Tag Archives: advertising

Own the Dictionary: Build the Glossary Nobody Else Bothered to Write

Leave a reply

Everybody wants to write the definitive guide. Almost nobody wants to write the definition. That gap is the cheapest authority play left on the internet, and it is sitting there unclaimed like a live straddle nobody noticed.

  1. Definitions are the most retrievable prose there is. Roughly 70% of Google’s featured snippets are paragraph snippets – the definition box – and Semrush pegs the winning ones at 40 to 60 words. Sixty words. You have written longer text messages declining a dinner invitation.
  2. Nobody owns the vocabulary of your niche. There is no rights-holder for “polarized range” or “minimum defense frequency.” The dictionary of your industry is unsurveyed land, and the deed goes to whoever bothers to walk it first.
  3. Forty terms, not four. Ahrefs found a single article holding 4,658 featured snippets across the keyword universe. Forty entries is forty front doors, each catching phrasings you would never have thought to target on purpose.
  4. The long tail is where definitions live. Backlinko found 54% of featured snippets come from searches doing under 50 queries a month. Semrush found 55.5% of ten-word queries trigger a snippet versus 4.3% of one-word queries. Nobody wins “poker.” Anybody can win “what is a donk bet and why is it called that.”
  5. Write the disagreement – that is the entire moat. Any intern can type “a donk bet is a lead into the previous street’s aggressor.” The version that gets quoted for a decade adds that purists restrict it to the flop, that it was coined as an insult, and that solvers have since made it standard – meaning half the internet is using a slur to describe correct play.
  6. Three more you can steal tonight. “GTO” strictly means a Nash equilibrium in a heads-up zero-sum game, and its guarantees quietly evaporate three-handed – yet it gets used daily to mean “whatever the solver said.” Sklansky defined a semi-bluff as a bet that is probably not best but can improve; today people call king-high a semi-bluff. And a “merged” range means value-plus-marginal to one camp and close to the opposite to another. Say so. Cite both. Now you are the referee.
  7. Front-load or forfeit. About 44.2% of AI citations are pulled from the first 30% of a page. Term, one clean sentence, then the nuance. Bury the answer and you are bluffing into a machine that folds instantly and never pays you off.
  8. Mark it up. Schema-tagged pages are cited roughly 2.3x more often in AI Overviews. Definition schema takes fifteen minutes. That is the best price-to-equity ratio on this page.
  9. The door is open right now. Ahrefs’ March 2026 data shows only 38% of AI Overview citations come from top-10 organic results, down from 76% a year earlier. You no longer have to outrank Wikipedia. You have to out-define it – and Wikipedia is not allowed to have an opinion about which definition is better. You are.

Guides get skimmed and forgotten. A glossary gets bookmarked, linked, argued with, and quoted – by people and by models – for years. Write forty definitions, note who disagrees and why, and stop playing the hand. Become the house.

Drop the AI Prefix

Leave a reply

Why “AI-powered” stopped being a magnet in 2026 and started working like a coupon.

1. The label now costs you money. Researchers at Washington State and Temple put the same products in front of 1,000 people and changed exactly one thing: whether the description said “artificial intelligence.” Across eight product and service categories, purchase intent fell every single time. It fell hardest where the stakes were highest, a shrug for a television, a wince for a car or a medical diagnostic. Parks Associates asked roughly 4,000 Americans a similar question: 18% said AI made them more likely to buy, 24% said less likely, and 58% said it changed nothing. You are spending the best three seconds of your homepage on a word that repels more people than it attracts and bores the majority.

2. We have seen this movie. It came out in 1998. Cooper, Dimitrov and Rau published the autopsy in the Journal of Finance: companies that simply bolted “dot.com” onto their names earned about 74% abnormal returns in the ten days after the announcement. No new product. No new customers. A new suffix. Books-A-Million announced a redesigned website the day before Thanksgiving 1998 and watched its stock run from $3.06 to an intraday high of $47. Two weeks later it was $10. By 2000 it was back around $3.60. Same bookstore the entire time.

3. The real tell is the deletion trade. Here is the part everyone forgets. When the bubble burst, the same researchers tracked firms that stripped “dot.com” back out of their names and found cumulative abnormal returns of roughly 64% over the sixty days around the announcement. The word paid 74% going on and 64% coming off. A label that is valuable in both directions is not an asset. It is a costume, and eventually the audience notices the zipper.

4. The prefix is now a legal exposure, not a marketing one. The SEC has a name for it, “AI washing,” and a growing docket. Enforcement opened in March 2024 against two advisory firms and reached its first public company in early 2025: Presto Automation, whose drive-thru voice AI turned out to belong to a third party and to lean heavily on humans. The founder of Nate Inc. was charged over more than $42 million raised on claims of an AI shopping app. Six named cases since March 2024, over $44 million in alleged fraud, and securities class actions over AI claims roughly doubled from 2023 to 2024. “Everybody says it” has never been a legal defense.

5. The buyer already assumes it, and has already been burned. In 2019 MMC Ventures reviewed 2,830 European “AI startups” and found no material evidence of AI in about 40% of them, even as the AI label pulled 15% to 50% more funding than plain software. That arbitrage is closed. MIT’s 2025 enterprise study weighed $30 to $40 billion in spending against a blunt finding: 95% of generative AI pilots produced no measurable impact on profit and loss. Your prospect has a dead pilot in a drawer. “AI-powered” no longer excites her. It starts an interrogation.

6. So name the job you finish. “AI-powered claims processing” describes your plumbing. “We close claims in four hours instead of nine days, and you pay only for closed claims” describes her Tuesday. Run the ninety-second audit: delete every instance of “AI” from your homepage. If what remains still names a verb, a number and a deadline, you had positioning. If the page goes blank, you had a costume.

The bottom line. Nobody bought a book in 2004 because the store was “internet-enabled.” They bought it because it showed up in two days. The prefix was never the product, and in 2026 it is not even the hook. Tell me what you finish, how fast, and what it costs. I will assume the robots.

Guest, Do Not Host: Twenty borrowed audiences beat one built from scratch

Leave a reply

Guest, Do Not Host

Twenty borrowed audiences beat one built from scratch. Disco figured this out in 1977.

1. The math nobody runs before buying a microphone. There are roughly 4.7 million podcasts registered worldwide and about 407,000 still publishing anything at all. That is one in fifteen. Of the shows launched in the first half of 2026, 42% were already dead within six months. Only 7% of active shows ever reach 10,000 listeners, and exactly 1,381 crack a million. Starting a show is not entering a market. It is joining a cemetery with better cover art.

2. Twenty doors, an afternoon each. A weekly show costs you 52 Tuesdays, an editor, and a year of talking to nobody. Twenty guest spots cost maybe fifteen hours, and each one hands you a crowd some stranger spent five years assembling, plus the host’s endorsement, which you cannot buy and cannot fake. You are not renting attention. You are being vouched for.

3. Studio 54 turned Chic away. Chic did not open a nightclub. New Year’s Eve 1977: Nile Rodgers and Bernard Edwards are stuck on the sidewalk because Grace Jones forgot to leave their names at the door. They went back to an apartment, furious, and wrote “Le Freak.” Seven million copies. Still the best-selling single in the history of Atlantic Records. The lesson is in what they did not do. They did not build a rival club. They built something that got played inside every club that already existed.

4. Rodgers made a forty-year career out of other people’s records. Chic never had a No. 1 album. His guest work did: Bowie’s Let’s Dance (about 10 million), Madonna’s Like a Virgin (21 million and counting), Sister Sledge, Diana Ross, Duran Duran, and three decades later, Daft Punk. He kept showing up on other people’s stages, and a little piece of every audience became his.

5. The Bee Gees were finished until they took a job on somebody else’s movie. Mid-seventies, career in the ditch. Robert Stigwood asked for a few songs for a small film about kids dancing in Brooklyn. They wrote most of it in about two and a half weeks in a French château, having never read the script. Saturday Night Fever sold 40 million-plus copies and held No. 1 for 24 straight weeks, the best-selling album on the planet until Thriller. They did not produce the movie. They got on it.

6. The transcript is the asset. The episode is just the excuse. Every appearance gets transcribed, indexed, and quoted back for years by search engines and AI assistants that never once ask what your download numbers were. Otterly’s 2026 study of 100 million citations found 40.8% of AI-cited videos had under 1,000 views. Small rooms, permanent record.

7. Hosts are starving, and you are lunch. Roughly 480,000 active shows need a guest every single week, forever. That is not a competitive market, that is supply desperation. So skip the burnished bio. Lead with one specific claim you will defend on tape, one number nobody else has, and one story where you come off badly. Bookers hire tension, not résumés.

The doorman always wins eventually. July 12, 1979, Comiskey Park: 47,795 people crammed into a stadium built for 44,492 to watch a radio DJ detonate a crate of disco records between games. The White Sox forfeited the nightcap. A billion-dollar genre evaporated in one evening, and everyone who owned a disco went down with the building. The session players, the guests and the writers went to work the next morning like nothing happened. Own nothing. Appear everywhere.

Teach, Don’t Speak: Why your next conference slot should be a workshop

Leave a reply

Why your next conference slot should be a workshop – and what NFL coaching staffs figured out decades ago

A talk ends in applause. A workshop ends in a deliverable. Only one of those changes the phone call you get on Tuesday.

1. A speech is the broadcast. A workshop is practice.

The Wall Street Journal once stopwatched an NFL game: 3 hours and 12 minutes of programming wrapped around roughly 11 minutes of live football. Nobody has ever won during the other three hours. Bill Walsh scripted his first 15 to 25 plays so that by Sunday they felt like a rerun of Wednesday. Your keynote is the broadcast. Be the practice.

2. The research is not a close call.

A 2014 PNAS meta-analysis pooled 225 studies. Failure rate under traditional lecture: 33.8%. Under active learning: 21.8%. Exam scores rose about 6%, and lecture students were roughly 1.5 times likelier to fail. (Note whose numbers those are. The “we remember 90% of what we do” pyramid on LinkedIn has never been traced to an actual study.)

3. A good workshop feels worse in the room. Run it anyway.

Harvard physicists tested this in 2019: students in active classrooms scored measurably higher and rated their own learning lower. Polish produces the feeling of learning; struggle produces the fact of it. Highlight reels feel fantastic. Walk-throughs feel like chores. December decides which one mattered.

4. Scarcity is what forces you to actually teach.

The CBA allows NFL teams just 14 padded practices all season, 11 of them in the first 11 weeks. Coaches who once ran three-hour hitting sessions now install concepts in walk-throughs and meeting rooms – and the constraint made them better teachers, not worse. Give yourself 45 minutes instead of 60. Cut content, keep reps.

5. Their numbers, not your demo file.

A demo on your clean sample data is a magic trick: impressive, entirely non-transferable. The same method run on their messy Q3 export is a skill they now own. That one swap rewrites the follow-up – instead of “interesting, send the deck,” you get “what do I do about row 14?” Players study their own film, not just the coordinator’s cut-ups.

6. If they can leave empty-handed, you gave a speech with the chairs moved.

Put the deliverable on slide one and work backwards, ruthlessly: one filled-in model, one finished template, one draft they would be annoyed to lose. Everything that doesn’t survive contact with that goal is a story you can tell at dinner instead.

The Bottom Line

Lombardi opened camp holding a football: “Gentlemen, this is a football.” That wasn’t a keynote. That was the first rep. Stop performing your expertise and start handing it over – someone who has already run your method on their own numbers, in front of you, is not an audience member anymore. They’re a user. Users call back.

Answer in Spanish: The Second Court Nobody Is Playing

Leave a reply

Your English authority stops at the net. In U.S. insurance distribution, the other court is empty.

1. Clay is not grass. Rafael Nadal won Roland Garros fourteen times. That bought him zero free points at Wimbledon, where he won twice. Same forehand, different surface, different bounce, different everything. Language models work the same way: each language is a partially separate knowledge space. Ranking #1 in English does not seed you in the Spanish draw. Nobody carries a ranking across the net for free. Surfaces are learned.

2. The 97-to-19 problem. Researchers taught a model new facts in English. Quizzed in English, it recalled them 97% of the time. Quizzed on the identical facts in another language, accuracy fell to 19%. Same model, same facts, one language apart. A second study estimates perfect cross-language sharing could lift accuracy up to 150% – a polite academic way of saying it does not happen today.

3. Untranslated means invisible. If your annuity explainer and licensing FAQ exist only in English, a Spanish query does not rank you low. It does not see you. You are not losing the point; you never entered the tournament. English is 49.7% of identifiable web content, Spanish just 6.0% (W3Techs, June 2026).

4. The crowd is already in the stadium. 44.9 million U.S. residents speak Spanish at home (2024 ACS) – 1 in 7 people age five and up. That group grew 21.3% from 2010 to 2024 while the population grew 11.2%. And 41% say they speak English less than “very well.” They are asking chatbots about deductibles, beneficiaries and IUL right now, in Spanish.

5. The number that should sting. Hispanic life insurance ownership fell from 51% in 2021 to 40% in 2025, the lowest of any group LIMRA tracks. Roughly 20 million Hispanic adults say they are underinsured, and 72% overestimate what term costs. That is not a demand problem – the demand is already there. It is an answer-supply problem, and the answers are sitting in the wrong language.

6. Do not practice against a ball machine. Auto-translating your English pages is hitting against a ball machine and calling it a match. “Rider,” “surrender charge,” “face amount” – rendered literally, they land as nonsense that no Spanish speaker and no model treats as authoritative. Write it natively, in the vocabulary agents actually use in Miami and Houston.

7. Serve first. Carlos Alcaraz did not inherit a grass game. He built one on purpose, then won Wimbledon in 2023 and 2024. The Spanish-language retrieval surface in insurance is wide open and lightly defended. Every month you leave it untranslated is a free point handed to whoever translates first.

The Referee Advantage

Leave a reply

Why the Company That Keeps Score Never Has to Rent an Audience

Fifteen years of financial startups that quit buying attention and started measuring the game.

WHAT IT COSTS TO RENT A CROWD

Chime spent $519.7M on sales and marketing in 2024 – 35% of gross profit – for 1.4 million new members: $109 a head by its S-1, $91 without brand ads, $371 with salaries and stock comp. PNC Bank: $337 million, 2% of revenue, 2,200 branches.

It also paid $33 million over three years for Dallas Mavericks jerseys. Super Bowl LX hit a record $8 million per thirty seconds ($10 million for prime slots); eleven fintech brands paid it. Robinhood pays customers instead: a 3% IRA match for Gold members, an uncapped 2% bonus on transfers and 401(k) rollovers in 2026.

All of it works – while you keep writing checks. Anish Acharya of a16z named the problem in 2020: “increasingly generic product experiences and rising acquisition costs.” It is inventory any competitor can buy at the same price. Stop paying and it is gone. You rented a crowd and the crowd went home.

THE REFEREE’S JOB

Never charge a problem head on; work the edges while everyone else runs at the same wall. When everyone bids on the same thirty seconds, the winner is whoever found a door nobody guarded. One seat in the stadium is free: the scorekeeper’s – unpaid, unwatched, and read by everybody.

The move: find the question your market argues about and cannot settle. Answer it with a number only your systems produce. Name it so a normal person will say it out loud. Publish it the same date every period, free, forever.

It compounds: year one, a forgettable press release; year five, people chart against you; past ten, a citation – and citations do not churn. Nobody outbids you for a ten-year series. Peter Thiel’s “competition is for losers,” run by one analyst with a calendar invite.

A billboard expires with the invoice. A scoreboard ends up in a Federal Reserve working paper.

STEP ONE – REFEREE A FIGHT PEOPLE ARE ALREADY HAVING

“Purchasing power parity” is a subject. “Is your currency overpriced?” is a fight. Nobody forwards a subject. Every market has an argument that recurs and never settles because nobody collected the data. That argument is your product.

1. The Big Mac Index (1986). The Economist’s Pam Woodall priced currencies with a hamburger on September 6, across 13 countries. Same recipe in roughly 120 countries makes it a control variable. Now in textbooks, academic papers, IMF and central-bank valuation debates – and it coined burgernomics.

2. Truflation (2022– ). Is CPI honest, and why wait a month? A $100,000 bounty project became a daily read from 13 million+ data points across 30+ sources, claiming a 45-day lead on the BLS. It reached the Bloomberg Terminal in early 2026; then the 37-day shutdown stopped official data and proved the point.

3. Ramp (2023– ). Others asked companies if they used AI; Ramp measured what 70,000+ businesses paid for. Census Bureau: 20%. Ramp’s cards: 46.6%. A startup corrected the federal statistical system – the New York Times, Wall Street Journal, Financial Times, Bloomberg and NPR’s Planet Money all ran it. It bought a Super Bowl ad too; which gets cited in 2030?

STEP TWO – A SCOREBOARD READABLE FROM THE CHEAP SEATS

Two questions: can a producer fit it in a chyron, and will a normal person repeat it at a bar? “Eleven pipers piping” passes. “Core CPI ex-shelter” fails. For a number nobody must read, the name is most of the distribution.

4. PNC’s Christmas Price Index (1984– ). A Pittsburgh bank pricing a carol, fifth decade running. 42nd edition: $51,476.12 for twelve gifts, up 4.5% against a BLS October CPI of 3.0%; all 364 gifts, $218,542.98. Five gold rings up 32.5%, pear tree up 14.3%, plus a straight-faced “core” index excluding swans.

5. Toast (2023– ). Average tip from roughly 171,000 locations, quarterly. Q1 2026: 19.3% full-service, 15.8% quick-service; Delaware highest at 22.1%, California lowest at 17.3%. Axios covers the state ranking nationally, then locally; LendingTree analyzes it unasked. No rival tipping figure exists to rebut it.

6. Robinhood’s Investor Index (2022– ). Outside sites scraped its most-owned holdings for years; shutting the API in 2020 killed the coverage along with the leak. On September 9, 2022 it rebuilt the insight as its own: 20 million+ funded accounts, weighted by conviction, not dollars. Bloomberg covered it that day. If strangers will steal your data, publish it yourself.

STEP THREE – LOSE CONTROL ON PURPOSE

You are not trying to keep the number. You are trying to get it stolen. The day somebody quotes your figure with no idea your company exists is the day it started working.

7. Case-Shiller (1980s). Two economists built a repeat-sales housing index for an honest read on home prices. On May 16, 2006 the CME listed futures at $250 per index point – $62,500 notional per contract at 250. Shiller took the Nobel in 2013. The futures flopped (7,579 contracts through September 2008), but the number got away clean. Getting away is the win.

8. Homebase (2020– ). A free scheduling app published what its dashboard knew: which small businesses were open and hours actually worked – the most valuable data in America in spring 2020. The St. Louis Fed built a labor-market coincident index on it (60,000 businesses, ~1 million hourly employees); Drexel and Chicago Booth published papers; CNBC used it. A monthly report then brought 300+ pieces of earned coverage and the Wall Street Journal front page. Media budget: zero.

How it fails: publishing on schedule with nothing to say. Twice and it is a chore; reporters stop opening the email. Find the slice inside the flat number, or skip the month and say why.

STEP FOUR – SHARE THE STADIUM

You rarely own both halves – data without an audience, an audience without credibility, credibility with half a dataset. Do not buy the missing half. Find whoever has it and has an equally embarrassing gap.

9. WWF and MTV (1984). The WWF could not buy teenagers; MTV could not make live drama. On July 23, “The Brawl to End It All” aired from Madison Square Garden: 23,416 in the building, one match, a 9.0 Nielsen – MTV’s highest-rated program to that point. Eight months later: WrestleMania.

10. Plaid and the Harris Poll (2020– ). Plaid borrowed credibility, not eyeballs. The 2021 Fintech Effect survey put U.S. adoption at 58% to 88% in a single year. Six editions later, still the default citation.

11. Ramp and Revelio Labs (2026). Ramp borrowed a dataset, matching spending records to Revelio’s workforce data across about 21,500 companies: heavy AI adopters grew headcount roughly 10%, entry-level hiring about 12% – against the story that AI deletes junior jobs. Neither had the picture alone.

THE FIFTEEN-YEAR LEDGER

Primary Venture Partners noted in 2026 that the largest fintech businesses all have “proprietary data assets at their core.” Incomplete – plenty of companies hoard proprietary data and die anyway. Publishing it is what turns it into a company.

12. Credit Karma (2007– ). It gave away the number itself, back when credit scores cost money – then the full report, then a business built on the traffic. By 2020: 100 million+ users, 37 million monthly actives, just under $1 billion in revenue. Intuit paid $7.1 billion.

13. CB Insights (2008– ). Front door: a free newsletter and free quarterly State of Fintech report, co-branded with Money20/20 so somebody else supplies the audience. The platform behind it did an estimated $100 million in revenue by 2022. Not marketing – what the product advertises with.

14. Wise (2011– ). The fight: your bank’s hidden exchange-rate markup. Wise publishes a table pricing its own competitors, added quarterly mission reports in 2018, and put the cost to Americans at roughly $8.7 billion a year. Pricing your rivals is aggression dressed as public service.

15. Chainalysis (2015– ). Its annual Crypto Crime Report is the reference document for illicit crypto. 2026: $154 billion in illicit inflows (up 162%), sanctioned entities up 694%, stablecoins 84% of illicit volume – all still under 1% of crypto activity. It bought a $170 million Series F at an $8.6 billion valuation, 1,300 customers across 70 countries, and revenue tilted to the DoD, FBI and IRS. Rivals now benchmark against it line by line. When competitors argue with your number, you are the referee.

16. Carta (2019– ). State of Private Markets, quarterly – the down-round rate every founder checks. From a 22% peak in 2023 to 11.4% in Q1 2026, with $30.4 billion raised that quarter and 60%+ of every venture dollar going to AI. Every VC newsletter reprints the charts free, Carta’s name attached.

Not one of the sixteen bought the coverage. Each published something nobody else could compute, on a fixed schedule, free, and let reporters, academics, rivals and Federal Reserve banks distribute it. Everyone else was at the front door, bidding.

STARTING MONDAY, AND WHAT TO EXPECT

  • Referee a fight, not a subject.
  • Use data only you have. If a consultant could buy the same numbers, the moat is not real.
  • Name it so a stranger will repeat it.
  • Fix the date and never move it. Dull consistency is the strategy, not a flaw in it.
  • Publish the methodology. Credibility is cheaper than a media buy and lasts longer.
  • Give it all away. A paywall kills citations; you are selling the byline, not the data.
  • Find whoever holds the other half of the stadium. MTV had airtime, Vince had wrestlers, Plaid had Harris, Ramp had Revelio.

Then be patient: about two years to a dependable press cycle, five to a series people chart against, ten to a citation. The early signals are unglamorous – an unasked-for citation, someone analyzing your data, a competitor benchmarking against you, an institution adopting it. Nobody gets there in a quarter, which is why so few follow you through the side door – and why it stays open.

A Super Bowl spot costs $8 million and is over in thirty seconds. A scoreboard costs one analyst and a calendar, and forty years later the whole league is still checking it.

Own a Number: Why the smartest play in your market is to publish one statistic on the same day, forever

Leave a reply

Everybody is writing think pieces. Ahrefs studied 14 billion web pages: 96.55% get zero traffic from Google. Your insight isn’t competing with other insights; it’s competing with oblivion. The way out is a number with your name welded to it.

1. Pick the argument, not the topic

Every market has a question people argue about with no data to settle it. In 1986, The Economist’s Pam Woodall asked whether a currency was mispriced and answered with a hamburger. The Big Mac Index now turns up in textbooks and IMF papers, and spawned a word: burgernomics.

2. Measure it identically every year

The 2024 and 2025 Dodgers repeated by running one plan twice: stars, pitching, and the biggest checkbook in baseball. The day you “improve” your formula, your time series dies and you’re a blogger again. An imperfect number measured identically for a decade beats a perfect one measured once.

3. Publish on a fixed date

The Dodgers turn up every October and the sport schedules around them. The Edelman Trust Barometer lands every January against Davos: 26 editions, 33,938 respondents, 28 countries. Case-Shiller drops the last Tuesday monthly. Once reporters build in your date, they call you.

4. Give it a name people say out loud

PNC’s Christmas Price Index is a Pittsburgh bank pricing a carol; the 42nd edition hit $51,476.12, up 4.5% against the BLS’s 3.0%. Eleven pipers piping get on morning shows; “core CPI ex-shelter” does not.

5. Show the receipts, itemized

Nobody says the Dodgers spent a lot. They say $417.3 million payroll, a record $169.4 million luxury tax, $514.7 million all in, more than the bottom six clubs combined. PNC also publishes a “core” index stripping its volatile swans, up 6.1%. Flagging your noisiest input is why anyone believes the headline.

6. Expect it to get away from you

Two economists built a repeat-sales housing index in the 1980s. By 2006 the CME listed futures on it at $250 per index point; Shiller took a Nobel in 2013. When your number gets traded or misquoted by people who’ve never heard of you, you’ve won. The failure mode is publishing on schedule with nothing to say: October with no rotation.

7. Stop being a source, become the source

Ahrefs reports 66.5% of links created in the last nine years are dead. Posts rot; indices compound. AI systems cite content 25.7% fresher than typical top-ranking pages, which is exactly what an annually refreshed figure is.

One last data point. Yoshinobu Yamamoto started Game 6, closed Game 7 on no rest, and finished 3-0 with a 1.02 ERA. Same arm, same method. Still the number everyone quotes.