Do UGC hooks matter?
The caption hook type barely does. Across 11,849 brand UGC videos from 46 creator programs (1 August 2026 to 30 October 2026), every caption hook type landed between 0.87x and 1.12x of its brand's typical views, and the 1,098 videos that hit 10x their brand's median used the same mix of hooks as all videos.
What did help, when we compared videos from the same program, creator size and length:
- A number in the first line (9.1% vs 5.9% hit 10x).
- A real question (7.5% vs 5.1%).
- All-lowercase writing (7.5% vs 5.6%).
- Longer openings, or no caption at all (7.7% vs 6.4% for 11+ words; 11.3% vs 7.4% with no caption text).
Captions that talk at the viewer ("you need this") did slightly worse: 5.1% vs 6.2%.
- Videos
- 11,84946 creator programs
- 10x hits
- 1,0989.3% of videos
- Hook type range
- 0.87x to 1.12xmedian views vs brand typical
- Median opening
- 6 words1 August 2026 to 30 October 2026
Hook type moved views less than length or creator size
Each caption's first line, with hashtags and mentions removed, was sorted into a hook type by the same classifier behind the hook mix on our UGC Brands pages. Every type sat within a narrow band of its brand's typical video.
- Other statement45% of videos1.00x
- Relatable reaction or POV14% of videos1.05x
- Benefit or result claim10% of videos0.98x
- Hashtags or tags only9% of videos0.98x
- Tip, hack or how-to6% of videos1.07x
- Question6% of videos0.87x
- Routine, food or workout moment5% of videos1.12x
- Story, skit or prank4% of videos1.05x
1.00x is a typical video for that brand. The same numbers are in the table below.
| Caption hook type | Share of videos | Median views vs brand typical | Hit 10x |
|---|---|---|---|
| Other statement | 45% | 1.00x | 9.2% |
| Relatable reaction or POV | 14% | 1.05x | 9.2% |
| Benefit or result claim | 10% | 0.98x | 6.2% |
| Hashtags or tags only | 9% | 0.98x | 11.1% |
| Tip, hack or how-to | 6% | 1.07x | 5.6% |
| Question | 6% | 0.87x | 11.3% |
| Routine, food or workout moment | 5% | 1.12x | 13.0% |
| Story, skit or prank | 4% | 1.05x | 11.7% |
These are raw figures. The like-for-like check further down changes several of the hit rates. For scale, video length in the same videos runs from 0.87x to 3.14x and creator size from 0.78x to 12.18x (see the UGC video benchmarks).
The 10x videos used the same hooks as everyone else
If one hook type made breakouts, it would be overrepresented among the 1,098 videos that reached 10x their brand's median. None is by more than 4 points.
| Hook type | Share of all videos | Share of 10x hits |
|---|---|---|
| Other statement | 45.2% | 44.9% |
| Relatable reaction or POV | 13.6% | 13.5% |
| Benefit or result claim | 10.2% | 6.8% |
| Hashtags or tags only | 9.3% | 11.1% |
| Tip, hack or how-to | 6.4% | 3.9% |
| Question | 6.3% | 7.6% |
| Routine, food or workout moment | 4.9% | 6.8% |
| Story, skit or prank | 4.2% | 5.3% |
Most raw hook gaps are creator and length effects
Captions are not written at random. Openings with numbers come from accounts with a median 2,710 followers, against 408 for the rest. Benefit claims and "!" endings are common on high-volume accounts made for the brand.
So we reran each comparison inside the same program, creator-size band and length band, and weighted the strata.
| Comparison | Raw hit 10x | Like for like | Programs |
|---|---|---|---|
| Benefit or result claim vs other written captions | 6.2% vs 9.4% | 3.6% vs 3.2% | 15 |
| Ends with "!" vs not | 5.8% vs 9.5% | 4.1% vs 4.1% | 14 |
| Tip, hack or how-to vs other written captions | 5.6% vs 9.3% | 3.3% vs 4.3% | 8 |
| Has a number vs not | 16.8% vs 8.2% | 9.1% vs 5.9% | 16 |
| Question vs other written captions | 11.3% vs 8.9% | 7.5% vs 5.1% | 11 |
| All lowercase vs not | 9.4% vs 8.9% | 7.5% vs 5.6% | 13 |
| 11+ words vs 1 to 10 words | 12.2% vs 8.0% | 7.7% vs 6.4% | 20 |
| No caption text vs any written caption | 12.4% vs 9.0% | 11.3% vs 7.4% | 11 |
| Talks to the viewer (you, your) vs not | 8.1% vs 9.2% | 5.1% vs 6.2% | 18 |
- With the trait
- Without
- Has a numberwith it9.1%
- Has a numberwithout5.9%
- A real questionwith it7.5%
- A real questionwithout5.1%
- All lowercasewith it7.5%
- All lowercasewithout5.6%
- 11+ word openingwith it7.7%
- 11+ word openingwithout6.4%
- No caption textwith it11.3%
- No caption textwithout7.4%
- Talks to the viewerwith it5.1%
- Talks to the viewerwithout6.2%
Within the same program, creator size and video length. Compare the two bars in a pair, not across pairs.
Like-for-like rates are lower overall, because the strata that qualify sit mostly in large, high-volume programs. Compare within a row.
| Opening line | Videos | Median views vs brand typical | Hit 10x |
|---|---|---|---|
| No words | 847 | 1.00x | 12.4% |
| 1 to 5 words | 4,206 | 0.97x | 7.7% |
| 6 to 10 words | 4,008 | 1.01x | 8.2% |
| 11 to 20 words | 1,908 | 1.05x | 12.1% |
| 21 or more words | 880 | 1.05x | 12.4% |
Questions: lowest median, more breakouts
Question captions had the lowest median of any hook type (0.87x) but a higher 10x rate: 11.3% raw, and 7.5% vs 5.1% like for like. They are a high-variance bet: a lower floor, with more breakouts.
What to put in a UGC brief
- Drop mandatory caption hook lists. Give creators the product truth and let them write the opening.
- Ask for a true, specific number in the first line: a price, a count, a day in a series or a time saved.
- Allow real questions.
- Avoid "you need this" openings.
- Don't force a caption.
- Put the brief's weight on video length and creator fit. See the length benchmarks, and write it up with the free UGC campaign brief generator.
To see which hooks a specific program's creators use, open its page in the top UGC programs ranking.
Method
- Universe: 46 fully scanned brand creator programs. Brand videos from 1 August 2026 to each program's measured date (latest 30 October 2026), from confirmed or likely accounts, with a public view count, in programs with 15+ videos. This is the same set as the video benchmarks.
- Score: each video's views divided by its program's median views in the window. A 10x hit is a score of 10 or more.
- Hook: the first line of the caption, with hashtags and account mentions removed. On-screen text and the spoken first line are not public data, so they're not measured.
- Hook type: a keyword classifier, the same one behind the hook mix on our brand pages. The first matching rule wins, and "Other statement" is kept as an honest bucket.
- Published segments: at least 150 videos from at least 8 programs.
- Like for like: comparisons within strata of program, creator-size band (under 1K, 1K to 10K, 10K to 100K, 100K+, unknown) and length band (under 15s, 15 to 30s, 30 to 60s, 60s to 2 min, over 2 min, slideshow, unknown). Only strata with 5+ videos on each side count, weighted by nA·nB/(nA+nB).
- Privacy: aggregates only. No caption, handle or video link is published. The examples on this page are generic templates, not quotes.
Programs with the most videos in the study include StudyFetch, Cantina, Celsius, Poppi and Wispr Flow.
Limits
- Captions only. Not the spoken or on-screen hook.
- Keyword classification. It is not human-coded; 45% of openings land in "Other statement".
- Small samples. The like-for-like comparisons above cover 8 to 20 programs each.
- Correlation, not experiments.
- Organic public views only. No sales, installs or watch time.
- Coverage. Mostly consumer apps and DTC, and 85% TikTok.
Cite and download
Cite this study: UGC Agent, "UGC hooks study 2026" (11,849 brand-program videos, 1 August 2026 to 30 October 2026), UGC Agent research, published 1 November 2026, https://ugcagent.app/research/ugc-hooks-study-2026/
Download the segment aggregates (hook type, opening length, caption ingredients) as ugc-hooks-2026.csv (28 rows) and the like-for-like comparisons as ugc-hooks-like-for-like-2026.csv (17 rows). Free to reuse with attribution under CC BY 4.0.