A registration wall asks a reader for an email address or login before the full article loads. The case for it has always rested on one calculation: whether the revenue from fewer, identified readers beats the revenue from more, anonymous ones. AI search changes the inputs to that sum. The anonymous traffic that used to arrive from Google is shrinking, the traffic that replaces it from assistants is tiny, and the machines that now read publisher pages cannot register at all. That makes registration a bet on the human audience a publisher can still reach directly, and a reason to be precise about which visitors are worth gating.
Why are publishers revisiting registration walls now?
Because the referral base is eroding faster than most plans assumed. The Reuters Institute's Journalism and Technology Trends and Predictions 2026 report, using Chartbeat data from 2,576 publisher sites, found Google Search referrals fell 33 per cent globally between November 2024 and November 2025, and 38 per cent in the US. The same report put ChatGPT at 0.02 per cent of publisher referrals and Perplexity at 0.002 per cent, so assistants are not backfilling the loss. These figures were summarised by Digital Content Next in a 5 October 2026 article; DCN is a publisher trade body and argues one side of the debate, but the underlying numbers come from the Reuters Institute.
On 8 October 2026 Digiday reported that publishers are responding by reconsidering registration walls. The Verge plans to roll out a hard registration wall to a cohort of readers later in October, with its director of audience and subscriptions, Esther Cohen, describing the aim as turning visitors from "lower converting but high traffic sources" into engaged members. She accepts some loss of volume: "Even if there's a loss of volume, there's hopefully a growth of quality." Digiday also cited World History Encyclopedia reporting Google traffic down 50 per cent since 2024, largely due to AI Overviews, and Tom's Guide reporting its UK audience down 55 per cent since July 2024.
What does registration actually earn a publisher?
Three things, in descending order of how well they are evidenced. The first is subscription conversion. Piano's Subscription Benchmark Report, reported by Digiday in July 2022, found anonymous visitors converted to paid at just over 0.2 per cent while registered users converted at about 10 per cent, which Piano described as 45 times more likely. Only about 3 per cent of registered users converted within their first year, so registration is a bridge that pays back slowly, not a quick win. This is the oldest of the figures cited here and comes from a vendor's client base of more than 500 companies, so treat it as directional.
The second is engagement. Gannett told Digiday in 2022 that registered users viewed five times more pages than non-registered readers. The third is advertising yield, which is the least well evidenced in public. Unwind Media, a casual gaming publisher, told Digiday in April 2024 that its authenticated users generated 42 per cent more programmatic revenue than unauthenticated users, with a 90 per cent lift in cookieless environments such as Safari. In the October 2026 Digiday briefing an anonymous executive at the Digiday Publishing Summit said rates on registered users can run "as much as eight times" higher. That is a single unattributed claim and should not be used as a planning assumption.
What does a registration wall cost?
Volume, and sometimes more than expected. Unwind reported that a hard registration wall after 10 to 20 games cut page views by more than 20 per cent and reduced overall revenue by about 15 per cent, which is why it moved towards softer prompts instead. Gannett's Jeff Burkett put the risk plainly: "If you're too aggressive with your login requirements, you'll actually send a lot of people away." The October 2026 Digiday briefing also records a food publisher whose registration wall produced disappointing sign-ups, and a local publisher whose models never found a subscription price that offset the lost visitors, which now uses registered data for personalised newsletters instead.
The pattern across these cases is consistent. Registration pays where the reader receives something in return, such as a newsletter, personalisation, saved content or a comment account, and where the publisher can use the resulting identity commercially. It underperforms where the wall is a toll with nothing behind it.
Hard wall or soft wall?
The published evidence favours soft prompts for most publishers, with a hard wall reserved for traffic that is already low-value. Unwind's post-game pop-up reminder lifted registrations by 20 per cent without the revenue penalty of a hard gate. Snopes found each additional single sign-on option it tested lifted authentication by about 0.5 per cent. The Verge's approach is notable because it is considering gating only certain referral sources or a percentage of visitors, which is a way of spending the hard wall on the traffic least likely to have converted anyway.
Digiday's July 2022 briefing also carried a warning from FTI Consulting's Justin Eisenband: publishers that add a registration wall in front of a metered paywall can lose some of the urgency to subscribe. Sequencing matters as much as the wall itself.
How do AI agents and crawlers interact with a registration wall?
They do not register, and that is the point most registration plans have not yet absorbed. A wall changes what a person sees; it does not change the fact that automated retrieval from AI assistants, Live Search Agents and crawlers is arriving at the same pages. What those requests receive depends on how the site is configured, and several of the biggest sources are covered separately in our guides to AI crawler behaviour. The practical consequence is that a registration strategy and an AI-traffic strategy are two separate ledgers. Registration monetises humans the publisher can identify. It does nothing for the growing volume of machine reads that carry no login, no cookie and no click.
That second ledger is the one blankspace works on: detecting AI and Live Search Agent retrieval at the CDN edge and attaching commercial value to it, independent of whether the reader ever registers or visits. The two approaches are complements, not substitutes. A publisher that builds a first-party audience and also prices machine access is covering both halves of the traffic it actually receives.
Who should consider a registration wall?
Publishers with a repeat-visit habit, a newsletter or app that gives registration a purpose, and a meaningful share of traffic from referral sources that convert poorly. Publishers whose audience is mostly one-off search visitors, or whose content is easily substituted, should expect larger volume losses and should test on a cohort before committing. Whichever route is taken, the sensible design is a staged one: measure the baseline for anonymous visitors, test a soft prompt, then reserve a hard wall for defined segments, tracking registered-user revenue per session against the pageviews given up.
Frequently asked questions
What is a registration wall?
A registration wall requires a reader to create a free account or provide an email address to continue reading. It differs from a paywall because no payment is taken. Publishers use it to identify readers, build first-party audiences and prepare readers for a later paid subscription.
Do registered users really convert at higher rates?
Yes, according to the available benchmarks. Piano's report, covered by Digiday in 2022, found registered users converted to paid at about 10 per cent against just over 0.2 per cent for anonymous visitors. The figure is dated and vendor-sourced, and registered users still convert slowly, with only about 3 per cent doing so in their first year.
Do registration walls increase programmatic revenue?
They can. Unwind Media reported 42 per cent more programmatic revenue from authenticated users in April 2024, with a larger lift in cookieless browsers. Public data is limited to a handful of publisher disclosures, so results will vary by audience and ad stack.
Will a registration wall reduce my traffic?
Usually, yes, and sometimes materially. Unwind saw page views fall by more than 20 per cent and revenue by about 15 per cent after a hard wall. Soft prompts and selective gating limit the loss, which is why The Verge is considering restricting its wall to certain referral sources or a share of visitors.
Does registration help with AI search visibility?
There is no published evidence that it does. Registration is about the human audience a publisher can identify, not about how assistants cite or retrieve a page. Treat it as a first-party revenue decision, and handle AI visibility and machine-traffic monetisation separately.