Media Briefing: As AI search shifts referrals, publishers revisit registration walls 

This Media Briefing covers the latest in media trends for Digiday+ members and is distributed over email every Thursday at 10 a.m. ET. More from the series →

This week’s Media Briefing looks at why shrinking referral traffic is giving registration walls a new purpose, one that publishers are testing in different ways: turning fewer, anonymous visitors into known — and potentially more valuable — users.

  • The registration wall revival
  • 3 Qs with Joy Robins, The New York Times’ global chief advertising officer
  • Apple plans to block programmatic tech companies, Google gives publishers “Preferred Sources” counts, and more.

The registration wall revival

As referral traffic shrinks, publishers can no longer afford to let the readers who do show up to their sites remain anonymous. That’s prompting many to revisit a familiar tactic: registration walls. 

The new question is whether a reader who isn’t ready to subscribe can still be valuable as a known user, helping publishers deepen engagement, understand their audience and potentially improve ad targeting. The payoff isn’t guaranteed: Asking readers to register can also drive them away. 

The Verge is putting that question to a test. Later this month, it will roll out a hard registration wall to a cohort of readers. Esther Cohen, director of audience and subscriptions at The Verge, told Digiday she wants to learn how to turn visitors in “lower converting but high traffic sources” into “engaged and known members of our audience.”

The publisher is weighing different approaches, such as requiring registration from a percentage of site visitors or from visitors arriving from specific referral sources or channels, Cohen said. If they don’t register and create an account on The Verge, they won’t be able to access the site.

“The idea is to bring those folks who aren’t ready to subscribe – whether it’s because the source they’re coming from, whether it’s because [of] the type of story they’re clicking on – to bring them into the fold a little bit further down the funnel… and then be able to give them tastes of what they could get by just engaging with us,” Cohen said.

Cohen said she wasn’t worried about creating friction that might turn away some readers. “Even if there’s a loss of volume, there’s hopefully a growth of quality,” she said.

Reg walls as part of an ad strategy

Other publishers see another potential pay-off: registered readers may be more valuable to advertisers, even if they never subscribe. By getting readers to register, publishers can turn anonymous visitors into identifiable audience cohorts. That first-party data can be more valuable to advertisers and drive higher programmatic ad rates, publishing execs told Digiday. 

“Those reg walls give you the kind of data that allows your programmatic ad sales guy to get higher yield, and that higher yield from cohorts that have registered will offset the lower yield that you get from the decline in your drive-by Google traffic,” said one publishing exec during a closed-door town hall session at the Digiday Publishing Summit last month. 

“That’s the calculation that we’re beginning to see. If you can get them to register first and then continue to give them free content, we get a much higher rate. Our rate on registered users – obviously it changes multiple times a day – but it can be as much as eight times,” they added.

A smaller pool of known users can potentially generate more ad revenue per user, and that makes registration valuable even when publishers continue giving those readers content for free.

Data from reg walls can help segment audiences

Data from reader registrations can also be used to segment audiences across multiple revenue streams. One publishing exec in the town hall said registrations were serving two purposes: They use registration data to determine whether someone is a prospective subscriber and move them further down the funnel, while also charging advertisers more to reach known registered newsletter readers.

Publishers can also use data from registered users to test how much free, unpaywalled content to give to improve engagement (such as a few stories a month), as well as test giving free gift links in newsletters and on social to high-traffic, low-conversion paywalled content, another exec said.

Registration walls don’t always lead to subscriptions 

Naturally, getting readers to register for free doesn’t guarantee they’ll ever pay.

One food publisher launched a registration wall last year to disappointing results, despite adding features and other benefits to its offering. “We were shocked by how few people actually wanted to register with it,” an exec said during the town hall.

But when the publisher launched a digital subscription this year (which included a print magazine and access to new recipes), it was “a totally different game,” they said. “It shocks me how many people want to subscribe.”

Another tactic that has worked well for one news publisher is adding the word “exclusive” to headlines, which has driven a double-digit increase subscription conversions, driving 40 to 60 new subscriptions on exclusive stories a day, an exec said in the town hall. 

Not enough to offset lost revenue

Subscriptions aren’t a universal solution for every publisher. They can create meaningful new revenue, but are unlikely to come close to replacing lost traffic – and those economics are particularly difficult for some local news sites. 

The food publisher exec admitted that while their new subscription offering was becoming a “significant” new revenue line, it was not yet anywhere near enough to offset revenue lost from recent traffic declines. 

Another publishing exec said they had repeatedly run models and never found a price point where subscription revenue beats the loss in site visitors.

“We continually do research on: Is there a benefit to putting this behind a paywall? What could we gain from that in terms of revenue procession as we see monthly active users diminish in the way that the entire open market is seeing? And the math never works. We can never figure out, ‘Okay, if we charge X, that’s going to reduce our monthly active users by Y. Are we going to increase revenue per session per user by enough to at least break even, and God forbid, make more money than we were in the previous model?’ I’ve not found any locally-focused companies that have done a good job in doing that,” they said.

Instead, the publisher is using registered users’ data to create “more robust” audience profiles and offer personalized newsletters, which are “more marketable” to advertisers, they said.

The exec also felt that adding a paywall to their sites would just worsen the traffic problem many publishers are facing.

“People are not landing on our sites because they’re getting the information they need from a Google search response or jumping into an LLM themselves. You’re just throwing gas on that fire by putting any sort of blockage between the consumer getting the information that they’re coming to you for,” they said.

What we’ve heard

“It’s a concern because it’s a dynamic list that can be updated at will by Apple to impact businesses it believes are circumventing ITP…It’s a vendor issue that makes it a publisher issue too…It’s also very clearly designed to give advertisers more of an incentive to use their ad network.”

– Publishing executive who spoke anonymously about Apple’s blocklist. 

3 Qs with Joy Robins, The New York Times’ global chief advertising officer 

Robins spoke with Digiday about selling its portfolio in the AI era, at Advertising Week New York on Tuesday. The conversation has been lightly edited and condensed for flow.

SG: In my recent reporting, I’ve been hearing that AI is making brand authority and credibility more important. Do you agree, and if so, how are you trying to tie that even more intentionally into the Times’ ad sales strategy?

Robins: I agree with that. I think it’s more important than ever. Having direct relationships with consumers is going to be more important than ever. For the Times, what we do… is going to be rare and more valuable in this particular kind of ecosystem, and I think that the relationship that we have with our consumers and readers and trust and love can translate directly to our advertisers, and really help them understand how they can bring value to those communities, to those readers, to those audiences. That’s the difference.

SG: What are you doing in particular to help advertisers see the impact of their ads?

Robins: We’re continuing to evolve our own targeting capabilities and really develop increased capacity of measurement capabilities. So we’ll have more to share on that in the new year, which we’re excited about… BrandMatch [The Times’ AI-powered contextual targeting tool] is one of the ways that we’ve done that, and I think that there’s a whole way of continuing to evolve the success that we’ve seen with BrandMatch into more and more identity solutions and targeting people.

SG: Are you testing anything with agentic media buying?

Robins: That’s not really something we’re doing. We’re always evaluating. I think the thing that we really want to make sure of is, what is going to be something that is right both for our advertisers and for us. We’re really keeping an eye on it. But that’s kind of the evaluation phase, and that’s where we are right now. 

Numbers to know

$110 billion: The cost of the Paramount Skydance takeover of Warner Bros. Discovery, creating a company spanning CBS, CNN, HBO Max, Paramount+, Warner Bros., DC Studios and Paramount’s film and television assets

50%: The decline in World History Encyclopedia’s Google traffic since 2024, after falling 25-35% in each of the past two years, largely due to AI Overviews.

55%: The decline in Tom’s Guide’s UK audience since July 2024. 

440,000: The average pageviews The Daily Mail’s “Deep Dive” visual stories now bring in, typically generating 60–100 new subscribers.

~150: The number of AI-related copyright cases in the U.S., according to The Association of American Publishers.

What we’ve covered

Apple’s secret blocklist is the talk of Advertising Week New York (mostly off the record)

  • Out in the snaking lines outside the Advertising Week New York conference venue, the talk often circles back to Apple — the biggest brand/platform provider in the industry — which reopened its war with the digital ad industry just a week before.
  • The news – that Apple’s iOS 27 updates meant it had started cutting off companies that track people around the web to target them with ads – came too late for the official agenda, so ad execs have been picking it apart over coffee and drinks all week, leaving many in a tizzy.

Read more here.

Dhar Mann’s $100M AWNY deal target for creators is already pledged (so he says)

  • Dhar Mann is AWNY’s first chief creator officer, and his job this week is to broker deals between creators and brands. He and the event’s organizers want those introductions to produce $100 million in creator-brand business within 12 months.
  • Mann said after several conversations with CMOs, their combined pledges for AWNY already exceed the $100 million benchmark.

Read more here.

SPUR publishes AI content tracking standard, pitches OpenAI and Google to join advisory board

  • A coalition of media organizations, including the Guardian and the Financial Times, has released a new standard for tracking how AI tools use publishers’ content.
  • The Standards for Publisher Usage Rights initiative, or SPUR, has invited OpenAI, Anthropic, Google, Meta, and Microsoft to help shape how the standard is implemented by joining an AI Licensing Advisory Board.

Read more here.

The creator economy’s stance on AI is shifting — again

  • At the beginning of the year, the creator economy denounced generative AI content in favor of authenticity. YouTube nuked AI slop accounts, and Instagram CEO Adam Mosseri released a lengthy memo about improving ranking for originality.
  • But in the last 10 months, AI has steadily crept into the creator economy. TikTok editor CapCut has a suite of AI editing tools. Adobe Firefly is billed as an “all-in-one creator studio.” YouTube is adding new agentic tools.

Read more here.

What we’re reading

Apple plans to block hundreds of programmatic data companies from iOS

Apple is expanding an iOS 27 privacy feature that could block hundreds of programmatic data companies from collecting data or serving ads on Apple devices, AdExchanger reported/

Google tells publishers how many users picked them as a ‘Preferred Source’

Google has begun emailing publishers showing how many users selected them as a Preferred Source, according to PPC Land. One of publishers’ biggest complaints about Preferred Sources has been the inability to determine whether the feature actually changes traffic or engagement.

Australia’s ABC rejects AI copyright carveout

National ​broadcaster the Australian Broadcasting Corporation is pushing back against a proposed AI copyright opt-out, arguing existing copyright law should require AI companies to license content rather than putting the burden on publishers to prevent their work from being used for AI training, Reuters reported.

Poland’s antitrust regulator accuses Google of unfair market practices

A Polish regulator said Google isn’t playing by the rules when the tech giant displays articles from local media publishers in Google Search, News and Discover. It could hit Google with a fine amounting to up to 10% of Google’s annual revenue, Engadget reported.

More in Media

Apple’s Safari ad tech blocklist sends publishers back to IAB Tech Lab’s Trusted Server

What publishers once filed under someday now has a deadline, and Apple set it.

SPUR publishes AI content tracking standard, pitches OpenAI and Google to join advisory board

Publishers are pushing for AI transparency with a new SPUR standard to track how AI tools use content, and inviting frontier AI labs to help.

Media Briefing: Publishers are turning GEO from an experiment into a business

Publishers are turning GEO into a new revenue stream as clients increasingly look to boost their brands’ visibility in AI-generated answers.