs 8 to 20 live variations in rotation.
That is what Genyad is for. You upload footage once, the system transcribes and tags every clip, and each variation is a fresh script, shot selection, voiceover, caption set and export. It is not a re-cut of the same timeline, so you can kill a weak ad at 48 hours and brief a replacement from the same library without a shoot or editor.
One standard variation costs 1 credit. The free plan gives five variations plus one AI-generated video with no card. Starter is €29 for 15 credits, Growth is €99 for 65 credits. Genyad does not give predicted performance scores and does not publish directly to Meta or TikTok. You still make the kill calls from ad account data, which is why the table above exists.
For the testing cadence after a kill, see our creative testing workflow.
Frequently asked questions
How much data do I need before killing an ad?
For Meta feed or Reels, wait for about 2,500 impressions before judging hook rate and about 5,000 impressions before judging CTR. On TikTok, use 3,000 impressions for hook rate and a lower CTR benchmark, since the median video CTR is 0.84 percent. With less than this, a kill decision is mostly noise.
Is a low CTR with high completion a kill?
Not automatically. If completion rate is above the account median, give the ad a second week. Low CTR means weak thumbstop or broad placements. High completion means the message holds the right viewers once they start. Kill only if both CTR and completion are weak.
When should frequency override good metrics?
On Meta prospecting, kill or pause a creative when weekly frequency passes 2.5. On TikTok, the ceiling is about 3.0. Frequency can be high even while CTR still looks healthy, and the decline begins at that point. Do not wait for the week three cliff.
What is the difference between pausing and killing?
Pausing preserves the ad for later testing or a different placement. Killing means you stop spending and replace it with a new variation. Frequency fatigue often calls for pausing or rotating the current placement, while a 48 hour hook rate failure is usually a permanent kill.