---
title: Reading creative metrics — spend, thumbstop/hook rate, hold rate, CTR, conversion
slug: creative-metrics-hook-hold-ctr
summary: The hook analytics funnel (spend → thumbstop → hold → CTR → conversion), exact thumbstop formula, primary vs "storytelling" metrics, diagnostic patterns (high hook + low CTR = wrong audience), spend thresholds, new-visitor % as an incrementality proxy, and Dara Denney's four-step UGC analysis (metrics, content, comparison, feedback).
domain: marketing
tags: [ad-creative, metrics, hook-rate, creative-analytics, meta-ads]
related: [hook-anatomy, paid-social-creative-testing-framework, iterating-on-winning-ads, ugc-ad-formula-and-formats, ecommerce-ad-creative-metrics, judging-creatives-by-spend-not-roas]
sources: [yt:DqFjak694PQ, yt:yNTlVKvuX0o, yt:FcI1R-oz-EI, yt:4wM-i1cxtAM, yt:pV0tM1gSJiU, yt:aSF0GiHUZxo, yt:u4B0GGBBDXY]
updated: 2026-10-24
---

# Reading creative metrics

## TL;DR
- Read an ad top-down: **spend** (does Meta think it deserves budget?) → **thumbstop** → **hold rate** → **CTR** → **conversion**; the first break shows where it failed.[^1][^2]
- **Thumbstop / hook rate = 3-second video views ÷ impressions** (video only — statics have no 3 s).[^3]
- Evan Lee: spend ~**80% of analysis on hooks** — are you even attracting the right person?[^1]
- Spend, CPA, purchases, ROAS = **primary** (did it work); hook, hold, CTR = **storytelling** metrics (why).[^4]
- Set a minimum spend before judging (Motion uses ~$300).[^5]

## Details
**Diagnostic patterns (Dara Denney):**[^6]
- Low hook + low hold → creative doesn't resonate; back to the drawing board.
- High hook + high hold, low CTR → attracting the wrong audience, or little of the target audience is available for that creative.
- High hook rate on a winner → you reached the right demographic; make more hooks like it (visuals, VO, messaging).
- Low thumbstop but high CTR can mean people clicked within 3 s — rare (Evan).[^7]

**Benchmarks.** Dara uses Varos for niche benchmarks; Evan points to Motion's benchmark report.[^8][^7] One investigative-journalism creator ad Dara shows had a **77% hook rate** — exceptional, not a target.[^9] (App-focused target in [[paid-social-creative-testing-framework]]: 30–40%.)

**Dara's 4-step UGC analysis:**[^10][^11][^12]
1. *Metrics* — primary + storytelling.
2. *Content* — messaging, visuals, creator, setting, pacing; look for trends across tests, not 1:1 metric links.
3. *Comparison* — vs campaign average, vs top performers, vs competitors.
4. *Feedback* — ad comments, recurring objections, CS tickets, sales/CX team patterns.

**Savannah Sanchez's weekly Motion routine.**[^16][^17][^18]
- Each week she opens the *Top creatives* report for the last **14 days**, filtered to prospecting. She watches the ad inside the tool and uses *creative insights* (watch-time curve, age/gender, spend, ROAS, CPA).
- In the *Video performance* report she ranks ads by **thumbstop ratio**, then **CTR**, then **click-to-purchase** (purchasers ÷ clickers).
- Low thumbstop + high click-to-purchase → the offer/body works, so keep format and CTA and swap the hook.
- High thumbstop + low purchases → try a stronger CTA, different talent, or check that ad's landing page against the page used by better-converting ads.

**Segment before you judge.** Results in Meta's audience-segments breakdown (new vs engaged vs existing customers) can flip which ad looks best. Barry Hott ranks creatives by results with the *new* audience ([[judging-creatives-by-spend-not-roas]]).[^19]

**Clicks aren't customers.** Sarah Levinger: CTR "lies sometimes"; compare with post-purchase surveys/quizzes to see which creative drove purchases.[^13]

**Incrementality proxy (Tushy).** Without a holdout test, compare an ad's **new-visitor %** (Triple Whale/Northbeam) with the account average — e.g. 85% vs 70% suggests it unlocks a new part of the auction. Formal option: inverse/geo holdout on whitelisted content.[^14][^15]

## Disagreements / caveats
- Hold-rate definitions vary (15 s ÷ impressions vs 15 s ÷ 3 s views) — see [[paid-social-creative-testing-framework]]; Dara defines it loosely as retention through the rest of the ad.[^4]
- Metric roles differ on LinkedIn (different ad types), per Evan.[^7]

## See also
- [[hook-anatomy]] · [[iterating-on-winning-ads]] · [[judging-creatives-by-spend-not-roas]]

[^1]: yt:DqFjak694PQ @ 27:55 — Evan Lee
[^2]: yt:DqFjak694PQ @ 29:29 — Evan Lee
[^3]: yt:DqFjak694PQ @ 35:42 — Evan Lee
[^4]: yt:yNTlVKvuX0o @ 16:42 — Dara Denney
[^5]: yt:DqFjak694PQ @ 34:08 — Evan Lee
[^6]: yt:yNTlVKvuX0o @ 16:42 — Dara Denney
[^7]: yt:DqFjak694PQ @ 32:34 — Evan Lee
[^8]: yt:yNTlVKvuX0o @ 18:17 — Dara Denney
[^9]: yt:4wM-i1cxtAM @ 13:37 — Dara Denney
[^10]: yt:yNTlVKvuX0o @ 15:14 — Dara Denney
[^11]: yt:yNTlVKvuX0o @ 18:17 — Dara Denney
[^12]: yt:yNTlVKvuX0o @ 19:52 — Dara Denney
[^13]: yt:pV0tM1gSJiU @ 1:08:53 — Sarah Levinger
[^14]: yt:FcI1R-oz-EI @ 56:50 — Jarelle (Tushy)
[^15]: yt:FcI1R-oz-EI @ 56:03 — Jarelle (Tushy)
[^16]: yt:aSF0GiHUZxo @ 00:00 — Savannah Sanchez
[^17]: yt:aSF0GiHUZxo @ 05:15 — Savannah Sanchez
[^18]: yt:aSF0GiHUZxo @ 06:50 — Savannah Sanchez
[^19]: yt:u4B0GGBBDXY @ 11:55 — Barry Hott