Why naming is a second move now, what the eighteen-month cold-start tax actually costs, and what a night watchman in Kansas had figured out by 1980

A questions and answers briefing. Every figure dated and attributed.

Section One: The Advice, and the Year It Quietly Stopped Working

1. What is the standard advice, and who wrote it?

Al Ries and Jack Trout gave us the law of categories: if you cannot be first in a category, invent one you can be first in. Play Bigger put a number on it in 2016. The category king takes 76% of the market capitalization and everybody else divides the remaining 24%. The firm tracked 35 category kings whose combined market cap grew from $465 billion in 2014 to $1.92 trillion in 2021, a 22.46% compound annual rate. The book sold north of 100,000 copies and the 76% figure reached Harvard Business Review.

That advice suited the world it was written in: analysts publishing taxonomies on a schedule, trade press with subscribers, and ad budgets that could rent a market’s attention for as long as it took to teach it a new word.

2. So is the 76% number wrong?

The number is fine. The use of it is the problem. It is a survivorship statistic describing what happens to kings after the coronation, and it says nothing about the odds of getting crowned. The figure that belongs beside it rarely appears beside it: roughly 70% of SaaS unicorns never created a category. They arrived as the second or third wave answer to a problem somebody else had already named. Zoom is a multibillion-dollar company living inside the words “video conferencing.” Play Bigger itself says category creation is not for everyone.

3. What actually changed between 2016 and now?

Retrieval changed. In 2016 a buyer typed a word into a box and got ten links. A brand new word returned nothing useful, which was survivable, because the buyer would then call a peer and the word got a second chance at a human. In 2026 the buyer asks a machine, the machine runs a live lookup, and it answers out of whatever comes back.

The scale is not in dispute. 94% of B2B buyers used generative AI somewhere in the purchase process (6sense, n≈4,510). 68% of US Google searches ended with no click in early 2026, up from 60.4% in 2024 (SparkToro with Similarweb). ChatGPT hit 900 million weekly actives in February 2026, double the year before. Semrush measured 93% of Google AI Mode sessions ending with no click at all.

4. Why does a coined word fail specifically in that environment?

A retrieval system needs something to retrieve. Ask an assistant about “revenue orchestration” and it finds hundreds of pages written by people who work for no vendor in particular. Ask about a phrase minted six weeks ago in a positioning workshop and it finds a homepage, a press release and a LinkedIn post. So it does the sensible thing: it maps the unfamiliar phrase to the nearest familiar concept and answers about that instead.

The nearest familiar concept is your competitor’s category. You have paid a consultancy to make a machine describe your rival.

The naming space is crowded and churning. Scott Brinker’s May 2026 Marketing Technology Landscape counted 15,505 products, up 0.79%, with 1,488 added and 1,367 removed. Fifteen thousand vendors are hunting for a word. The good ones are taken and the rest are unlit.

Section Two: Where the Answers Actually Come From

5. If I publish forty pages explaining my new category, does that fix it?

Your own pages are a minority shareholder in the answer. The evidence is consistent across independent studies with large samples, most of them collected in “You Lost the RFP Before Anybody Wrote It” at 1000startups.com.

Finding	Figure	Source and sample
AI citations coming from earned media rather than brand-owned pages	84%	Muck Rack, May 2026
Branded-query citations to reviews, listicles and forums (12% to product pages)	57%	Omniscient Digital, 23,387 citations
“Best X” listicles as a share of page types ChatGPT cites	43.8%	Ahrefs, 26,283 source URLs
Commercial-investigation answers leaning on at least one list post	100%	AIVO Research, 138 queries
Overlap between ChatGPT citations and Google’s top organic results	8%	Ahrefs Brand Radar, 15,000 prompts

TABLE 1 | You cannot write the list you need. Somebody who does not work for you has to write it.

6. What is a “retrieval surface,” in plain terms?

Every place your word appears that you do not own: review platform category pages, analyst taxonomies, roundups, job listings, conference tracks, forum threads. Domains absent from G2, Capterra and Trustpilot average 1.8 citations; listed domains average 4.6 to 6.3, and G2 alone commands roughly 84% of review-platform citations. A category name with none of that behind it is a word in a deck.

7. How long does that surface take to build, and what is the currency?

About eighteen months, payable in third-party coverage. Two independent lines of evidence land on that figure from opposite directions, which usually means a number is real.

Stage	Elapsed	What the evidence says
Page indexed and first appearing in retrieval	2 to 8 weeks	Comparison pages often move inside 14 days
Citations begin to decay	4.5 weeks	Median half-life across 3.5M citation events; ChatGPT fastest at 3.4
Review platforms and analyst taxonomies absorb the term	2 to 4 quarters	Listing cycles, not ad cycles. A cited URL has ~15% odds of surviving 3 weeks
Market education before pipeline materializes	18 months	Category-design practitioners, at 2 to 3x normal content spend
Full category payoff	3 to 5 years	Play Bigger puts the IPO sweet spot at 6 to 10 years of age

TABLE 2 | The cold-start tax. The bill arrives in writer-weeks and listings, and media spend does not accelerate it.

Section Three: The Night Watchman Who Waited Three Years to Name It

8. Why is a baseball statistician in a positioning article?

Because Bill James ran the experiment and left receipts. He built a genuinely new category of thought, named it himself, and the sequence is recoverable year by year with unit sales attached. Few category creation stories come with a ledger.

9. What did he do first?

In 1977, working night shifts as a watchman at a Stokely-Van Camp pork and beans cannery in Lawrence, Kansas, James self-published 68 mimeographed pages held together with staples: 1977 Baseball Abstract: Featuring 18 Categories of Statistical Information That You Just Can’t Find Anywhere Else. He bought a one-inch ad in the back of The Sporting News and sold 75 copies at about $3.50 each. Norman Mailer ordered one. So did the screenwriter William Goldman.

Read that title again. Baseball. Statistics. Categories. Information. Every word already lived in his buyer’s vocabulary. He spent four years describing radical work in language a Sporting News subscriber could already search for inside his own head.

10. When did he coin “sabermetrics,” and what did he name it after?

1980, three years and four editions in. He named it for SABR, the Society for American Baseball Research, founded 1971, an organization nine years old at the time with members, a journal and a mailing list. James hung his coinage on something that already had a footprint. In 2026 vocabulary, he attached a new term to an existing retrieval surface, using a typewriter. Anybody naming a category this quarter should answer the same question: what existing thing does my word attach to?

11. What actually made the word travel?

Somebody else wrote about him. Daniel Okrent bought the 1978 Abstract off that Sporting News ad and struck up a correspondence. His profile, “He Does It by the Numbers,” ran in Sports Illustrated on May 25, 1981, after two years stuck in editorial politics. James’s cumulative sales stood near 2,600 copies. That piece won him a Ballantine Books contract, and the distributed Abstract sold more than 100,000.

Year	What happened	Copies	Whose words
1977	Self-published, mimeographed, one-inch ad in The Sporting News	75	Existing
1978	Second edition. Okrent buys a copy off the ad	325	Existing
1979	Third edition. Still a mailing-list operation	600+	Existing
1980	James coins “sabermetrics,” after SABR, founded 1971	~750	New, attached
1981	Okrent profile runs in Sports Illustrated, May 25	~2,600 total	Third party
1982	Ballantine Books picks it up for mass distribution	100,000+	Third party
2003	Michael Lewis publishes Moneyball	Bestseller	Everybody’s

TABLE 3 | The Bill James ledger. One writer-week from an outsider in 1981 did what four years of self-publishing had not.

12. When did “sabermetrics” become a word ordinary people looked up?

2003, when Michael Lewis published Moneyball, 23 years after the coinage and 26 after the first Abstract. The term went mainstream because a bestselling author needed a word for what he was describing and found one sitting there already defined, documented and attached to a research society. James joined the Red Sox in 2003 and stayed 17 years across four World Series titles.

The cold-start tax on “sabermetrics” ran twenty-three years, and James paid it in other people’s coverage rather than his own promotion.

13. Is there a shorter version of the same lesson inside baseball?

WAR. Wins Above Replacement won its argument the day Baseball-Reference and FanGraphs each put it in a column on every player page. Once a metric lives in a table other people maintain and everybody consults, it is retrievable, and the debate is over. Your category needs a column on somebody else’s page. That is the assignment.

Section Four: What to Do on Monday

14. So never create a category? And what does the alternative look like in practice?

Sequence it. Win the category everybody already searches, get described by outsiders in existing language until you are the reflexive answer inside it, then rename from a position where the rename is itself news. Naming from strength gets written about. Naming from obscurity produces a word with no retrieval surface. Seven moves:

1. Audit prompts, not keywords. Run the twenty questions a buyer types and log every source retrieved. 33% open with a category search, 31% with a competitor search.

2. Use the existing term as the container. Put your difference inside it as a modifier. “Actuarial-grade pricing analytics” retrieves. “Pricing intelligence orchestration” does not.

3. Get listed before the quarter you need it. Absent from review platforms averages 1.8 citations. Listed averages 4.6 to 6.3.

4. Get covered by somebody else. 84% of citations are earned. Your own roundup with you at number one is a beige suit in a pastel show.

5. Load the pages with numbers. Pages with 19 or more data points averaged 5.4 citations against 2.8 for thin ones. A Princeton-led KDD 2024 study found statistics and citations each lifted AI visibility up to 40%.

6. Refresh on a calendar. Content updated within 90 days carries a 3.2x citation multiplier against a 4.5-week median half-life.

7. Change the metric on the wall. Report share of answer and criterion presence. Leads lag the decision by months.

15. How do I know I have earned the right to name it?

Call ten people who know your company and are not on your payroll, and ask what you do. If the answer comes back in your language unprompted, the market already carries the words and a name ratifies what exists. If it comes back in the incumbent’s language, coverage repairs that and a new name will not. The test comes from “The Telephone Game” at 1000startups.com, which argues your real position is whatever your least-informed advocate says over dinner.

16. What is the strongest argument against all of this?

Two, and the second is good. Genuinely novel technology sometimes has no honest container, and jamming it into one gets you scored against the wrong criteria. More interesting: a coined term faces almost no competition in retrieval, so once a handful of third-party pages carry it, the machine has one obvious answer and it is you. That edge is real, and it concedes the point, since the coined term only pays after third parties adopt it. The coverage work is mandatory either way. The argument is about order, and about who funds the eighteen months of silence.

17. What happened to B2B software’s most famous category creator?

Drift coined “conversational marketing” across 2016 and 2017, published the book, ran the conference, raised more than $100 million, served over 50,000 businesses, and sold to Vista at a $1 billion valuation in 2021. Salesloft acquired it in February 2024 for a reported $500 million. On March 6, 2026, Clari and Salesloft announced Drift’s sunset. Founder David Cancel has publicly discussed the failure of the follow-up category, “revenue acceleration.”

Meanwhile buyers in that market type “AI SDR,” an acronym for a job that existed before the software did. The vendors ranked against that phrase are getting recommended this afternoon, in conversations no marketing team will ever see.

Drift got the trophy. The boring words got the traffic.

18. One sentence for the board?

Rent the category everybody already searches until enough outsiders describe you inside it, then buy the naming rights with the coverage you accumulated, because in a retrieval market a word with no third-party pages behind it is indistinguishable from silence.

Sources

1. Penland, “You Lost the RFP Before Anybody Wrote It,” 1000 Startups, Sept. 4, 2026. Retrieval figures, citation half-life, review multipliers. 1000startups.com/2026/09/04/you-lost-the-rfp-before-anybody-wrote-it/  ·  2. Penland, “The Telephone Game,” 1000 Startups, Sept. 5, 2026. The ten-calls test. 1000startups.com/2026/09/05/the-telephone-game…  ·  3. Play Bigger, “Category Contenders.” The 76% figure and 35-king cohort. playbigger.com/media/category-contenders  ·  4. Growigami, “Category Creation,” March 2026. Eighteen-month education estimate. growigami.com/glossary/category-creation  ·  5. SABR, Henry Chadwick Award citation for Okrent. The 75-to-2,600 sales run. sabr.org/journal/article/henry-chadwick-award-daniel-okrent/  ·  6. Britannica, “Sabermetrics,” plus Smithsonian Lemelson Center, “Sports Analytics Before Moneyball.” britannica.com/sports/sabermetrics  ·  7. Salesloft, “Salesloft Acquires Drift,” Feb. 2024, plus 2026 sunset coverage. salesloft.com/company/newsroom/salesloft-acquires-drift  ·  8. Brinker, 2026 Marketing Technology Landscape, May 2026. chiefmartec.com  ·  9. Aggregated 2026 AI search data: SparkToro zero-click, Semrush AI Mode, Muck Rack, Ahrefs, Princeton KDD 2024. contextbolt.com/ai-search-statistics/

More at 1000 Startups: 1000startups.com/category/ai-search-visibility  |  1000startups.com/category/positioning-messaging