Reading ad metrics: CTR, hook rate, and hold rate
CTR, hook rate, and hold rate each diagnose a different part of your ad. Learn what each means, what good looks like, and how to fix creative from the numbers.
CTR, hook rate, and hold rate each measure a different part of the same ad: hook rate is the share of impressions that stop the scroll in the first ~3 seconds, hold rate is the share who keep watching to a key point (often 15 seconds or completion), and CTR is the share who click through to your store. Read together, they pinpoint exactly which part of the creative to fix — the opening, the middle, or the offer.
The three metrics, defined
Most ad managers surface a dozen numbers. These three tell you the story of how a viewer experiences your ad, in order.
- Hook rate — usually 3-second video plays divided by impressions. It answers one question: did the first frame and first line stop the thumb? This is your scroll-stopper score.
- Hold rate — viewers who reach a later checkpoint (commonly 15 seconds, or completion / average watch time as a percentage). It answers: once they stopped, did they stay?
- CTR — clicks divided by impressions (link CTR, not all clicks). It answers: did watching turn into wanting? This is where entertainment becomes intent.
Definitions vary slightly by platform, so baseline against your own account rather than a blog's number. The pattern across the three matters more than any single value.
Hook rate: your first three seconds are the whole game
On a feed, you are not competing with other ads. You are competing with the next swipe. A weak hook rate means the ad never got a fair shot — everything downstream is measuring an audience that already left.
When hook rate is low, the problem is almost always visual or the first spoken line. Things that move the number:
- Open on motion or a face, not a logo or a slow product pan.
- Lead with the problem, a bold claim, or a pattern interrupt in the first line — not a brand intro.
- Put a punchy on-screen caption in the first frame so sound-off viewers get the hook too.
- Test 5-10 different openings against the same body. The hook is the single highest-leverage variable in the whole ad.
Hold rate: the middle is where ads quietly die
A strong hook with a collapsing hold rate is the most common failure we see. It means the opening wrote a check the rest of the ad didn't cash. The viewer leaned in, got bored, and left.
Read the retention curve, not just the average. A sharp cliff at a specific second tells you exactly where to cut. Fixes that raise hold rate:
- Pay off the hook fast — if you promised a result, show it before people leave to find it elsewhere.
- Cut dead air and re-cut for pace: new shot or new idea every 2-3 seconds.
- Keep captions on the whole way through; word-by-word text keeps sound-off viewers reading.
- Restack the story so the most interesting beat isn't buried at the end.
CTR: did the ad actually sell?
Good hook, good hold, low CTR is a specific and useful signal: the ad was entertaining but not persuasive. People watched and felt nothing they needed to act on. This is a messaging and offer problem, not a production one.
- Sharpen the offer and make it concrete — the reason to click, and the reason to click now.
- Add proof: results, reviews, before/after, the product visibly in use.
- Make the CTA explicit and land it while attention is still high, not only in the last second.
- Check message-to-landing-page match — a click that fights a mismatched page depresses CTR over time as the platform learns.
Read them together: a diagnostic table
No single metric diagnoses a creative. The combination does. Use this as a quick lookup when you open a report.
| Hook | Hold | CTR | Diagnosis | What to change |
|---|---|---|---|---|
| Low | — | — | Scroll-stopper is weak; nothing else is being measured fairly | Test new openings, first frames, and first lines |
| High | Low | — | Opening over-promises; the middle sags | Tighten pace, pay off the hook, cut dead air |
| High | High | Low | Entertained but didn't sell | Sharpen offer and CTA, add proof, show product in use |
| High | High | High | Winner | Scale spend and spin off variations before fatigue |
| Low | Low | Low | Wrong audience or wrong angle entirely | Rethink the angle and who you're targeting |
One caution: judge on enough volume. A hook rate off 400 impressions is noise. Give each variant a few thousand impressions before you trust the read, and change one variable at a time so you know what moved the number.
Turn the diagnosis into action fast
The playbook only works if you can act on it. If low hook rate means "test ten openings," you need ten openings by this afternoon, not next sprint. The teams that win at paid social are not smarter readers of metrics — they close the loop faster. Diagnose, produce variants, re-test, kill losers, scale winners, repeat weekly.
Why teams pick Polaris
Polaris is an AI ad studio built only for ecommerce, and it's designed for exactly this loop: read the metric, produce the fix, ship the next test. Paste a product image or link and generate UGC-style video and image ads in minutes.
- Volume of variations, cheaply. Generate many hook openings and full video ads using multiple top models chosen per shot (Google Veo, Kling, Seedance, Nano Banana), so testing ten hooks is a normal Tuesday.
- Captions that fix hold rate. One-click Auto-Captions in TikTok-native styles (TikTok, Hormozi, Beast, Neon), transcribed word-accurate and burned into the video — the retention lever, done for you.
- Recreate winners. Paste a reference ad that's performing and get your own version to build variants around a proven structure.
- Works inside Claude. Our MCP connector lets you generate ads right in chat, next to the analysis.
- Credit-based pricing. Pro is $49/mo for 1,500 credits, and cheaper on longer terms ($42/mo on 3-month, $37/mo on 6-month).
Reading the metrics tells you what to change. Polaris lets you change it before the day is out. Create your first ad or see pricing.