Blog

Notes from a fact-driven publisher.

Perspectives on fact-driven publishing, search, and how data-driven content earns citations from AI answer engines.

A single upright report card with an orange bar chart standing among dozens of plain gray documents scattered flat on a white surface

AI Search Sep 15, 2026 Josh McWilliam

Why Answer Engines Cite Data-Driven Content (and How to Become the Source)

AI assistants don't cite opinions. They cite the pages that supply facts they can verify. Here is what that means for anyone who publishes for a living — and why content generated from data has a structural head start.

A long row of gray blocks shrinking from tall to tiny across a white surface, with one small orange block standing out far down the tail

Search Strategy Sep 14, 2026 Josh McWilliam

The Long Tail Is Where the Answers Are

Almost 93% of search queries are asked fewer than ten times a month, 15% have never been asked before, and AI answers appear on nearly half of seven-word searches but under a tenth of one-word searches. Those facts describe one opportunity: the specific, conversational questions people now ask assistants are the long tail — and only a system that covers a topic completely can answer them.

Five small crawler tokens stopped at a concrete wall marked with orange X marks, while one orange lane passes through a gap to a row of documents

Publishing in the Age of AI Sep 13, 2026 Josh McWilliam

Should Publishers Block AI Crawlers? What the robots.txt Data Says About the Trade

79% of top news sites block AI training bots — but 71% also block the retrieval bots that decide who gets cited, and a growing share of AI requests ignore robots.txt anyway. We separate training crawlers from answer crawlers, tally who blocks what, and lay out the actual trade: a licensing negotiation you may never get, versus a citation you can measure.

A single document on the left fanning orange ribbons out to a scattered cluster of small cards on the right

AI Search Sep 12, 2026 Josh McWilliam

Where the Citations Go: Ranking and Being Cited Are Coming Apart

A year ago, three in four pages cited by Google's AI Overviews ranked in the top ten. Now it is roughly one in three, and a third of citations come from outside the top hundred. ChatGPT leans on Wikipedia, Perplexity on Reddit, and only about one domain in nine is cited by both. The data on the new rules of attention.

A rising orange path of document steps leading to an arrow, beside a cliff edge where older documents have crumbled and grown moss

Publishing in the Age of AI Sep 10, 2026 Josh McWilliam

The Evergreen Retreat Is a Mistake

Facing AI summaries, publishers plan to cut service journalism and evergreen content and double down on paid content and events. The instinct is understandable and, we think, backwards: the content AI can replace is thin evergreen. Evergreen built from data — comparisons, rankings, entity depth — is the content assistants must cite. The numbers on what survives.

A search bar above a row of gray bars that shrink from left to right, with an orange arrow tracing the decline downward

Decline of Organic Search Sep 8, 2026 Josh McWilliam

The State of Organic Search Traffic for Publishers, 2026

Search referrals to news sites fell a third in a year. Clicks halve when an AI answer appears. Publishers expect to lose another 43% within three years. We assembled every credible measurement from the last fourteen months into one report — and asked which pages are still being clicked and cited.

A crumpled pile of gray papers on the left, a single verified report with an orange bar chart raised on a pedestal in the center, and a neat stack of documents on the right

AI Search Aug 25, 2026 Josh McWilliam

What Makes a Fact Citable? The Anatomy of a Statistic an AI Assistant Will Repeat

A peer-reviewed benchmark found that adding verifiable statistics lifted a page's visibility in generative engines by a third. But not every number qualifies. We break a citable fact into its eight parts and show the same statistic written to be ignored and written to be cited.

A stack of four gray data modules emitting straight orange streams through a gate, which fan out into a burst of ordered documents

Content Quality Aug 18, 2026 Josh McWilliam

Data-Generated Is Not Generative: The Distinction That Decides Whether You Get Trusted

Two systems can each publish a million pages a year. One predicts plausible sentences; the other renders verified facts through templates people wrote. Search engines, answer engines, and readers are getting better at telling them apart — and the evidence shows up in engagement, in citations, and in what Google penalizes.

A tall narrow stack of cards spiking on the left beside a long orange path lined with evenly spaced document stacks rising gently toward an arrow

Media Strategy Aug 11, 2026 Josh McWilliam

News Wins the Day. Evergreen Wins the Year.

The median article earns 90% of its lifetime traffic in 2.6 days, while the average page ranking first on Google is five years old. Strong media brands are winning headlines and losing the evergreen queries beneath them to sites with a fraction of their authority. The gap, measured — and how news and evergreen content reinforce each other when built together.

A stone flywheel ringed by four stacks of cards and orange arrows, turning beside a rising orange bar chart

Audience Development Aug 4, 2026 Josh McWilliam

Rankings Are a Flywheel: How Share-Worthy Data Replaces Link Building

Two-thirds of web pages have no backlinks, and referring domains remain the strongest ranking correlate there is. Rankings solve both problems at once: the ranked have a reason to share, the local press has a reason to write, and each release compounds the last. The mechanics, the methodology rules that make rankings defensible, and what one release looks like at scale.

Put the engine behind your brand.

See how we partner with brands and media companies.

Partner Services Get in Touch