ALERT: Early (discounted) registration deadline for Helsinki Media Innovation Week is today

AI’s impact on subscriptions may nudge paywall, direct traffic investment changes

By John DeFriest

FTI Consulting

Washington, DC, USA

By Justin Eisenband

FTI Consulting

Washington, DC, USA


In our most recent INMA Media Leaders blog post, we dove into ways for publishers to use Artificial Intelligence (AI) to automate workflows and improve performance across the organisation.

On the content side, that might include producing commoditised articles, versioning source material to fit different distribution outlets (Web site, newsletter, social, audio/video, etc.) or automating editing tasks.

On the distribution side, it could be curating Web site content placement to maximise engagement or using contextual signals to tag metadata for search optimisation.

And on the customer experience side, it would be personalising Web site and newsletter content or adjusting marketing messaging.

Conversation about AI is largely centered on content and advertising, but its reach extends even further in newsrooms.
Conversation about AI is largely centered on content and advertising, but its reach extends even further in newsrooms.

We’ve also seen great discussions across INMA about use cases for AI within advertising. Publishers are already applying AI models to automate programmatic ad pricing and maximise yield/sell-out rates, in addition to improving contextual and user-based targeting capabilities by building new segmentation and behavioural signals.

These AI solutions help free up time for employees in both the newsroom and business operations to tackle more complex, value-add tasks and also driver stronger performance. We’ve already seen publishers share case studies about the improved engagement, click-throughs, and ad yield they’ve achieved with AI.

Fortunately, these use cases and benefits don’t just stop with content and advertising.

There are numerous opportunities to apply AI to accelerate digital subscription acquisition, pricing, and retention:

  • Acquisition: News publishers have been applying AI-supported dynamic paywalls for years to tailor article counts and subscription offers at the individual user level. These paywalls use demographic, behavioural, and contextual clues to adjust the customer journey to improve conversion rates and offer more balanced content sampling than strict one-size-fits-all settings.
  • Pricing: Similar to dynamic paywalls, AI decision trees can create customised pricing road maps for users to maximise take rates and overall average revenue per user (ARPU). This could include both the initial introductory offer presented, the length of the discount and initial price step-up, or long-term repricing decisions. Combining data points around historical take rates and user-level information, AI models can autonomously execute pricing strategies in real time.
  • Retention: AI systems can also be fed subscriber-level data around engagement over time, content preferences, pricing, tenure, and other demographic information to proactively identify users prone to churn. These multi-variable propensity-to-churn models could be combined with content mechanisms (newsletter, e-mail/SMS notification, pricing, etc.) to build an integrated subscriber re-engagement and retention programme. AI could also help prevent non-pay/involuntary churn by optimising card dunning processes automatically to maximise completion rates.

On top of integrating directly into operations, Artificial Intelligence could become a major disruptive factor to digital subscriptions through another channel: search.

For most engines, the ranked search results list is the primary user-facing experience that funnels users to publisher Web sites. Although most news publishers have moved past the maximising pageviews approach from more than a decade ago, search referrals remain a strong driver of audience traffic and monetisation.

However, the introduction of large language models (such as ChatGPT) to AI is poised to create a new era for search. Instead of providing ranked lists, search engines will shift toward providing written responses to search queries through a chat interface. These AI responses are created by ingesting crawlable wWeb pages and summarising them into direct responses.

This sounds great for users, but the trouble lies with publishers who will no longer enjoy search referrals and may have their content indexed to create these responses without appropriate compensation.

Both Microsoft and Google have announced chat-based AI search integrations. Microsoft Bing added a chat API using ChatGPT to provider written responses to answer use queries in a self-content chat interface.

Meanwhile, Google is using a phased rollout of Bard, which will use PaLM 2 to provide similar summary responses instead of ranked links. Even smaller players, such as the news aggregator Artifact, are adding AI-generated news summaries that limit required time on site and page depth and may eliminate the need to visit publishers’ Web sites.

Any impact on search referral traffic will affect advertising business models first as they depend on ad impressions to generate revenue.

However, there are additional implications for digital subscriptions. If traffic declines due to the reduction of search links in favour of summarised responses, publishers will have fewer users to potentially convert into paying subscribers.

More indirectly, AI models have reportedly been able to access paywalled content (due either to the paywall tech load time or differences between Web crawling and user browsing) and incorporate that content into summarised responses, which would undermine the need for users to pay for content.

Publishers may begin considering tactical changes to combat AI search, including:

  • Investing in direct and newsletter traffic: As search declines as an audience driver, direct-to-site and e-mail newsletter channels will gain outsized importance for maintaining traffic. Reminiscent of the impact changing social media algorithms have had on referral traffic, building direct, proprietary relationships with users through newsletter signups, push notifications, and other direct-to-site strategies will become important. It will also be imperative to maximise each user’s first site visit to build a long-term relationship that can be moved down funnel into paid subscribers.
  • Reducing search indexing/SEO resources: Publishers today allocate resources toward optimising page load, metadata tagging, and other SEO variables to ensure content is highly ranked in search results. If search engines no longer prioritise referral links, there’s a much smaller business case for investing in these strategies which may be deployed elsewhere — such as helping on-site customer experiences and building newsletter programmes — to drive engagement through other channels.
  • Switching to hard paywalls: If AI models continue to access content behind paywalls, publishers may feel compelled to institute hard paywalls with modified tech that prevents search crawling and indexing. Hard paywalls can refer to required registration or a full subscription approach. Registration provides more flexibility for content sampling and relationship building, but if ad rates continue to decline and non-subscriber monetisation is minimal, then publishers may switch to hard subscriber-only paywalls.
  • Prioritising commentary over basic facts: Finally, as basic fact gathering becomes further commoditised by AI summaries in search, adding value-add and unique commentary to content will be a key driver of audience brand recognition. Simple reporting will no longer drive site traffic. Instead, users will look to the newsroom’s unique perspectives on events and potential impacts.

As is the case for seemingly all potential use cases, Artificial Intelligence will likely have a mix of positive and negative impacts on publisher digital subs models. Ultimately, success will be determined by those publishers proactive enough to see the positive solutions and apply them early, while monitoring the negative impacts and implementing protective measures to insulate.

About the Authors

By continuing to browse or by clicking “ACCEPT,” you agree to the storing of cookies on your device to enhance your site experience. To learn more about how we use cookies, please see our privacy policy.