We earn citations for a living. Then we build you the same desk.

Original research from data we own, published in the open. It earns mentions, citations, and answer-engine pickup, because engines quote sources, not sites. We point the same engine at your market.

This week · Legal desk Cited

Who AI actually cites when someone asks how to hire a law firm

We sampled 236 answer-engine citations across the law-firm-hiring question set. 67% pointed at review and comparison publishers, not at any law firm or any single agency.

67% of AI citations went to
publisher/review sites2
Read the study →
From the desk1 published · 11 queued

What's on the desk right now.

One study is published and cited today. The rest of this slate is queued: the beat, the angle, and the dataset are set, dated for when the desk actually runs them. Nothing here is shown as a finding before it's earned one.

Data Studies · Legal

Who AI Actually Cites When Someone Asks How to Hire a Law Firm

n=236 sampled AI citations on the law-firm-hiring question set. 67% pointed to review and comparison publishers, not to any firm or agency.1

Published · Jul 6, 2026 Cited · DR84 →
Algorithm Watch

The May 2026 Core Update Finished June 2. Here's Who It Actually Took Down.

A dated read of the confirmed May 23–Jun 2 update window, run against real tracked positions once the desk publishes it.

Queued · on the desk Read →
Algorithm Watch

The “AI Citation Count” Everyone Is Bragging About Might Be Fake. Here's the Test.

The same gameable-vendor-vocabulary test behind our own citation study, aimed at the industry's favorite vanity number.

Queued · on the desk Read →
AI Search & Citations

The Backlink That Doesn't Exist Still Gets You Cited. Here's the Number Behind It.

The unlinked-mention correlation already anchoring our own citation study, expanded into its own explainer.

Queued · on the desk Read →
AI Search & Citations

Why Your YouTube Channel Might Matter More to ChatGPT Than Your Website Does

What the strongest single predictor of ChatGPT visibility actually measures, and why it isn't a website metric at all.

Queued · on the desk Read →
Discover & Distribution

The Headline That Ranks and the Headline That Gets Read Are Not the Same Line. Most Sites Still Write Only One.

Why an SEO title and a Discover headline do two different jobs, and what happens on the sites that only write one of them.

Queued · on the desk Read →
Discover & Distribution

One Publisher Changed Nothing but the Headline. Discover Click-Through Went From 2% to 14%.

The same-story, same-week test the desk runs before trusting any headline advice, once we've run our own version of it.

Queued · on the desk Read →
Data Studies · Mass tort

The Docket-Momentum Tracker: Which Mass-Tort Cases Are Actually Gaining Steam

A standing, monthly-refreshed read of case-count growth per active MDL, off the JPML Pending MDL Report and FDA adverse-event data.

Queued · on the desk Read →
CRO

High Impressions, Zero Clicks: The One-Line Diagnosis Most Sites Skip

The impressions-high, clicks-zero pattern, and the one-line diagnostic the desk runs before it touches a page.

Queued · on the desk Read →
Paid Media Craft

Meta's Own Five-Step Algorithm, Explained in Plain English

How ad delivery actually ranks a piece of creative, and why the industry rarely explains it past the jargon.

Queued · on the desk Read →
Attribution

Last-Touch Attribution Is Lying to You About What Actually Sold

Why the model most dashboards default to structurally undercounts the content that started the deal.

Queued · on the desk Read →
Agency Buyer Ed

In-House vs Agency Isn't a Budget Question. It's a Bandwidth Question. Here's the Actual Math.

Reframing the build-vs-buy decision buyers actually face, away from sticker price and onto real capacity.

Queued · on the desk Read →

See the full newsroom feed

The engine & the ask

The same desk, pointed at your market.

A low-commitment ladder. See the product working for free, test our honesty cheaply, then run the engine. Every rung feeds the next.

1
Free

AI Visibility Snapshot

Enter a domain, get a per-platform read: who cites you, who doesn't, mention rate and link rate as two numbers.

Start free
2
Paid wedge

AI Visibility & Market Position Audit

A fixed-fee, dated, sourced read of your true starting position, frozen as the benchmark every later claim is scored against. The fee credits into the engagement if you continue.

Get the audit
3
Doors

Productized service pages

Law Firm SEO, Programmatic Advertising, Generative Engine Optimization and more. Each ranks and converts on its own, and feeds the flagship.

See the doors
4
Flagship

Full-Service Authority Engagement

The owned-data newsroom engine, with an independent measurement firewall, holdout-verified billing, and a public results ledger written into the contract.

See the ledger

Start with the honest read. It's the whole product in miniature.

Sources & method

  1. Gamed-citation case: one tracked site logged 31,000–37,000 "AI citations" against 75 human clicks, almost entirely off one boilerplate word. Documented public case, read Jul 6, 2026.
  2. Legal citation study: n=236 sampled AI citations on the law-firm-hiring question set; 67% pointed to review and comparison publishers, not to any law firm or agency. Our legal desk, Jul 6, 2026; earned pickup: Search Engine Roundtable (DR84). See the full study.
  3. Unlinked brand mentions track AI-overview visibility at 0.664, stronger than backlinks, referring domains, or domain rating; 70–72% of AI mentions carry no link. Public correlational analysis of AI-search visibility, read Jul 6, 2026.
  4. YouTube presence tracks ChatGPT visibility at 0.737, the strongest single factor measured. Public correlational analysis of AI-search visibility, read Jul 6, 2026.
  5. Top-cited domains overlap only about 14% across ChatGPT, Google AI Overviews, and Perplexity; each platform runs its own citation index, so a blended score obscures the picture. Public platform-citation research, read Jul 6, 2026.
  6. Freshness acts as a retrieval gate for AI systems, not a ranking nicety. Public research on retrieval-augmented generation, read Jul 6, 2026.
  7. AI models fall back to primary, seed-tier sources. Public research on model citation behavior, read Jul 6, 2026.
  8. Editorial slate: beats and angles queued from the confirmed 2026 Google update timeline, public keyword-demand data, and the desk's own craft notes on Discover packaging, CRO, paid media, and attribution measurement. Every title is dated for when the desk plans to run it; no finding is shown before the desk earns one, and each piece clears the standard fact-check pass before it publishes.