Subscription teams have spent a decade optimizing one variable: the meter. How many free articles before the wall, which cohorts get looser thresholds, how the countdown resets. That variable is no longer the one doing the most work. Increasingly, the strategic fight is happening earlier \u2014 at the moment of first contact, before a visitor ever sees a meter \u2014 and it’s built around three distinct levers: when the registration wall fires, what the on-site loop does with a reader after that, and how fast identity converts into revenue.<\/p>\n
That shift matters because organic click-through rates have fallen sharply on queries where an Overview appears<\/a>, per Similarweb, and users who encounter an AI Overview click a traditional search result only about 8% of the time, versus 15% when no summary is shown<\/a>, according to Pew Research Center. But the more interesting story isn’t the traffic decline itself \u2014 it’s what publishers are actually building to compensate, and the mechanics are more granular than “put up a wall sooner.”<\/p>\n The Atlantic’s approach illustrates the calibration involved. CEO Nicholas Thompson has described treating AI platforms as a distribution channel worth negotiating with directly<\/a> \u2014 a licensing deal with OpenAI \u2014 while simultaneously pulling the identity-capture moment forward on-site: registration prompts, app installs, and newsletter sign-ups positioned before a reader has burned through a meter allotment, not after. The logic is a bet on sequencing: an email address captured on visit one is worth more than a pageview that may never recur if the query that produced it doesn’t send a second visitor.<\/p>\n That sequencing question \u2014 first visit versus third, hard wall versus soft prompt \u2014 is now the actual A\/B test subscription teams are running, and it has replaced meter-threshold tuning as the primary lever teams pull. A registration wall placed too early on a cold, first-touch visitor from an unfamiliar referral source can suppress the very session data a publisher needs to decide whether that visitor is worth pursuing at all; placed too late, the visitor is gone before an identity is ever attached to them. Publishers testing this are increasingly using cohort-based rules \u2014 visitors arriving from bookmarked or direct-navigation sessions get softer registration nudges than single-query, no-referrer sessions, on the theory that the latter group is least likely to return organically and therefore has the least downside to a harder ask.<\/p>\n Business Insider has been explicit about the second lever: architecture designed to be revisited rather than discovered. Executives told Digiday the company is restructuring site architecture and story selection around “durable” content rather than news-cycle traffic<\/a>, shifting away from disposable posts optimized for a single search snippet toward evergreen explainers and franchise packages built for repeat, direct visits. The mechanical difference is real: a story engineered to answer one query and die has no reason for a reader to come back to the same URL; a franchise package, updated section front, or serialized explainer gives a returning visitor a reason to land on the homepage or app rather than search again \u2014 which is exactly the behavior a registration prompt needs to convert.<\/p>\n On-site loops in practice look like related-content modules tuned for topic affinity rather than recency, curated section fronts that function as a destination rather than a search result, and app or newsletter cross-promotion embedded inside the article body itself rather than in a footer. The goal isn’t engagement for its own sake \u2014 it’s manufacturing the repeat-visit signal that used to arrive for free via search, and doing it inside a session the publisher fully controls.<\/p>\nTiming the wall, not just building it<\/h2>\n
Building the loop that replaces the click<\/h2>\n
The conversion math behind identity capture<\/h2>\n