The Weekly Bulletin | April 7, 2026

Catch up on your members' content, check out the community buzz, and browse through job opportunities

Author

Vahe Arabian
April 07, 2026

Hi SODP community,

Let's recap on what's been happening, the new content, industry updates, tips, and more.

.TIP OF THE WEEK.

Build to Own: A Publisher's Step-by-Step Guide to First-Party Data Infrastructure

Developing a first-party data stack is essential for publishers to regain control of audience insights, increase ad revenue, and prepare for a cookieless future. 

A robust stack enables the collection, organization, and activation of data directly from user interactions. Here is a step-by-step guide to developing a first-party stack:

  1. Identify Data Collection Points & Define Strategy
    Start by mapping every touchpoint where your brand interacts with users to gather data. This involves shifting from passive audience measurement to active, permission-based collection.
    • User Registration/Logins: Implement universal login systems (e.g., Auth0, Okta) to encourage user registration.
    • Newsletter Subscriptions: Create interest-based newsletters allowing users to self-select into segments.
    • Interactive Content: Use quizzes, surveys, polls, and calculators to collect declared "zero-party" data.
    • Engagement Tracking: Monitor behavioral data such as click-through rates (CTR), scroll depth, time-on-page, and article completion rates.

  2. Choose the Core Technology Components
    A first-party stack typically includes several integrated technologies:
    • Customer Data Platform (CDP): To unify data from various sources (site, app, email) into a single, cohesive user profile.
    • Customer Relationship Management (CRM): To store and manage authenticated user profiles and registration data.
    • Consent Management Platform (CMP): To ensure compliance with regulations like GDPR and CCPA while managing user opt-ins.
    • Data Warehouse (optional, recommended): For complete ownership and advanced analysis of data.

  3. Implement Progressive Data Collection
    Avoid overwhelming users with long forms. Instead, use "progressive profiling" to gather more information over time as trust builds.
    • Initial Visit: Only ask for an email address.
    • Subsequent Visit: Ask for content preferences or job title once they are engaged.

  4. Activate Data for Monetization
    Once data is collected, activate it to increase the value of your inventory.

    • Create Audience Segments: Segment users based on behavior (e.g., "Auto-intenders," "Frequent Investors").
    • Private Marketplaces (PMPs) & Deals: Use Deal IDs to offer premium advertisers access to specific, high-value audiences.
    • Programmatic Guaranteed (PG): Sell audience-based, first-party data campaigns directly through your ad server (e.g., Google Ad Manager).
    • Data Clean Rooms: Partner with technology companies (e.g., InfoSum, LiveRamp) to safely match your data with advertiser data for targeting, without sharing personal information (PII).

  5. Ensure Privacy and Compliance
    First-party data relies on trust. Clearly communicate your privacy policy, provide easy consent options, and ensure data governance is tight.

Summary Checklists

  • Identify key engagement points (polls, login walls).

  • Deploy a Consent Management Platform (CMP).

  • Select a CDP to unify data.

  • Build an authenticated user base.

  • Segment audience into high-value cohorts.

  • Activate segments in Ad Manager or a DS

.NEWS OF THE WEEK.

➡️Your Owned Content Is Losing to a Stranger’s Reddit Comment. The next time you ask an AI what product to buy, which agency to hire, or which software platform actually works, pay attention to where the answer comes from. Increasingly, it does not come from the vendor’s own website. It comes from a stranger’s Reddit comment written eighteen months ago, upvoted 847 times by people who tried the thing themselves. This is not an accident. It’s architecture. The financial architecture behind Reddit’s presence in AI answers became public in early 2024. Google signed an initial licensing agreement with Reddit worth a reported $60 million per year, with total disclosed licensing across multiple AI companies reaching $203 million.

➡️ The Stack: AI Surges while Social Platforms Face Scrutiny. Regulatory tensions kicked off the week in Australia, where the eSafety Commission flagged “major gaps” in how platforms are enforcing the country’s under-16 social media ban. TikTok, Facebook, Instagram, Snapchat, and YouTube were all found to be falling short of compliance requirements, raising fresh concerns about child safety and platform accountability. Meanwhile, OpenAI made headlines with a USD$122bn funding round, pushing its valuation to an eye-catching $852bn. Backed by major investors including SoftBank, Andreessen Horowitz, and Microsoft, the raise signals strong market confidence as the company edges closer to a potential public listing.

➡️Gemini Overtakes Perplexity, Becomes No. 2 Bot Referral To Websites. Google Gemini has overtaken Perplexity to become the world’s second-largest source of AI chatbot referrals to websites for the first time, according to independent web analytics company Statcounter. Publishers have struggled with the lack of referral traffic from AI chatbots to websites from one of the most used traditional search engines worldwide, but recently this has changed. March 2026 data released in April shows Google Gemini accounted for 8.65% of referrals -- up from 2.31% in the year-ago month. It measured all AI chatbot referrals globally, surpassing Perplexity, which fell to 7.07%.

➡️ Inflation is back: Should you raise subscription prices for news? Amid the war in the Middle East, inflation is heading to 4% in 2026. Your subscribers’ energy bills are already reflecting that. As the cost of living and publishing rises, should you raise subscription prices too? In March, the OECD revised its 2026 forecast for inflation across the 20 biggest economies by 1.2 percentage points. A prolonged war in Iran could add another 0.9 points in 2027. Consumers in the European Union and the United States are growing less confident about their future finances, even if they have not yet cut spending. Marketers are already bracing. WARC estimates that, in a pessimist scenario, US$94 billion could be wiped from expected growth in the US$1.4 trillion global advertising market by 2027.

➡️ Newsrooms face AI control crossroads. At the end of February, Anthropic made the unusual move of publicly rebuking the Pentagon during an active contract negotiation. In a statement published on its website, Anthropic laid out two proposed use cases the company found objectionable: Using its technology for fully autonomous weapons and mass surveillance. “We believe AI can undermine, rather than defend, democratic values,” CEO Dario Amodei wrote. “Some uses are also simply outside the bounds of what today’s technology can safely and reliably do.” This was a stunning development, not just because the Trump administration moved to punish the private company soon after, but also because of the timing.

➡️ WordPress 7.0 Release Team Pauses Schedule to Finalize Real-Time Collaboration Database Table Design. Work on WordPress 7.0 is effectively on hold as the release team and core committers head into the Easter long weekend unsure about how to proceed with versioning and when a new release date might be set. On Monday, WordPress co-founder Matt Mullenweg called for a delay to 7.0 — which had been planned for release on April 9 — to allow for more time to implement new database tables to address cache invalidation issues with real-time collaboration.“Given the scope and status of 7.0, I think we should go back to beta releases, get the new tables right, lock in everything we want for 7.0, and then start RCs again,” he wrote in the #core-committers channel in WordPress Slack.

.SODP POSTS.

Social media giants are not complying with under‑16s social media ban, new report finds

Nearly four months into Australia’s social media ban for under-16s, the online regulator today released its first detailed compliance update report on how the world-first policy is progressing.

eSafety’s report comes at a crucial time, with many other countries eyeing the progress of the ban. Since the ban took effect on December 10 last year, I have spoken with journalists from Canada, France, Germany, Japan, New Zealand, the United Kingdom and elsewhere. Everyone asks two questions: how successful is the ban, and are children still accessing social media platforms?

The new report paints a complicated picture – and leaves other key questions about the social media ban unanswered.

A number of compliance concerns
The report acknowledges social media companies have taken “some steps” to comply with the social media legislation (which restricts account holders to those aged 16 and older). Some 4.7 million accounts were removed by mid-January and another 310,000 by early March.

However, the report also highlights “compliance concerns” in four key areas:

  1. Messaging to under-16s on some platforms encouraged children to attempt age assurance even where they declared themselves to be underage

  2. Some platforms enabled under-16s to repeatedly attempt the same age-assurance method to ultimately pass age checks

.UPCOMING EVENT .

SODP Publisher Revenue Dinner in London

We're hosting an off-the-record dinner for senior publishing revenue and commercial leaders on Wednesday, June 10 at Cornus Restaurant, London (6:30 PM GMT.) The focus is on how to extract a reliable commercial signal as AI reshapes content creation, audience discovery, and inventory measurement simultaneously. You'll get candid peer exchange on revenue strategy, content licensing, inventory quality, and how to future-proof your commercial model at a moment when the rules are changing fast. You will have access to dinner, drinks, strategic insights that won't be shared publicly, and our private post-dinner network. Many thanks to Ezoic for partnering with us to host this. Seats are limited.

.JOB BOARD.

➡️ NBCUniversal (United States) is looking for a Sr. Manager, Consumer Insights who will play a key role in helping to develop forward-looking insights that deepen the business’s understanding of marketing, title level tracking, key audiences, and content opportunities across the evolving entertainment landscape. (U.S).

➡️ Nine (Australia) is seeking a Chief of Staff to join their team, who will be responsible for running the newsgathering operation of the newsroom by identifying the key stories to cover, assign reporters, deploy camera crews across the day and respond to breaking news. (Australia)

➡️ The Telegraph (United Kingdom) is looking for Head of Breaking News who will be responsible for overseeing a team of editors and reporters who monitor, identify, verify, commission and publish breaking news, with an initial focus on domestic news. (U.K)

.SOCIAL MEDIA.

➡️ Olivier de Segonzac on LinkedIn:

We just published "Inside ChatGPT Search", a full technical study on how ChatGPT Search works, what broke with GPT-5.3 and 5.4, and how to monitor your visibility in this new landscape.

Built on hundreds of prompts tracked daily over 14 weeks, reverse engineering of internal tools, a honeypot experiment, and a reconstructed system prompt

What’s inside:
1/ The Bigfoot Effect (cc Dr. Pete Meyers ^^)
When GPT-5.3 became the default model on March 4, average unique domains per response dropped 20% (thanks Idriss Khouader from Meteoria for data !). Fewer sites now share the same visibility surface.

Lily Ray and Chris Long documented and explained the emergence of site: operators in GPT 5.4 fan-outs. We extend this with new data and analysis of how this mechanism behaves in production

Jérôme Salomon (Oncrawl) recently confirmed the crawl contraction through server log analysis

2/ "web run" dissected
ChatGPT accesses the web through a single internal tool called web.run. Since 5.3: 12 operations, including the new find, click, and screenshot for navigation. Full syntax documented and a new type of fan-out discovered!

3/ Same cutoff, different behavior
GPT 5.2, 5.3, 5.4... all share the same training cutoff (August 2025). Yet the same prompt produces different fan-outs, retrieves different sources, and cites different brands. Four divergence layers: weight changes, post-training (RLHF/SFT), compute budget, product orchestration...

These are the highlights for the last week.

Until next!

Vahe Arabian and the editorial team at SODP