The largest local news publisher in the US reported its third quarter on 5 November, and the most useful thing in it is not the headline number. It is two small lines moving in opposite directions.
USA TODAY Co. (formerly Gannett) booked $21.3 million in "digital other" revenue, the category it says is driven by AI partnerships and content licensing, up 24.8% year on year (named-party figure). In the same quarter, CEO Michael Reed told analysts that indirect digital advertising, roughly $100 million a year and fed largely by search, fell 32% (named-party figure).
That is the shift in one quarter. Machines are now a revenue line. Search-driven open web advertising is shrinking fast. The question is whether the first can ever replace the second.
What did USA TODAY Co. actually report?
Total revenue was $519.2 million, down 7.4% (named-party figure). Digital revenue was $249.3 million, 48% of the total, and the company expects digital to pass 50% during 2027 (named-party figures). Adjusted EBITDA fell to $36.6 million from $57.2 million a year earlier (named-party figure).
Digital advertising came in at $75.0 million (named-party figure). Against the $87.2 million in the prior year quarter shown in the company's 10-Q, that is a fall of about 14% (blankspace estimate, calculated from company filings).
Reed said the company "felt the impacts of the changes in consumer search". It answered with a $100 million annualised cost reduction programme, implemented at the end of September (named-party figure). Full year guidance moved to a mid-single-digit same store revenue decline.
What is "digital other", and why does it matter?
"Digital other" is where USA TODAY Co. books AI partnerships, content licensing and syndication. The company does not break out which AI counterparties sit inside it this quarter, or on what terms.
At $21.3 million a quarter, it is now worth roughly 28% of the company's entire digital advertising line (blankspace estimate). Two years ago that comparison would have been meaningless.
Reed said he expects sequential growth in Q4 "as we sign additional AI licensing agreements and grow our commerce business". The company said it expects to announce at least one, and potentially two, more AI licensing deals before year end (reported, per the call transcript).
Why is search-fed advertising falling so fast?
Indirect advertising is programmatic and sponsored link inventory: pages reached through search, monetised through auctions. Reed called it the company's "lowest value advertising inventory" and said search referral traffic "has historically been one of our lower value audience sources".
That is a publisher talking down the channel that is leaving. It may also be true. Direct sold advertising grew in the quarter, digital-only subscription ARPU rose 23% to $10.82 (named-party figure), and unique visitors ticked up to 161 million from 158 million in Q2 (named-party figure, Comscore and Adobe Analytics).
But the audience did not vanish. It moved. More of the questions that used to send a click now get answered inside an AI interface, often using publisher reporting. Across the market, AI bots now account for 1 in 50 publisher visitors (OPA Europe / Rossel, January 2026), and IAB Tech Lab puts publisher losses to AI bot traffic at $2 billion a year (reported).
Is AI licensing the replacement for search revenue?
Not yet, and the structure of the money explains why.
Licensing arrives in deals. Deals are negotiated, timed and lumpy. According to the call transcript, management attributed part of the guidance cut to AI licensing revenue shifting later in the year (reported). A revenue line that depends on when a contract gets signed is not the same as a revenue line that grows with usage.
Licensing also concentrates. A company with 161 million monthly uniques and hundreds of mastheads can get a meeting with Microsoft or Meta. Most publishers cannot. The long tail of the open web is read by the same AI systems and has no deal to sign.
And licensing prices access, not consumption. A flat agreement pays the same whether an AI assistant reads a page ten times or ten million times. The reads keep scaling; the contract does not.
What does this mean for publishers?
Three things.
First, the AI read is now provably worth money. A public company is reporting it as a growing line next to advertising. That ends the argument that machine traffic is just cost.
Second, the inventory being lost is the inventory AI is consuming. Indirect, search-fed pages are exactly the pages Live Search Agents read when a user asks a question. Those agents do not run JavaScript, so no ad tag loads and no auction runs. The read happens; the impression does not.
Third, blocking is a negotiating tactic, not a business. USA TODAY Co. says it continues to block scrapers (reported) while it signs deals. That works for publishers with leverage. For everyone else, blocking simply removes them from the answer, while the user gets served from whoever stayed open.
What does this mean for advertisers?
The audience that used to arrive via search is now forming its decisions inside AI answers. Budgets that chased that audience through programmatic search pages are following a shrinking pool: one indirect channel at one large publisher fell by a third in a year.
Display, paid search and social do not appear inside an AI answer. If a brand is not present at the moment the answer is generated, it is invisible at the moment of decision. The supply that matters is moving from the page view to the read.
the blankspace read USA TODAY just showed the industry both halves of the problem in one quarter. The read is worth paying for: $21.3 million says so. But it is being paid for the old way, in deals signed by the biggest publishers and priced for access rather than use. Meanwhile the search-fed inventory that funded the open web is falling by a third. The fix is not a bigger licence. It is a price on every read, at the moment it happens, for every publisher an agent touches. Monetise the read, not the referral, and stop waiting for a contract to tell you what your audience is worth.
Key figures
| Figure | Value | Provenance |
|---|---|---|
| Q3 2026 total revenue | $519.2m, down 7.4% YoY | Named-party figure (USA TODAY Co. earnings release) |
| Q3 2026 digital revenue | $249.3m, 48% of total | Named-party figure (earnings release) |
| Digital share of revenue, 2027 target | 50%+ | Named-party figure (company guidance) |
| Q3 2026 digital advertising | $75.0m | Named-party figure (earnings release) |
| Digital advertising change YoY | About minus 14% (vs $87.2m) | blankspace estimate (calculated from company 10-Q figures) |
| Indirect digital advertising | About $100m annualised, down 32% YoY | Named-party figure (CEO, Q3 call) |
| Q3 2026 "digital other" (AI partnerships, licensing) | $21.3m, up 24.8% YoY | Named-party figure (Q3 call, as reported in transcript) |
| Digital other as share of digital advertising | About 28% | blankspace estimate |
| Additional AI licensing deals expected by year end | 1 to 2 | Reported (call transcript summary) |
| Average monthly unique visitors | 161m (Q2: 158m) | Named-party figure (Comscore, Adobe Analytics) |
| Digital-only subscription ARPU | $10.82, up 23% | Named-party figure (earnings release) |
| Q3 2026 Adjusted EBITDA | $36.6m (prior year $57.2m) | Named-party figure (earnings release) |
| Cost reduction programme | $100m annualised | Named-party figure (earnings release) |
| AI bots as share of publisher visitors | 1 in 50 | Reported (OPA Europe / Rossel, January 2026) |
| Annual publisher losses to AI bot traffic | $2bn | Reported (IAB Tech Lab) |
Primary source, opened and verified: USA TODAY Co. Q3 2026 earnings release (Form 8-K, Exhibit 99.1). Supporting: Form 10-Q for the quarter ended 30 September 2026 and the Q3 2026 earnings call transcript.