Pro Strategy Summary
Scale a video ad when ROAS exceeds your target threshold, CTR holds above 1.5%, and cost per result has trended down across three or more consecutive days. Kill it when ROAS drops below break-even, ad frequency surpasses 3 on the same audience segment, and hook rate falls below 25%. No single metric tells the full story. Combine spend thresholds with creative performance signals to build a repeatable decision framework. Most e-commerce founders wait too long to cut losers and too long to scale winners. This data guide gives you the exact benchmarks to act with confidence on both.
Knowing when to scale vs. kill a video ad is one of the highest-leverage skills in paid media. You can have a solid product, a healthy budget, and a capable team, but poor creative decisions will drain your ad account faster than any algorithm change. This guide breaks down the data signals that separate a winner from a waste of spend.
What Does “Scale vs. Kill” Mean for Video Ads?
Scaling means increasing budget, broadening the audience, or duplicating a campaign to extract more value from a proven creative. Killing means pausing or stopping an ad that is no longer delivering results worth the spend. Both decisions are time-sensitive. A winning ad left unscaled loses momentum. A losing ad left running burns budget that could fund a better test.
The challenge for most e-commerce founders is that ad platforms are designed to encourage spending, not disciplined culling. Understanding the difference between a learning-phase dip and a genuine creative failure is what separates profitable scaling from guesswork.
Key Metrics That Drive the Scale vs. Kill Decision
Before setting thresholds, you need to know which numbers actually matter. Not every metric in Ads Manager deserves equal weight.
ROAS and Cost Per Result
Return on ad spend (ROAS) is the headline metric, but cost per result (CPR) often tells the more actionable story. A declining CPR alongside stable or rising ROAS is a strong scale signal. A CPR climbing faster than revenue is a kill signal. Set your break-even ROAS before you launch, not after results come in. Refer to our breakdown of what is a good ROAS for social media video ads to calibrate your baseline.
Hook Rate and Hold Rate
Hook rate measures how many people watch the first three seconds of your video ad compared to total impressions. A hook rate above 30% is strong. Below 25% signals a creative problem at the very start of the ad. Hold rate measures retention through to 50% or 75% of the video. Low hold rates confirm that even those captured by the hook are dropping off before the CTA. Both metrics point directly at the creative, not the targeting.
Frequency and CPM
Frequency above 3 within a 7-day window typically signals audience saturation. When CPM starts rising alongside falling CTR at the same frequency level, the audience has seen your ad enough times. Rising CPMs with no lift in conversion rate mean you are paying more for less attention. Our industry data on video ad CPM benchmarks by industry can help you spot when your costs are abnormally high for your vertical.
When to Scale a Video Ad
A video ad is ready to scale when it clears all three of these gates consistently:
- ROAS is at or above your target for at least three consecutive days with sufficient spend (minimum 2x your average order value per day)
- Hook rate is above 30% and hold rate is above 40% at the 50% watch point
- Frequency is below 2.5 within a 7-day window, indicating the audience is not saturated yet
When an ad meets these criteria, scale the budget no more than 20% every 48 to 72 hours on Meta to avoid triggering a new learning phase. On Google and YouTube, scaling can be more aggressive because the algorithm is less sensitive to budget changes.
When to Kill a Video Ad
Kill a video ad when it meets any two of these conditions after clearing the learning phase (typically 50 conversions or 7 days of spend, whichever comes first):
- ROAS falls below break-even for three or more consecutive days
- Hook rate drops below 25%
- Frequency exceeds 3.5 with no corresponding drop in CPM
- CTR falls below 0.8% on a cold audience
A common mistake founders make is giving underperforming ads too long a runway, hoping the algorithm will “figure it out.” After the learning phase, the platform has already shown the ad to the most receptive portion of your audience. If it is not working at that point, more budget rarely fixes the underlying creative problem.
A Repeatable Framework for Scale vs. Kill Decisions
Build a weekly creative review cadence. Every Monday, pull the last 7 days of data and score each active ad against your benchmarks. Assign each ad a status: Scale, Hold, or Kill. “Hold” means the ad is in the learning phase or has borderline metrics that need one more week of data before a decision. This prevents both premature killing and budget waste.
Document every kill decision with a reason code: poor hook, audience fatigue, low hold rate, or below break-even ROAS. Over time, this log becomes a creative brief generator. Patterns in what fails directly inform what to test next.
Expert Tips for Smarter Scale vs. Kill Decisions
When we analyze hook rates for clients, the single most common issue is a weak opening three seconds. The visual does not stop the scroll, or the spoken hook arrives too late in the frame. Fixing the first three seconds of an existing ad often produces a winning variant without rebuilding the entire creative.
A common mistake that kills ROAS is scaling too fast. Founders double or triple budgets overnight, the algorithm resets into a new learning phase, and performance tanks. They then kill the ad, assuming it failed, when the real culprit was the scaling method.
The secret to a high-converting creative rotation is having two to three ready-to-launch variants for every ad you are scaling. The moment frequency climbs past 2.5, introduce a new variant into the ad set so performance carries forward while the creative refreshes.
Build Video Ads Worth Scaling with videoadstop.com
The scale vs. kill decision only matters when you have creatives worth scaling in the first place. videoadstop.com is a leader in professional video ad production built for performance marketers. We specialize in high-impact visuals and data-backed storytelling designed to stop the scroll and drive conversions across Meta, TikTok, and YouTube. Our team understands hook mechanics, retention design, and the creative variables that move your metrics, so you spend less time killing ads and more time scaling winners.
Frequently Asked Questions
How much spend should I give a video ad before deciding to scale or kill?
A standard rule is to spend at least 2x your average order value per day, or wait until the ad has generated 50 optimization events, whichever comes first. This gives the platform enough data to exit the learning phase before you draw conclusions.
Can a video ad recover after a performance dip?
Yes, but only in specific scenarios. If the dip is tied to external factors like a seasonal slowdown or a budget change that triggered a new learning phase, recovery is possible. If hook rate and hold rate have also declined, the creative itself is the problem and recovery without changes to the video is unlikely.
What is the right frequency cap before killing a video ad?
On Meta, most e-commerce brands see diminishing returns after a frequency of 3 within a 7-day window. On YouTube, a frequency of 4 to 5 per week is generally acceptable for in-stream ads. Once frequency climbs past those thresholds alongside rising CPM and falling CTR, the creative needs to be rotated or retired.
Should I kill an ad that has great engagement but low ROAS?
High engagement with low ROAS usually points to a disconnect between the ad message and the landing page, or a targeting mismatch. Before killing the ad, test the same creative against a different audience segment or review the post-click experience. If both tests fail, then kill it.