Robert Rose argues that marketers should let the click economy burn. Average, search-engineered content is losing traffic as AI answers satisfy more queries without a visit. His prescription is to publish ideas worth citing, build direct audiences, participate in trusted communities, and measure relationships rather than page views. The diagnosis of industrialized, interchangeable content is persuasive. Traffic was often treated as the goal when it was only a rough proxy for attention and intent.
The trouble is the proposed replacement. Calling citation the new ranking carries the old logic into a less observable system. A citation inside an AI answer may produce awareness, or it may be a tiny link that nobody opens. It may attribute a useful idea accurately, strip away its qualifications, or place it beside a competitor's recommendation. The publisher usually cannot identify the reader, continue the relationship, or measure whether the mention influenced a decision.
For a software company, that ambiguity may be tolerable because content supports a separate product. For a publisher whose reporting is the product, it is existential. Being the trusted source behind an answer does not pay the reporter, fund the investigation, or create a subscriber relationship. The article criticizes Cloudflare's crawler controls as a defense of pages that host ads, but that framing is too dismissive. Advertising and subscriptions are imperfect mechanisms; they are also how many original facts that models summarize are financed.
Direct audiences are a sensible hedge, not a universal escape. Newsletters, communities, events, and podcasts require discovery, labor, and often platforms of their own. They work best for organizations with an established reputation or a narrow professional audience. A new publication cannot simply decide to have distribution independent of intermediaries. Search once supplied some of that discovery, even while distorting editorial incentives.
The evidence also needs more precision. Falling click-through rates for pages exposed to AI answers do not mean all search traffic is gone, nor do publisher forecasts establish a single future for every category. Navigational, local, commercial, and complex research queries behave differently. A traffic decline can coexist with higher-quality visits. Declaring the old model dead may inspire needed change, but it can also encourage teams to abandon measurable demand before replacement channels prove durable.
A better measurement system would keep several imperfect signals together: citations, qualified visits, subscriber growth, repeat direct use, branded demand, assisted conversions, and revenue. It would also distinguish between a source mention and a faithful use of original work. Publishers need licensing and attribution mechanisms that preserve incentives to create evidence, not merely advice to become more quotable.
The addendum is that traffic should lose its monopoly, not disappear from the ledger. Citation is another proxy, and a particularly opaque one. Content that matters still needs a path from public value to economic support. Until AI platforms make that path visible and compensable, celebrating the fire risks confusing the destruction of bad content incentives with the destruction of the institutions that produce knowledge worth citing.