Metrics to Track for Video Ad Performance: A Practical Guide

If you're spending money on video ads and only looking at cost per purchase, you're flying with one gauge on the dashboard. The metrics to track for video ad performance span three stages — did people stop scrolling, did they keep watching, and did they act — and each stage tells you something the others can't. Get this wrong and you'll kill a creative that was actually working, or keep funding one that was never going to convert.

The metrics to track for video ad performance, grouped by funnel stage

Every video ad has to do three jobs in sequence: stop the scroll, hold attention, and move someone to act. Tracking metrics by stage makes it obvious which job your ad is failing at, instead of just knowing that it "isn't working."

Attention metrics: did it stop the scroll?

• Hook rate (or thumb-stop rate) — the share of people who watch past the first few seconds instead of scrolling immediately. This is your clearest signal on whether the opening frame and first line of copy or voiceover are doing their job.
• 3-second view rate — a close cousin of hook rate, available on most ad platforms by default. Use it the same way: a low number almost always points at the first 1-3 seconds, not the rest of the ad.
• Impression-to-view ratio — how many impressions it takes to generate a view. A widening gap over time can mean creative fatigue, even if other numbers still look fine.

Retention metrics: did it hold attention?

• Average watch time — how long, on average, people stay on your video. Useful as a trend line across creative versions more than as a single absolute number.
• Video completion rate — the share of viewers who reach the end. For a 30-second ad, this tells you whether your offer and call to action are even being seen by most of your audience.
• Retention by quartile (25%, 50%, 75%, 100%) — the most diagnostic metric on this list. A steep drop at a specific point usually maps to a specific moment in the ad: a slow problem statement, a proof section that drags, or a confusing transition.

Action metrics: did it move someone?

• Click-through rate (CTR) — the share of viewers who click through to your site. Pair this with retention: strong retention and weak CTR usually means the call to action or offer is the problem, not the hook.
• Landing page view rate — how many clicks actually resolve into a loaded page. A gap here points at page speed or tracking issues, not the ad.
• Cost per result (CPA) — cost per purchase, lead, install, or whatever your conversion event is. This is your bottom-line efficiency number, but it's the least useful metric for diagnosing why an ad is underperforming.
• Return on ad spend (ROAS) — revenue generated per unit spent. Treat it as the scoreboard, and treat hook, retention and CTR as the stats that explain the score.

How to set up tracking before you spend budget

Metrics are only useful if you can trust them and compare them fairly across creatives. Before you launch a round of video ad tests, lock down the basics:

1. Confirm your pixel or tag is firing correctly on the specific conversion event you care about, not just page views.
2. Decide your primary metric per objective ahead of time — hook rate for a brand-awareness test, CTR for a traffic test, CPA or ROAS for a conversion test. Don't judge an awareness test by purchase volume.
3. Give every ad a consistent naming convention (hook version, angle, date) so you can actually compare rows in the ads manager later instead of guessing which creative is which.
4. Set a minimum spend or impression threshold before you look at results. Early numbers on small samples swing wildly and will mislead you.
5. Isolate one variable per test where possible — one new hook, one new call to action — so a change in metrics has an obvious cause.

Benchmarks to treat as a starting point, not a rule

Every account, audience, and offer is different, so hard benchmarks for video ad performance don't really exist. What does help is a rough sense of what each metric is for, so you know where to look first when a number disappoints you.

Reading metrics together to diagnose the real problem

The most common mistake in creative testing is treating metrics in isolation. A single number rarely tells you what to fix; a combination usually does. Use these as decision rules when you're triaging an underperforming ad.

• High hook rate, low completion rate — the opening works, but something in the middle loses people. Check pacing and whether the proof or benefit section is too slow or repetitive.
• Low hook rate across every version of an ad — the problem is likely the product angle or targeting, not the edit. No hook rewrite will fix the wrong audience.
• High completion rate, low CTR — people watch the whole thing but don't act. The call to action is probably too soft, too late, or missing from the on-screen text (remember that many viewers watch on mute, so the ask needs to be visible, not just spoken).
• Good CTR, weak CPA or ROAS — the ad is doing its job; the landing page, offer, or pricing is likely the bottleneck.
• Metrics decline over time on a previously strong ad — this is usually creative fatigue rather than a sudden audience shift. Plan to refresh the hook or visuals before performance drops too far.

Building a creative testing cadence around your metrics

Metrics only compound in value if you review them on a schedule and act on what they tell you. A simple cadence that works for most small and mid-size advertisers:

1. Weekly: scan hook rate and retention by quartile across all active ads. Flag anything with a hook rate noticeably below your account average.
2. Every two weeks: review CTR and CPA/ROAS trends per creative. Pause or rework ads that are consistently below your other variants once they've hit a reasonable spend threshold.
3. Monthly: look at the pattern across your best-performing ads. Are strong hooks sharing a format (question, bold claim, before/after)? Are weak ads sharing a flaw (slow pacing, unclear offer)? Feed that pattern into your next batch of creative.

Because retention problems are usually pacing or structure problems, it helps to test several hook and structure variants quickly rather than guessing with one ad at a time. Tools like FrameNotion can generate A/B hook variants of a finished ad without re-editing from scratch, which makes this weekly cadence realistic even for a small team producing its own creative.

Common mistakes when tracking video ad performance

• Judging an ad before it has enough spend or impressions — early numbers are noisy; wait for a reasonable sample before deciding.
• Comparing metrics across platforms as if they mean the same thing — a completion rate on one placement isn't directly comparable to another placement with different average view lengths.
• Optimizing for the metric that's easiest to see instead of the one that matters — views and impressions are visible everywhere but rarely tell you whether the ad is profitable.
• Blaming the ad when the problem is the landing page — a strong CTR with a weak CPA points away from the creative, not toward it.
• Changing multiple variables at once — swapping the hook, the voiceover, and the offer in the same test makes it impossible to know which change moved the metrics.

Where FrameNotion fits into creative testing

FrameNotion doesn't track ad performance or publish to ad platforms — you'll still read your metrics in your ads manager. What it does is remove the production bottleneck that keeps most teams from testing enough variants to act on those metrics. Paste a product link and FrameNotion AI writes and renders a 30-second vertical ad from scratch in about 10-20 minutes, complete with voiceover, captions, and a soundtrack cut to the beat, delivered in the aspect ratios you need for TikTok, Reels, Shorts and Meta ads. Once an ad is live and you've identified a weak hook from your retention data, you can request hook variants or edit on-screen text and colors without starting the edit over. If your testing cadence keeps surfacing the same bottleneck — not enough fresh creative to react to what the metrics are telling you — that's the gap worth closing first.

Putting it into practice

Start simple: pick one primary metric per test objective, track hook rate and retention by quartile on everything, and only look at CPA or ROAS once you've confirmed the ad is actually being watched. From there, use the decision rules above to figure out whether a disappointing ad needs a new hook, a tighter middle, a clearer call to action, or a better landing page — and keep producing enough fresh variants that you always have something ready to test next. For more on the creative side of that process, see how sound-off viewers change what your hook needs to do, and how safe zones for on-screen text affect whether your call to action is even visible on every placement.