.jpeg)
- Has a video ad that is underperforming but is unsure why
- Makes decisions about scaling or pausing based on whether the ad "feels good"
- Wants to know how to use hook rate and hold rate to target creative revisions
- Is tired of reshooting entire ads from scratch when only one part is actually the problem
Many decisions about ad performance are made based on the wrong signals: "the video feels good," "it looks slick," or "ROAS was good for the first few days." None of these tell you anything about how users are actually reacting to the video.
Performance isn't about what you think of a video; it's about how users actually experience it. ROAS is an aggregate result that doesn't distinguish between whether the problem is that no one keeps watching the video, or that they watch it all the way through but never click. Without breaking the video down into its individual layers, you are essentially guessing what needs to be fixed.
Hook rate: Are you grabbing attention?
Hook rate measures the percentage of viewers who watch the video past the first few seconds. It is the first and most fundamental layer of the diagnosis: if you lose the user here, the rest of the video rarely performs well, no matter how strong the story is.
A low hook rate points directly back to the opening itself: is it surprising, pattern-interrupting, or does it grab attention immediately? Is the product shown within the first few seconds? Is there a clear, easy-to-understand message in the opening? A low hook rate means you need to test new opening sequences, visual hooks, or messages with more immediate impact, rather than changing the rest of the video.
Hold rate measures how many people watch the video beyond the first few seconds, typically broken down into several checkpoints along the way (for example, at 25%, 50%, and 75% of the video's length).
A drop in hold rate at a specific point is a direct signal of poor pacing, an unclear narrative, or a lack of relevance at that exact moment. The advantage of looking at hold rate in multiple stages is that you can pinpoint exactly where in the video the viewer drops off, rather than just stating that "people stop watching it."
The real value comes when you look at hook rate and hold rate together, not in isolation. The combination of the two tells you which part of the video actually needs work:
High hook rate, low hold rate
You have done a good job of stopping the scroll, but a poor job of retaining the viewer. Here, the work should be focused on the body of the video, the story, the value proposition, and the pacing, not on the opening, which is already working.
Low hook rate, high hold rate
Those who watch the video past the opening stay, but too few make it past the opening in the first place. This is a distribution problem: the content itself is strong, but it doesn't get the chance to deliver because the hook isn't catching enough people.
High hook rate, high hold rate
The video is working on both levels. This is a signal to scale and study which specific elements are driving performance so the pattern can be reused in new concepts.
Low hook rate, low hold rate
Both layers are lagging. This is often a sign that the video needs to be rethought from scratch, rather than adjusted in individual parts.
Knowing the diagnosis means you can revise precisely instead of recreating the entire ad from scratch.
1. Isolate which layer is the problem before changing anything
Look at hook rate and hold rate together and identify which of the four combinations the situation matches before any creative decision is made.
2. In case of a low hook rate: revise only the opening
Test new visual approaches, a more pattern-interrupting opening image, or a message that lands faster, while keeping the rest of the video unchanged.
3. If the hold rate is low: revise the body, not the hook
Check if the story delivers on what the hook promised, if the pacing is too slow or too fast, and if the message is clear throughout.
4. Test one change at a time
If you change both the opening and the body at the same time, you won't know which change actually moved the numbers. Isolate the variable you are testing.
Three signs that it's time to break a video ad down into hook rate and hold rate, instead of evaluating it on ROAS alone: an ad is underperforming, but you are unsure if the problem is the opening or the rest of the video; decisions to scale or pause are based primarily on gut feeling or the first few days of ROAS; or you often recreate entire ads from scratch when only a part of the video is actually the problem.
We never make decisions about scaling, adjusting, or dropping a video ad based on how it "feels." We use three layers of creative metrics—hook rate, hold rate, and CTR—to determine exactly which layer of the video is driving or hurting performance. This is the systematic approach to creative intelligence that we build into our Paid Social work through E-COM OS.
1. What is hook rate, and how do you measure it?
Hook rate is the percentage of viewers who watch a video past the first few seconds, typically measured by 3-second views divided by the number of impressions. It is the primary indicator of whether your opening captures attention.
2. What is hold rate, and why is it broken down into multiple checkpoints?
Hold rate measures how many people watch the video past a specific point—often checked at 25%, 50%, and 75% of the video's duration. This breakdown allows you to see exactly where viewers drop off, rather than just noting a general decline.
3. What does it mean if the hook rate is high but the hold rate is low?
It means the opening is working well, but the rest of the video fails to retain the viewer. You should focus your efforts on the story, pacing, and value proposition in the body of the video—not the opening.
4. Do you always have to recreate the entire video if it underperforms?
No. By diagnosing which layer—hook or hold—is the problem, you can often revise only the relevant part instead of recreating the entire ad from scratch.
5. Why is ROAS alone not enough to evaluate a video ad?
Because ROAS is an aggregate result that doesn't distinguish whether the problem lies in the opening, the story, or the offer. Without breaking the video down into hook rate and hold rate, you are essentially guessing what needs to be fixed.
.png)
