Subscriptions are far and away the most consistent source of revenue for newspapers. However, to grow their subscriber numbers, sometimes they need to throw away consistency in favor of a more experimental approach.

The Independent has been a digital-only title for six years, having sloughed off its loss-inducing paper edition in 2016. While it has been solidly profitable for those years, its revenue has largely been generated through digital advertising; the same sort of business model that led to mass layoffs and bankruptcy at BuzzFeed and Vice respectively.

To head off problems with digital advertising, the Independent has made it a priority to diversify its revenue. While ecommerce and video advertising are core, its chair John Paton told me it sees a way clear to multiplying its direct reader revenue by five within three years.

To do that, he states the Independent is pursuing an “unknown to known” approach to its audience. Like many digital publications, it asks audiences to register to read. This provides the publication with data that can be used to track their propensity to ultimately pay for editorial content, among other things.

He said direct revenue is “now going to be a fast-growing. Part of that based on the back of the fact we’ve got these registered users who want to talk to us, and our marketing towards that as well. This improved our pitch and led to revenue of about 2 million. That’s going to be a big number like 10 million in two [or] three years.”

The process of turning a user from an anonymous, fly-by-night stranger to a potential customer is vital because every user’s likelihood of signing up for a subscription is wholly unique. The more data a newspaper has on its users the better its chances of converting them.

At the same time, many general-interest newspapers eschew that approach in favor of a one-size-fits-all hard paywall, sacrificing flexibility for clarity. So which approach, then, works best for different types of media brands?

Flexi and freemium

For many titles, flexible paywalls grew out of a freemium model. Austria’s Kleine Zeitung, for example, offers the traditional freemium model with a number of articles available to read before the paywall prompt appears. Other titles can also manually drop the paywall for certain premium articles, the better to entice people to pay.

In this model, eventually, inevitably, the paywall slams down. This is often paired with messages asking for support for independent journalism, or for registration to continue reading. This multi-pronged approach is vital, as research has demonstrated that the single-approach strategy of demanding payments rarely works.

Ultimately, however, these are largely blanket strategies, decisions made manually by human editors based on analytics and gut instinct. It is a beneficial approach both in terms of cost and messaging, as it reinforces the idea that the journalism behind the paywall is universally valuable.

Over the past few years, though, things have shifted. More personalized approaches have come to the fore, powered by AI and machine learning.

Tech and touchpoints

Analyst and founder of A Media Operator Jacob Donnelly explains that audience ID tech is making personalization of paywalls appealing to newspapers after years of scale-based priorities: “You have to start every conversation with the who, and really be clear about who the audience is, so that you can get everything else right.”

It’s a view echoed by Tim Rowell, general manager APAC for Piano, a paywall tech provider. He says that tools to model propensity to pay have been high up publishers’ agenda for years: “The reality is that, for most newspapers, these data are in disparate systems and the challenge is how to to turn all these data points into insight that can be acted upon. Those that can have built models to identify propensity to pay”.

One of the tech advances has been early deployment of proprietary AI tools. Swiss newspaper NZZ, for example, uses AI to personalize its paywall based around individual users’ propensity to pay.

Its managing director Stephen Neubauer explains: “We’re using machine learning today to derive dynamic segments, and to identify these patterns of preference” as a result of which the paywall can be rigidly enforced or relaxed on a per-person basis, the better to entice that reader to pay.

NZZ, however, is one of the most successful digital newspapers out there. As such it is well-positioned in terms of resources to develop and experiment with that AI tool. For many regional and local news groups that time and spend is beyond them, even as the cost of the tech falls. It is far from a foolproof system, as explained by the New York Times’ own applied/data scientist Rohit Supekar – but it is powerful provided you have the resources to do it.

Experimentation and confusion

But to what extent can newspapers try new things with paywalls without confusing their readership – or more likely causing them to wait for a better deal or free access during trial periods?

The same technology and techniques that allow for personalized paywalls is ameliorating that issue. For one thing a vanishingly small number of readers are likely to compare their own paywall offering with others, preventing confusion at a macro level. At the individual level, the ability to tailor messaging also helps users to understand that they are being served deals based on their usage habits rather than some arbitrary decision.

Rowell says: “It’s theoretically possible that audiences get confused by experiments. But the experiments would have to be pretty extreme to do that. We’d argue that it is a fallacy. Sound experiments based on insight and data and designed to improve conversion rates are likely to involve offers that are attractive to the audience”.

Despite all the advances in tech and strategy that allows papers to entice readers down an ever-shifting, personalized funnel, there is still one fundamental issue: a sizeable proportion of people say they will never pay.

According to the latest Digital News Report, 65% of UK news consumers said nothing could encourage them to pay – compared to 49% in the US, both higher than the 42% average across the 20 surveyed markets where paywalls are in effect. That is primarily ascribed to a lack of perceived value in the news products. And if audiences never seek out your site, the opportunity of going from unknown to known to paying subscriber vanishes.