The referral model is genuinely eroding
Answer surfaces resolve exactly the informational queries publishers depend on, and the effect is measurable: impressions hold while clicks decline. Unlike most sectors, where this is an annoyance, for publishers it removes the revenue mechanism entirely, because a summarised article generates nothing.
This makes crawler policy a real commercial decision rather than a technical default. Publishers have a legitimate case for restricting AI crawlers or licensing access, and we will say so plainly even though it runs against the general advice we give everyone else.
Archives are the underused asset. Large publishers hold decades of content that still attracts demand, and it is typically neglected in favour of new production. Refreshing and restructuring high-value archive material usually outperforms publishing more, at a fraction of the cost.
How readers arrive now
Search referral is declining as a share of arrivals, and the publishers holding up are those with direct relationships — newsletters, apps, subscriptions — that do not depend on an intermediary's ranking decisions. Search remains valuable for discovery of new readers rather than for sustaining existing ones.
Readers arriving from an assistant citation behave differently from search arrivals: fewer, but more deliberate. Whether that trade is acceptable depends entirely on your revenue model, and it is worth measuring rather than assuming.
Evergreen and reference content behaves quite unlike news. It sustains demand for years, benefits from refresh rather than replacement, and is generally the archive material most worth investing in.
Copyright, corrections and press regulation
Copyright and licensing are live commercial issues rather than background law, with the use of published content in AI training and retrieval being actively contested and litigated. Publishers need a deliberate position on what is licensed, what is blocked and what is open, and it should be a business decision.
Corrections and right-of-reply obligations affect how content is amended and republished. Silently editing a published article creates its own problems, and visible correction practice is both a regulatory expectation in many markets and a credibility signal that engines and readers both read.
Advertising and sponsored content disclosure is prescribed in most jurisdictions. Inadequate labelling of commercial content attracts regulatory action and undermines exactly the trust that makes a publication citable.
Where the weight sits
SEO carries most of the return in this sector. The immediate levers are conventional — archive value, technical health at scale, and a deliberate crawler policy — while the strategic answer is reducing dependence on referral altogether.
What goes wrong here
- Defaulting to allowing every AI crawler without treating it as a commercial decision
- Neglecting archives that still attract demand in favour of constant new production
- Editing published articles silently rather than correcting them visibly
- Under-labelling sponsored content, which undermines the trust that makes you citable
- Measuring only sessions when the revenue model depends on direct audience
Services that apply
- llms.txt & Crawler Policy
- Crawler policy is a genuine commercial decision for publishers, not a default
- Content Refresh Cadence
- Archive refresh usually outperforms new production at a fraction of the cost
- Technical SEO Audit
- Publisher estates are large and accumulate crawl and duplication problems
- Topical Authority
- Evergreen reference coverage sustains demand that news cycles do not
- SEO Analytics
- Distinguishes referral decline from ranking decline, which need different responses
Questions
Should publishers block AI crawlers?
It is a legitimate decision here in a way it is not for most businesses, because the content is the product. It should be made commercially, with licensing considered, rather than defaulted either way.
Are archives worth investing in?
Usually more than new production. Existing content holds accumulated authority and links, and refreshing high-value archive material typically returns more per hour spent than publishing additional articles.
How do we tell referral decline from a ranking problem?
Compare impressions against clicks per query. Impressions holding while clicks fall indicates answer surfaces absorbing the click. Both falling indicates a ranking problem, and the two need entirely different responses.
Work in Media and publishing?
Thirty minutes with a senior strategist who has worked in this sector. We pull your live visibility while we talk and tell you which constraint is actually binding. Book a discovery call →