The 90/10 Rule: Why One Post Wins

Pull last month's analytics for any creator publishing daily and the numbers usually look the same. Thirty posts went out. Total reach was 720,000 views. One post did 540,000 of them. The next two did roughly 60,000 each. The remaining twenty-seven posts split 60,000 views between them — an average of about 2,200 each, with several below 500. This is not a bad month. This is power law content working exactly as it works on every algorithmic feed in 2026, and the sooner you plan around it the sooner your calendar starts compounding.

The instinct, looking at that table, is to ask what went wrong with the twenty-seven. The more useful question is what went right with the one — and whether the four traits it shares with other breakout posts can be reproduced on purpose. That is what this post is about.

What is power-law content?#

Power-law content is a distribution pattern where a small number of posts produce a disproportionately large share of total reach, and the gap between the top item and the median item grows wider as the dataset grows. Mathematically, the frequency of outcomes follows P(x) ∝ x^-α — the probability of a result of size x falls off as a power of x, not exponentially and not symmetrically around an average. Plotted on a normal axis it looks like a cliff; plotted on a log-log axis it looks like a straight line.

That matters because the normal distribution most people carry in their head — the bell curve, where most outcomes cluster near the mean and extremes are rare — does not describe content performance. A bell curve says your worst post and your best post should be roughly equidistant from your average. A power law says your best post can be 100x your median while every other post sits in a long, thin tail close to zero. The 2009 SIAM Review paper Power-law distributions in empirical data by Clauset, Shalizi and Newman catalogued this shape across web traffic, citations, word frequencies and city populations. Social feeds belong to the same family.

The related, narrower observation creators usually quote is the 80/20 Pareto rule — that 80% of outcomes come from 20% of inputs. The honest version for algorithmic feeds is closer to 90/10, and Nielsen Norman Group's long-running work on participation inequality documents an even sharper 90-9-1 split for user contribution online. Creator output is steeper still, because on top of the natural inequality of what people make there is a second multiplicative layer: what the algorithm decides to amplify. The two stack.

Why does one post outperform the rest combined?#

Because every modern recommendation feed is a series of gates, and each gate is roughly multiplicative. A small edge at gate one becomes a large gap by gate four.

TikTok's own support article How TikTok recommends content describes the loop in plain language: a new video is shown to a small initial audience, the system measures engagement signals (completion, re-watch, share, comment), and if the signals beat a threshold the video is shown to a larger audience. Public reporting and TikTok's own documentation put the practical shape of that loop roughly like this:

1. A test pool of a few hundred viewers sees the post within minutes of publish.
2. Completion rate, re-watches per view, shares per view and comments per view are measured in the first 30–60 minutes against the creator's recent baseline.
3. If signals clear the threshold, distribution expands to a few thousand viewers and the same signals are re-measured against a stricter baseline.
4. Posts that clear three to five successive gates enter long-tail "For You" delivery, where they can keep accruing views for weeks.

The reason this produces a power law and not a bell curve is that the gates are multiplicative, not additive. A post with 62% completion against your 55% baseline does not get 13% more reach — it gets promoted to the next pool, where its 62% is measured again, against a tougher pool, and if it survives that it is promoted again. Four gates of small edges compound into a 5x–20x reach gap over a post that stalled at gate one. The math is the same on YouTube Shorts, Instagram Reels and X's For You tab; the thresholds and signal weightings differ, the shape does not.

Direct-answer summary: one post outperforms the rest because algorithmic feeds promote winners through a small number of multiplicative gates, and only a handful of posts ever clear all of them.

What do the top 10% of posts have in common?#

Four traits, repeatedly. They are not style choices — they are properties that map directly to the gates above.

1. A topical hook tied to a sub-24-hour event. The breakout post almost always references something whose search or mention velocity doubled in the last six hours: a news beat, a new feature ship, a culture-moment quote, a chart that went around that morning. The first gate punishes generic openings because viewers have already seen forty of them today; topical specificity beats their priors and lifts completion. If you are not watching velocity, you are guessing.
2. Format-native execution. Aspect ratio, caption length, pacing, on-screen text density and music choice match the host platform's median for that vertical. A 16:9 talking-head re-uploaded to TikTok loses the first gate on look alone; a 9:16 vertical with on-screen captions in the platform's default font does not. Format-fit is the cheapest gate to clear and the one most often skipped by creators repurposing without rewriting.
3. Emotional specificity, not generality. The hook names a specific person, object or moment in the first two seconds rather than a category. "My landlord raised my rent by $400 the day after I left a five-star review" out-performs "Renting is brutal right now" by every signal that matters — completion, comment depth, share rate — because specificity is what readers screenshot and quote.
4. Save/share utility. Posts that answer a question viewers would screenshot and send to a friend clear the share-per-view gate at multiples of the baseline. This is why list-format posts, single-frame infographics and "I wish I knew this earlier" teardowns over-index in the top 10%: they are functionally useful out of context.

A post does not need all four. Two reliably puts you in the top 10%; three is usually a breakout; four is the post that ends up screenshotted on X with no attribution.

What this means for your content calendar#

Three concrete shifts, all of which contradict the standard "post consistently and improve over time" advice.

Shoot more shots on goal. If roughly one in fifteen posts breaks out, a weekly winner needs about two publishes per day — not three a week. Volume is not vanity; it is the denominator of your power law. The creators who appear to have a "knack" for going viral are almost always shipping 3–5x what their peers ship and selectively highlighting the winners.

Be variance-seeking, not consistency-seeking. Mid-tier content is the cost of buying lottery tickets, and the goal is to make each ticket cheap enough to print at scale, not to polish every ticket. A 30-minute post that hits the four traits will outperform a 3-hour post that misses them. Most creators spend their hours backwards.

Post-mortem the winner, not the average. Reviewing the average post teaches you about the average post. Reviewing the breakout teaches you about the gates. The 90% are noise around your baseline; the 10% are signal about which traits actually compound on your account.

The 5-question breakout post-mortem#

Run this within 48 hours of any post that clears 5x your median reach. Output is a one-line recipe you commit to repeating three more times in the next 14 days.

1. What was the hook, verbatim? Write out the literal first two seconds of video or first seven words of text. Not a summary — the actual words.
2. What sub-24-hour event was the post adjacent to? Name the news beat, ship, chart or quote. If you cannot name one, mark "evergreen" and note the format choice instead.
3. What format choice differed from your average post? Length, aspect ratio, on-screen text, music, opening shot, caption length — list every variable that was non-standard for you.
4. What did watch-time, save rate and share rate look like vs. your last 10 posts? Pull the numbers, not a vibe. The ratio that is most above baseline is the gate the post cleared hardest, and the trait responsible.
5. What is the minimum-viable version you can ship in 72 hours? Write the next hook now, while the recipe is legible. Schedule it.

Output: a one-line recipe — "topical news hook + 9:16 talking head with on-screen captions + named-person specificity in first 2 seconds" — and three concrete next posts on the calendar.

What to stop doing#

A short list of things that feel like content strategy and are actually friction against the power law:

• A/B testing captions on dead posts. A post that failed gate one will not be rescued by a better caption. The signal was already measured.
• Polishing mid-tier content. A second editing pass on a post that is going to do 2,000 views moves it to 2,400 views. The same hour spent shipping a new attempt at the four traits has a real shot at 50,000.
• Chasing consistency-streak metrics. "I posted every day for 90 days" is a vanity metric if 88 of those posts cleared no gates. Variance, not streaks, is what produces growth.
• Treating the average as the target. Optimizing your median is optimizing the part of the distribution that does not move totals. Optimize for more breakouts; the median will follow or won't, and it will not matter.

The reframe#

The job is not to make every post good. The job is to make winners repeatable. Power law content is the structural reality of every algorithmic feed in 2026 — it is not a phase, it is not a sign that your strategy is broken, and it does not get fixed by working harder on the 90%. It gets compounded by getting more disciplined about the 10%: identifying which traits your breakouts share, post-morteming them within 48 hours, and printing three more attempts at the same recipe before the moment cools.

If you want help with the first half of that loop — spotting which sub-24-hour events have enough velocity to be worth a hook — read how to find viral content ideas and score trends against your six niche fields. If you want help with the format-native execution side, hooks that earn the first three seconds covers the twelve patterns we see most often in the top 10%.

One post will out-earn the rest combined. That is the rule. The work is making sure the one is on purpose.