The most-used video opening in five app categories has the top median views in one of them
Curiosity-gap hooks are 35.8% to 55.6% of the classified videos in these categories. They lead median views in Education and rank fifth of six in Productivity, at 18,650 against 47,074.
90 days to 14 August 2026, plus the all-time classified library for context
- 12,299
- Videos opening with a curiosity gap
- 44.3%
- Share of the five categories' classified videos
- 1 of 5
- Categories where that opening leads median views
27,756 classified videos across five categories · 90 days to 14 August 2026
12,299 of the 27,756 classified videos in these five app categories open with a curiosity gap. That is 44.3% of them, and it is the most-used opening in all five.
It has the top median views in one of them. In Education the curiosity gap leads at 35,050 median views from a 100-video draw. A different opening leads in the other four. Finance goes to the problem and Health & Fitness to the direct callout. Photo & Video and Productivity both go to the question.
Productivity is the sharp case. The curiosity gap is 1,686 of that category's 4,710 classified videos in the window, more than any other opening. It ranks fifth of six by median views: 18,650 against 47,074 for the question.
The curiosity gap is the most-used opening in every one of the five
Exact counts of classified videos published in the 90 days to 14 August 2026, matched to an app in that category.
How crowded the curiosity gap is in each category, against the next-most-used opening. The denominator is the classified videos in that category and window, which is the sum of the nine opening cells.
| Category | Classified videos | Curiosity gap | Share | Next-most-used opening |
|---|---|---|---|---|
| Education | 4,393 | 2,442 | 55.6% | Direct callout, 654 (14.9%) |
| Health & Fitness | 4,918 | 2,474 | 50.3% | Problem, 803 (16.3%) |
| Photo & Video | 6,626 | 2,992 | 45.2% | Problem, 1,669 (25.2%) |
| Finance | 7,109 | 2,705 | 38.1% | Problem, 1,548 (21.8%) |
| Productivity | 4,710 | 1,686 | 35.8% | Problem, 1,448 (30.7%) |
| All five | 27,756 | 12,299 | 44.3% | Problem, 6,012 (21.7%) |
The category denominator counts only videos the classifier has read. Unclassified videos in the same window are in neither column. The window is 90 days because browse counts stop reporting a number above 10,000. 90 days is the widest window that keeps the largest cell here under it.
27,756 classified videos across five categories · 90 days to 14 August 2026
The crowd is not a five-category quirk. Across every classified video in the library, all time, 185,843 of 419,998 open with a curiosity gap. That is 44.2%, against 44.3% for the five categories in the window.
Two denominators sit in that comparison and they are built differently. The 419,998 is every classified video in the library, with no category filter and no date cutoff. The 27,756 counts only videos matched to an app in one of these five categories and published in the last 90 days. The two agree to within 0.1 points, which is a coincidence worth naming and not a check on either figure.
The opening that leads two of our five categories is a thin lane library-wide. 8,767 classified videos open with a question, 2.1% of the total, against 185,843 that open with a curiosity gap.
Every classified organic video in the library, by the one opening the classifier assigned it. Nine values, one per video, summing to the denominator.
| Opening | Videos | Share of 419,998 |
|---|---|---|
| Curiosity gap | 185,843 | 44.2% |
| Problem | 81,596 | 19.4% |
| Bold claim | 66,267 | 15.8% |
| Direct callout | 61,191 | 14.6% |
| Result first | 10,000 | 2.4% |
| Question | 8,767 | 2.1% |
| Other | 4,662 | 1.1% |
| Social proof | 1,472 | 0.4% |
| Warning | 200 | 0.05% |
Other is a residual bin and not an opening, which is why it is excluded from every ranking on this page. Result first reads as exactly 10,000. We cannot tell from this counter whether that is the count or a ceiling, and the nine values still sum to 419,998. Shares are rounded and sum to 100.05%. This table is all-time and library-wide, so it is not comparable with the windowed per-category figures above it.
419,998 classified videos · Snapshot of 14 August 2026
A different opening leads median views in four of the five categories
One draw per opening per category: the 100 most recent matched videos in the window, or all of them where fewer exist.
The most-used opening in each category against the opening with the top median views. Every median carries the size of the draw it was computed from.
| Category | Most-used opening | Its median views | Its rank by views | Top median views | That median |
|---|---|---|---|---|---|
| Education | Curiosity gap, 2,442 videos | 35,050 (n=100) | 1st of 6 | Curiosity gap | 35,050 (n=100) |
| Health & Fitness | Curiosity gap, 2,474 videos | 64,950 (n=100) | 2nd of 7 | Direct callout | 67,490 (n=100) |
| Finance | Curiosity gap, 2,705 videos | 24,600 (n=100) | 4th of 7 | Problem | 49,550 (n=100) |
| Photo & Video | Curiosity gap, 2,992 videos | 27,708 (n=100) | 4th of 6 | Question | 44,800 (n=100) |
| Productivity | Curiosity gap, 1,686 videos | 18,650 (n=100) | 5th of 6 | Question | 47,074 (n=88, under the 100-row draw) |
Rank counts only the openings we rank: the residual other bin is out, and so is any cell whose draw came back under 20 rows. That drops warning in all five categories, at n=3, 8, 2, 1 and 5. It also drops social proof in Education (n=12), Photo & Video (n=12) and Productivity (n=3). Each category ranks six or seven openings.
2,976 videos in 32 ranked draws · 90 days to 14 August 2026
Two of the four leads are wide enough to brief against. In Finance the problem opening returns 49,550 median views against 24,600 for the curiosity gap, from 100-video draws on both sides. In Productivity the question returns 47,074 from 88 videos against 18,650 from 100. Both gaps are above two to one.
Health & Fitness is a tie and we will print it as one. The direct callout leads at 67,490 and the curiosity gap follows at 64,950, a difference of 2,540 views. 48 of the 100 curiosity-gap videos cleared the leader's median.
Education is the one category where the most-used opening also leads the views column, at 35,050. Second is the bold claim at 30,901, so that lead is narrow too.
The leading openings are thin lanes. The question opening leads Productivity on 88 matched videos in 90 days, against 1,686 for the curiosity gap. Where a draw equals the videos-in-window count, the cell is not a sample: it is every matched video with that opening in the window. Six of the 32 ranked cells are counted that way, including the Productivity question.
Both columns for every opening we ranked, in every category
Median views, median virality and median follower count, side by side, so a gap in one column can be checked against the others.
The 32 ranked cells, grouped by category and ordered by median views inside each. Videos in window is an exact count; the three medians come from the draw beside them.
| Category | Opening | Videos in window | Draw | Median views | Median virality | Median followers |
|---|---|---|---|---|---|---|
| Education | Curiosity gap | 2,442 | 100 | 35,050 | 27.8 | 58,587 |
| Education | Bold claim | 439 | 100 | 30,901 | 18.4 | 316,333 |
| Education | Direct callout | 654 | 100 | 25,386 | 17.1 | 286,836 |
| Education | Question | 197 | 100 | 19,356 | 21.4 | 48,219 |
| Education | Problem | 544 | 100 | 18,691 | 21.1 | 89,054 |
| Education | Result first | 45 | 45, under 100 | 16,800 | 17.8 | 107,402 |
| Health & Fitness | Direct callout | 719 | 100 | 67,490 | 22.1 | 220,400 |
| Health & Fitness | Curiosity gap | 2,474 | 100 | 64,950 | 31.1 | 61,973 |
| Health & Fitness | Bold claim | 546 | 100 | 55,546 | 26.2 | 55,436 |
| Health & Fitness | Problem | 803 | 100 | 39,511 | 19.9 | 114,561 |
| Health & Fitness | Social proof | 22 | 22, under 100 | 38,558 | 28.3 | 15,189 |
| Health & Fitness | Result first | 163 | 100 | 37,031 | 19.0 | 150,446 |
| Health & Fitness | Question | 102 | 100 | 31,851 | 19.6 | 114,561 |
| Finance | Problem | 1,548 | 100 | 49,550 | 27.5 | 57,031 |
| Finance | Bold claim | 1,446 | 100 | 36,650 | 24.2 | 175,490 |
| Finance | Question | 137 | 100 | 30,397 | 22.1 | 126,956 |
| Finance | Curiosity gap | 2,705 | 100 | 24,600 | 29.8 | 55,461 |
| Finance | Social proof | 32 | 32, under 100 | 24,463 | 17.2 | 124,492 |
| Finance | Result first | 95 | 95, under 100 | 19,600 | 20.5 | 33,499 |
| Finance | Direct callout | 1,096 | 100 | 17,650 | 23.6 | 102,815 |
| Photo & Video | Question | 105 | 100 | 44,800 | 28.7 | 66,630 |
| Photo & Video | Direct callout | 863 | 100 | 32,500 | 27.4 | 99,100 |
| Photo & Video | Result first | 129 | 100 | 28,842 | 18.8 | 79,125 |
| Photo & Video | Curiosity gap | 2,992 | 100 | 27,708 | 27.7 | 75,544 |
| Photo & Video | Bold claim | 785 | 100 | 27,146 | 24.2 | 56,170 |
| Photo & Video | Problem | 1,669 | 100 | 20,400 | 24.4 | 117,600 |
| Productivity | Question | 88 | 88, under 100 | 47,074 | 24.4 | 42,500 |
| Productivity | Direct callout | 599 | 100 | 30,638 | 24.2 | 113,345 |
| Productivity | Result first | 94 | 94, under 100 | 28,206 | 21.6 | 86,400 |
| Productivity | Problem | 1,448 | 100 | 26,300 | 25.2 | 80,385 |
| Productivity | Curiosity gap | 1,686 | 100 | 18,650 | 31.3 | 3,327 |
| Productivity | Bold claim | 754 | 100 | 14,244 | 21.5 | 91,663 |
Views are read as the greater of the two view fields, because TikTok reports its views in the play column and leaves the other null. Medians are rounded to the nearest view and virality to one decimal. A draw under 100 is the whole population of that cell in the window, not a shortfall. The residual other bin and every draw under 20 rows are excluded, and their counts are listed under the table above.
2,976 videos in 32 ranked draws · 90 days to 14 August 2026
Read the columns separately. Median views says how far a video travelled. Median virality says how far it travelled for the size of the account behind it. Half of that score is views divided by followers.
The median follower column is there so you can tell the two apart in a single row. Where an opening holds a high virality figure and a low views figure, look left at the follower count before you believe the virality.
The curiosity gap holds the top median virality in Education, Finance, Health & Fitness and Productivity. In Photo & Video it is second, behind the question at 28.7 against 27.7. It holds the top median views in Education alone.
In Productivity the crowded opening tops one column and sits near the bottom of the other
The same two draws, on both measures, with the creators' median follower count beside them.
- 31.3
- Median virality, curiosity gap
- 24.4
- Median virality, question
- 3,327
- Median followers, curiosity gap
- 42,500
- Median followers, question
188 Productivity videos: 100 curiosity gap and 88 question · 90 days to 14 August 2026
The curiosity gap posts the highest median virality of any opening in Productivity, at 31.3, and the fifth-of-six median views, at 18,650. Its creators are smaller than any other Productivity cell's by an order of magnitude.
Median follower count in that draw is 3,327, against 42,500 for the question opening. 15 of its 100 rows come from accounts under 1,000 followers. 36 of the 100 come from four accounts whose handles all carry the same app's name. That is one company's creator roster, not a category-wide pattern.
Virality is weighted toward views over followers by construction, so ranking by it partly ranks by whose creators are small. That is the mechanism behind the 31.3, and it is why we are not folding the two columns into one number.
The leader's draw has a concentration problem of its own in another category. In Health & Fitness the direct callout leads on median views. 38 of its 100 rows come from two accounts, 27 of them from one.
Two columns, two questions, and we are not resolving them for you
Median views asks how far a video travelled. Median virality asks how far it travelled for the size of the account that posted it. Virality is weighted toward views over followers by construction, so ranking by it partly ranks by whose creators are small. We print both and we do not combine them. If you are buying reach, read the views column. If you are looking for openings that travel without an audience behind them, read the virality column. Check the follower count in the same row before you believe it.
What to brief on Monday, and what to stop briefing
Four instructions, in the order they matter, with the condition that would make each one wrong.
Stop opening with a curiosity gap by default. It is the most-used opening in all five categories here, and in four of them another opening holds the higher median views. Being the crowded lane is the finding about it, and not an argument for it.
Brief your category's leader instead. The problem in Finance. The question in Photo & Video and in Productivity. The direct callout or the curiosity gap in Health & Fitness, and the curiosity gap in Education. Where the top two sit within a few thousand views of each other, pick on other grounds and do not tell yourself the data chose.
Change one thing. Give the same creator the same product beat and the same length, and move the opening. A hook comparison run across different creators compares creators.
Put a number in the brief and treat it as a middle. A question opening in our Productivity draw put the middle video at 47,074 views. Half of the 88 landed below it.
What would have to be true for this to be advice
Three things, and we have measured none of them. The ordering would have to hold on a second draw. We took one set of videos per cell and never repeated it, so we cannot say how much of a 47,074-against-18,650 gap survives a re-draw. The opening would have to be the thing that differs, and it is not. In Productivity the curiosity-gap draw runs a median 22 seconds against 44 for the question draw. 69 of its 100 rows are TikTok, against 54 of 88. And the labels would have to be right, which is a classification we spot-check and have not benchmarked. Until those hold, this is an observation about what travelled in 90 days and not a rule about what works.
A median is a middle, and it is not a promise
How far the two Productivity draws overlap, measured on the same rows the medians came from.
36 of the 100 Productivity curiosity-gap videos beat the question opening's median of 47,074. 35 of the 88 question videos landed below the curiosity gap's median of 18,650. The two draws overlap across most of their range.
The tails cross as well. At the 90th percentile the curiosity-gap draw reaches 1,000,000 views and the question draw reaches 600,500. The opening with the lower median holds the higher upside, in the same category and the same window.
A median view count describes the middle of one draw. It is not a forecast for the video you are about to brief. That difference is most of what a founder gets wrong with a table like this one.
What these numbers cannot tell you
Five limits, starting with the two that decide how much weight the table can carry.
This is five categories, not 31. There is no per-category facet aggregate in our API, so every cell is its own query. Nine openings across 31 categories is 279 queries against a box that also serves customers. We ran 45. Games, Shopping, Entertainment and 23 other categories are absent from this page, and nothing here should be read across to them.
Every opening is a machine label. Every video is classified by a language model into a fixed vocabulary of formats, hooks, tones and CTAs. It is a classification, not a human tag, and it is spot-checked, not benchmarked, so we publish it as a distribution across thousands of posts and never as a judgement about any single video. On TikTok we read the caption and the spoken transcript. On Instagram we read the caption alone. An opening that exists only in on-screen text is invisible to us on both platforms.
No outcome is attached to any of these views. We hold no installs per post, no link clicks and no store visits traced back to a video. A higher median view count is a higher median view count. Nothing on this page says an opening produced a download.
The draws are shallow and the creators repeat. The 3,311 rows behind the medians come from 2,129 distinct accounts across the five categories. An account here is one handle on one platform: we do not stitch the same person across TikTok and Instagram. Two cells lean on a handful of accounts, and both are named above. The draw does not record which app each video was matched to. So we cannot print the distinct-app count our own format rules ask for. A category's numbers can be a few companies' creative strategy wearing a category's name.
The library is a sample of our customers' questions. Content is linked to apps by store links where we can find them and by name matching otherwise. Name matching is deliberately generous and it is not perfect. Category filters resolve through that matched app, so unmatched content is absent from every count here. These figures describe Apptonic's own library as of 14 August 2026. The library is a growing sample of the app market, not a census.
How we counted
Every figure comes from one frozen snapshot of the library, harvested on 14 August 2026. Nothing on this page is an estimate: every count is a row count and every median was computed from rows we hold.
The per-category counts are exact. Our facet resolver takes no category argument. Each cell was counted on its own through the video browse, filtered to one category and one opening. Every cell in every category uses the same cutoff of 16 May 2026. It was chosen by probing the two largest expected cells across 90, 60, 30 and 14-day candidates. The widest one that stays under the 10,000 count cap won. Because the window is identical everywhere, the categories compare with each other. They do not compare with the all-time library table, which has no window and no category filter.
The medians come from one draw per cell: the 100 most recent matched videos with that opening in the window, sorted newest first. The feed's viral sort was not used. It applies a penalty per repeat of the same app, which makes it a diversity shuffle instead of a ranking. 3,311 rows were drawn in total, published between 16 May and 13 August 2026, across 2,129 distinct accounts. 2,976 of those rows sit in the 32 cells we rank. Views are read as the greater of the two view fields, because TikTok reports views in the play column and leaves the other null. Medians are computed locally over the returned rows and rounded to the nearest view; virality is rounded to one decimal.
Two classes of cell are excluded from every ranking. The residual other bin is not an opening type. And no median is quoted off a draw under 20 rows. That drops warning in all five categories, and social proof in Education, Photo & Video and Productivity. Their counts are printed beside the ranking table. Where a draw is smaller than 100 but at or above 20, it is the whole population of that cell in the window. The table labels it.
Virality is Apptonic's own 0-100 score, weighted toward view-to-follower breakout and recency. It measures what is travelling now, not lifetime performance, and it is a ranking signal, not a platform-reported metric. Half its weight is views divided by followers. That is why it is printed next to a median follower count and never merged with the views column.
Every video is classified by a language model into a fixed vocabulary of formats, hooks, tones and CTAs. It is a classification, not a human tag, and it is spot-checked, not benchmarked, so we publish it as a distribution across thousands of posts, never as a judgement about any single video.
Content is linked to apps by store links where we can find them and by name matching otherwise. Name matching is deliberately generous and it is not perfect. These figures describe Apptonic's own library as of 14 August 2026. The library is a growing sample of the app market, not a census.
- Classified organic videos matched to an app in one of five categories
- 27,756 rows · 90 days to 14 August 2026
- Videos drawn for the medians: the 100 most recent per opening per category
- 3,311 rows · Published 16 May to 13 August 2026
- Organic videos carrying an AI classification, whole library
- 419,998 rows · Snapshot of 14 August 2026
- Organic videos held in the Apptonic library
- 565,898 rows · Snapshot of 14 August 2026
Frequently asked
- Which video opening gets the most views in my app category?
- We measured five categories over the 90 days to 14 August 2026. Education: the curiosity gap, 35,050 median views from 100 videos. Finance: the problem, 49,550 from 100. Health & Fitness: the direct callout, 67,490 from 100. Photo & Video: the question, 44,800 from 100. Productivity: the question, 47,074 from 88. Health & Fitness and Education are near-ties and should be read as such.
- Should I stop using curiosity-gap hooks?
- The data says stop reaching for one by default, and it does not say the hook fails. The curiosity gap is the most-used opening in all five categories, and in four of them another opening posts higher median views. In Education it leads. It also holds the top median virality in four of the five. That is a different question, and partly a reflection of its creators being smaller.
- Why five categories and not all 31?
- Because there is no per-category facet aggregate in our API. Each cell is a separate query, so a full table would be 279 of them against a box that also serves customers. We ran nine openings across five categories, which is 45 queries, and the other 26 categories are absent from this page. Games, Shopping and Entertainment are among them.
- Why do median views and median virality rank the openings differently?
- Because they answer different questions. Median views is raw reach. Virality is our own 0-100 score and half its weight is views divided by followers, so it rewards a video that escaped its creator's audience. Ranking by virality partly ranks by whose creators are small. In Productivity the curiosity gap holds the top virality figure at 31.3 and the fifth-of-six views figure at 18,650. Its creators' median follower count is 3,327, against 42,500 for the question opening.
- If I brief a question hook, should I expect 47,074 views?
- No. 47,074 is the middle of one draw of 88 Productivity videos, which means half of them landed below it. 35 of those 88 fell below the curiosity gap's median of 18,650, and 36 of the 100 curiosity-gap videos cleared 47,074. The two distributions overlap across most of their range.
- How do you decide what a video's opening is?
- A language model reads the caption, plus the spoken transcript where one resolved. It picks one of nine values: question, bold claim, problem, result first, curiosity gap, social proof, warning, direct callout, or other. It is a classification and not a human tag, and it is spot-checked, not benchmarked. We publish it as a distribution across thousands of posts, never as a judgement about any single video.
- Do these figures cover every video in the category?
- No. Category filters resolve through the app a video is matched to, so a video we could not match to an app is invisible to them. The counts describe matched content and not the corpus. The medians go further and describe one draw of the 100 most recent videos per cell, which is a draw and not a census.