[{"data":1,"prerenderedAt":662},["ShallowReactive",2],{"post-\u002Fblog\u002Fthe-attention-market":3},{"id":4,"title":5,"author":6,"body":7,"date":649,"description":650,"extension":651,"meta":652,"navigation":653,"path":654,"seo":655,"stem":656,"tags":657,"__hash__":661},"blog\u002Fblog\u002Fthe-attention-market.md","YouTube is a market for time, and everyone is measuring clicks","SourceSignal",{"type":8,"value":9,"toc":639},"minimark",[10,19,31,34,39,46,49,59,78,89,92,96,99,234,241,248,255,259,262,269,272,334,345,352,358,365,369,372,398,409,416,420,423,434,437,542,549,571,578,582,585,588,594,608,611,617,620,623],[11,12,13,14,18],"p",{},"Every tool that measures YouTube measures ",[15,16,17],"strong",{},"views",". A view is a click: it tells you someone\nstarted. It says nothing about whether they stayed.",[11,20,21,22,26,27,30],{},"We store how long every video runs — ",[23,24,25],"code",{},"duration_sec",", on 100% of the 12.4M rows we serve. That\nmakes a second number computable: ",[15,28,29],{},"hours actually watched",", once you discount a view for the\npart nobody sits through. It sounds like a minor variation on reach. It isn't. It's a different\naxis, and it disagrees with reach almost everywhere we look.",[11,32,33],{},"Everything below is measured against our production data across 15 niches, deduplicated per\nvideo, and restricted to channels that actually compete in the niche. Where a number rests on\na modelling assumption rather than an observation, we say so.",[35,36,38],"h2",{"id":37},"how-we-discount-a-view","How we discount a view",[11,40,41,42,45],{},"Before any table, the obvious objection: ",[15,43,44],{},"a view is not a completed view."," Multiply views by\nduration and you credit a three-hour podcast with three full viewer-hours every time someone\nclicks it and leaves after ninety seconds. Do that and \"attention\" just becomes a leaderboard of\nwho publishes the longest videos.",[11,47,48],{},"So we discount every view by how much of the video actually gets watched:",[50,51,56],"pre",{"className":52,"code":54,"language":55},[53],"language-text","attention   = views × duration × retention(duration)\nretention(L_minutes) = clamp( 0.79 − 0.124·ln(L),  0.15,  0.85 )\n","text",[23,57,54],{"__ignoreMap":58},"",[11,60,61,62,65,66,69,70,73,74,77],{},"That curve gives roughly: a 3-minute video keeps ",[15,63,64],{},"65%",", ten minutes ",[15,67,68],{},"50%",", an hour ",[15,71,72],{},"28%",", three\nhours ",[15,75,76],{},"15%",". It averages ~43% across our corpus, which is inside YouTube's real 40–50%\naverage-percentage-viewed range.",[11,79,80,83,84,88],{},[15,81,82],{},"We did not measure this curve — nobody can."," YouTube publishes no completion data. What we did\nis fix its ",[85,86,87],"em",{},"shape"," from what's known publicly (retention falls with length, monotonically) and\ncalibrate the constants against published studies — Backlinko's 1.3M-video analysis and a\n10k-video\u002F1,000-creator study. Our curve tracks their reported retention for strong performers at\nevery length we can check: sub-1m 70–85%, under 5m 65–75%, 30–60m 25–35%, 60m+ 20–30%.",[11,90,91],{},"It matters enormously. Uncorrected, the >2h band looks like 34% of all attention; corrected, it's\n17%. Half the number was an artifact of assuming people finish podcasts. Every figure below has\nthe correction applied, and we flag the one place it doesn't matter.",[35,93,95],{"id":94},"shorts-win-the-clicks-and-lose-the-time","Shorts win the clicks and lose the time",[11,97,98],{},"Group every video we track by length:",[100,101,102,121],"table",{},[103,104,105],"thead",{},[106,107,108,112,115,118],"tr",{},[109,110,111],"th",{},"Length",[109,113,114],{},"% of videos",[109,116,117],{},"Share of views",[109,119,120],{},"Share of watch-time",[122,123,124,143,157,171,189,206,220],"tbody",{},[106,125,126,130,133,138],{},[127,128,129],"td",{},"\u003C1m (Shorts)",[127,131,132],{},"14.1",[127,134,135],{},[15,136,137],{},"34.0",[127,139,140],{},[15,141,142],{},"2.2",[106,144,145,148,151,154],{},[127,146,147],{},"1–5m",[127,149,150],{},"37.2",[127,152,153],{},"19.2",[127,155,156],{},"7.2",[106,158,159,162,165,168],{},[127,160,161],{},"5–10m",[127,163,164],{},"23.5",[127,166,167],{},"10.6",[127,169,170],{},"8.7",[106,172,173,178,181,184],{},[127,174,175],{},[15,176,177],{},"10–20m",[127,179,180],{},"18.0",[127,182,183],{},"17.9",[127,185,186],{},[15,187,188],{},"23.7",[106,190,191,196,199,202],{},[127,192,193],{},[15,194,195],{},"20–40m",[127,197,198],{},"5.5",[127,200,201],{},"11.2",[127,203,204],{},[15,205,164],{},[106,207,208,211,214,217],{},[127,209,210],{},"40m–2h",[127,212,213],{},"1.4",[127,215,216],{},"5.1",[127,218,219],{},"17.5",[106,221,222,225,228,231],{},[127,223,224],{},">2h",[127,226,227],{},"0.3",[127,229,230],{},"2.0",[127,232,233],{},"17.2",[11,235,236,237,240],{},"Read the last two columns against each other. Shorts are ",[15,238,239],{},"34% of all reach and 2.2% of all\ntime"," — they pull fifteen times more attention than they hold. At the other end, videos over two\nhours are 0.3% of the catalogue and 17% of every hour watched: over-represented roughly fiftyfold,\nbut not, it turns out, the main event.",[11,242,243,244,247],{},"The main event is the middle. ",[15,245,246],{},"The 10–40 minute bands are 23.5% of videos and 47% of all\nwatch-time"," — the deep review, the follow-along, the teardown. That's where the market actually\nis, and it's invisible if you rank by views.",[11,249,250,251,254],{},"Reach and attention aren't merely different. Across length, they're close to ",[85,252,253],{},"inversely","\ndistributed. A strategy optimised for one is actively wrong for the other.",[35,256,258],{"id":257},"every-niche-is-really-five-markets","Every niche is really five markets",[11,260,261],{},"Here's where it stops being trivia.",[11,263,264,265,268],{},"If you pool all lengths together and measure concentration, the number lies. It degenerates into\n",[85,266,267],{},"\"who makes the longest videos\""," — because the 40m+ bands carry 35% of all watch-time on 1.7% of\nthe videos, they drag any aggregate toward their own structure.",[11,270,271],{},"Take Coffee & Espresso. Measured as one market, it looks like a tight oligopoly: roughly seven\neffective competitors. Split it by length:",[100,273,274,284],{},[103,275,276],{},[106,277,278,281],{},[109,279,280],{},"Band",[109,282,283],{},"Effective competitors",[122,285,286,294,306,314,322],{},[106,287,288,291],{},[127,289,290],{},"Shorts \u003C1m",[127,292,293],{},"6.0",[106,295,296,301],{},[127,297,298],{},[15,299,300],{},"Short-form 1–10m",[127,302,303],{},[15,304,305],{},"28.7",[106,307,308,311],{},[127,309,310],{},"Mid 10–20m",[127,312,313],{},"8.0",[106,315,316,319],{},[127,317,318],{},"Long 20–40m",[127,320,321],{},"3.9",[106,323,324,329],{},[127,325,326],{},[15,327,328],{},"Long-form 40m+",[127,330,331],{},[15,332,333],{},"3.1",[11,335,336,337,340,341,344],{},"Coffee is not a concentrated niche. It's ",[15,338,339],{},"two markets stacked",": the short-form review market is\na ~29-way free-for-all, and the long-form market is a ",[15,342,343],{},"3-way lockup",". The pooled \"seven\" was\nthe long-form oligopoly bleeding through.",[11,346,347,348,351],{},"A creator who read the pooled number would conclude Coffee is closed and go elsewhere. The real\nanswer is: ",[85,349,350],{},"short-form is wide open, long-form is owned by three players."," Opposite conclusions,\nsame data. You cannot get this from view counts, and you cannot get it from pooled attention\neither — it exists only once you band by length.",[11,353,354,355],{},"That generalises. For any niche we can now say: ",[15,356,357],{},"short-form is open (enter here); long-form is\nowned by N players (here's who).",[11,359,360,361,364],{},"We also checked whether banding is a real division or a convenient one. Per creator, the\ncorrelation between short-form success and long-form success is ",[15,362,363],{},"~0.3 across every niche"," —\nthe people winning short-form are, empirically, different people from the ones winning\nlong-form. They're separate markets, not one market viewed at different zoom levels.",[35,366,368],{"id":367},"two-leaderboards-and-most-people-only-see-one","Two leaderboards, and most people only see one",[11,370,371],{},"If reach and attention disagree, then \"who owns this niche\" has two answers.",[373,374,375,386,392],"ul",{},[376,377,378,381,382,385],"li",{},[15,379,380],{},"Outdoor Boys"," — #2 in the entire corpus by watch-time. ",[15,383,384],{},"#1,132 by keyword footprint.","\nIt ranks for about 900 keywords and runs the second-largest attention economy we measure.",[376,387,388,391],{},[15,389,390],{},"Linus Tech Tips"," — #1 by watch-time, #8 by footprint.",[376,393,394,397],{},[15,395,396],{},"Formula 1"," — #13 by watch-time, #8,846 by footprint. Fifty-seven keywords.",[11,399,400,401,404,405,408],{},"Footprint counts ",[85,402,403],{},"coverage",": how much of the search surface you touch. Attention counts\n",[85,406,407],{},"consumption",": how much of the audience's time you hold. Both are real, both are useful, and\nthey are not the same leaderboard. A competitive analysis that reports one and calls it \"market\nshare\" is answering a question you didn't ask.",[11,410,411,412,415],{},"The niches themselves reorder too. Measured in watch-hours per day, they span a ",[15,413,414],{},"32× range"," —\nfrom Gaming Peripherals at ~1.5M down to Home Office at ~48k — and the ranking doesn't match the\nreach ranking, because format mix converts reach into time at very different rates.",[35,417,419],{"id":418},"wanting-to-buy-and-being-able-to-buy-are-different-things","Wanting to buy and being able to buy are different things",[11,421,422],{},"One more divergence, because it's the one with money attached.",[11,424,425,426,429,430,433],{},"Two independent questions decide whether a niche's attention is worth anything commercially.\nDoes the audience ",[15,427,428],{},"arrive wanting to buy","? And are creators ",[15,431,432],{},"wired up to capture it"," — are\nthere affiliate or commerce links on the videos holding the attention?",[11,435,436],{},"We can measure both, bottom-up. They disagree, and the gap is the interesting part:",[100,438,439,452],{},[103,440,441],{},[106,442,443,446,449],{},[109,444,445],{},"Niche",[109,447,448],{},"Arrives wanting to buy",[109,450,451],{},"Wired to capture it",[122,453,454,469,480,491,503,514,531],{},[106,455,456,461,466],{},[127,457,458],{},[15,459,460],{},"Running",[127,462,463],{},[15,464,465],{},"80%",[127,467,468],{},"26%",[106,470,471,474,477],{},[127,472,473],{},"Golf",[127,475,476],{},"38%",[127,478,479],{},"6%",[106,481,482,485,488],{},[127,483,484],{},"E-Bikes",[127,486,487],{},"41%",[127,489,490],{},"16%",[106,492,493,496,498],{},[127,494,495],{},"Coffee",[127,497,476],{},[127,499,500],{},[15,501,502],{},"67%",[106,504,505,508,511],{},[127,506,507],{},"Smart Home",[127,509,510],{},"31%",[127,512,513],{},"61%",[106,515,516,521,526],{},[127,517,518],{},[15,519,520],{},"Power Tools",[127,522,523],{},[15,524,525],{},"18%",[127,527,528],{},[15,529,530],{},"53%",[106,532,533,536,539],{},[127,534,535],{},"Gaming",[127,537,538],{},"46%",[127,540,541],{},"48%",[11,543,544,545,548],{},"Three regimes fall out. Where ",[15,546,547],{},"intent far exceeds wiring"," — Running at 80\u002F26 — the audience\nshows up ready to spend and nobody has built the path. That's either untapped money or a\ncategory that monetises somewhere off-platform.",[11,550,551,552,555,556,559,560,563,564,563,567,570],{},"Where ",[15,553,554],{},"wiring far exceeds intent"," — Power Tools at 18\u002F53 — it means our keyword signal is\nblind. Tool queries are model numbers, ",[23,557,558],{},"DeWalt DCD777",", with none of the ",[85,561,562],{},"best"," \u002F ",[85,565,566],{},"vs",[85,568,569],{},"review","\nwords that mark a shopping query. But the creators affiliate-link everything, because they know\nwhat the traffic is worth. Here the affiliate signal corrects the keyword signal.",[11,572,573,574,577],{},"And where the two ",[15,575,576],{},"agree"," — Gaming at 46\u002F48 — you can trust the read.",[35,579,581],{"id":580},"what-were-claiming-and-what-were-not","What we're claiming, and what we're not",[11,583,584],{},"The band table, the concentration numbers, the reach-attention rank flips and the intent-wiring\ngaps are all measured. They come from data we already collect; no new pipeline was needed.",[11,586,587],{},"Two honest caveats.",[11,589,590,593],{},[15,591,592],{},"The retention curve is calibrated, not measured."," Its shape is defensible and it matches every\npublished benchmark we could check, but the constants are a considered guess, because the ground\ntruth is private to YouTube. So we tested how much our results depend on it — two answers, and\nthey differ:",[373,595,596,602],{},[376,597,598,601],{},[15,599,600],{},"Within a band it's a rounding error."," Durations inside a band are similar, so the discount\ncancels out of the share maths: turn it off entirely and Coffee's short-form number moves 28.7\n→ 28.2, its long-form 3.1 → 3.2. The market-structure findings above don't rest on the curve\nat all — that's why we lead with them.",[376,603,604,607],{},[15,605,606],{},"Across bands it's load-bearing."," It halves the >2h band, and at niche level it flips the #1\nspot outright: raw, Golf leads on watch-time; corrected, Gaming does. Golf's lead was an\nartifact — 69% of its attention is 40m+ tour broadcasts that nobody finishes.",[11,609,610],{},"That's the honest split. The per-band conclusions are robust; the cross-band totals carry an\nassumption, and we say so every time we print one.",[11,612,613,616],{},[15,614,615],{},"The wiring number is a floor."," We detect affiliate and commerce domains in video\ndescriptions. Link shorteners hide some, discount codes and sponsorships don't leave a link at\nall. Real commercial wiring is higher than what we count.",[11,618,619],{},"This is a research finding, not a product feature. The attention model isn't in our reports\nyet — it's the next thing we're building, on data that's already sitting in the warehouse.",[621,622],"hr",{},[11,624,625],{},[85,626,627,628,633,634,638],{},"Every figure above traces back to specific videos and searches in our dataset. You can poke at\nthe same data yourself — ",[629,630,632],"a",{"href":631},"\u002Fniches","the niches we cover",", or ",[629,635,637],{"href":636},"\u002Ftools\u002Fyoutube-visibility-check","check a channel",",\nno signup.",{"title":58,"searchDepth":640,"depth":641,"links":642},2,3,[643,644,645,646,647,648],{"id":37,"depth":640,"text":38},{"id":94,"depth":640,"text":95},{"id":257,"depth":640,"text":258},{"id":367,"depth":640,"text":368},{"id":418,"depth":640,"text":419},{"id":580,"depth":640,"text":581},"2026-07-16","Shorts take 34% of the views and 2.2% of the watch-time. Once you measure hours instead of clicks, the leaderboard in every niche reorders — and each niche turns out to be five separate markets stacked on top of each other.","md",{},true,"\u002Fblog\u002Fthe-attention-market",{"title":5,"description":650},"blog\u002Fthe-attention-market",[658,659,660],"attention","market-structure","methodology","6z8DKHC0vGnfNWUEaXgkm-N8z6P31nzuARogkFCshfw",1784209707639]