The Economist are known for being one of the most forward-thinking publishing brands when it comes to data and audience monetisation. Their paywall is well over twenty years old, and has seen numerous evolutions during that time.
The Economist’s Head of Insight and Data Science, Adam Davison explored the brand’s paywall evolution at the PPA Festival on Thursday 9th May, and explained how data has been used to take it to its present iteration.
20 years of paywalls
“Over the past 5-10 years, we’ve been in this massive transition from an advertising – predominantly print advertising – model, to a model where we attract revenue through paid subscriptions,” explained Davison; something he has in common with a growing number of publishers. “We’re at the point now where paid subscriptions are the majority of our revenue.”
Since the launch of the Economist.com in 1996, there has always been some form of registration on the site. Initially, that involved access to the archive and an exclusive Economist screensaver (!), and has evolved over the years to allow various levels of access.
In 2012, visitors to the website were able to access five articles, with a further five allowed if they registered their details – a popular ‘soft paywall’ approach in use by many publications like The Telegraph today.
“Over time, we’ve been in this process where as we’ve matured as a digital business, and maybe also the market’s matured, and people’s expectations of paid content have developed, we’ve been sort of tightening up what people can read for free,” Davison explained.
By 2018, this had reduced further to one free article, with three allowed in return for registration. However, last year the publication began to ask whether they should be tightening that paywall even further.
“The natural evolution, the natural next step for this is obviously to say, should we be giving any of our content away? What’s the right strategy for us?”
Weighing up a harder paywall
There are many clear advantages to taking a harder approach to paywalls. Encouraging registration in return for article access builds a strong marketing database to encourage subscriptions further down the line, without having a lot of drive-by anonymous traffic.
Paywalls also reinforce the perception of the publisher being a premium brand, which in turn is good for trust.
However, registration walls or hard paywalls introduce another level of friction, and risk having a significant effect on advertising inventory, as well as having a knock-on effect of brand awareness if less content is being seen and shared online.
“There’s a risk that by damaging engagement, you are maybe limiting how much of an audience we have for our content. First of all, there’s an immediate risk on advertising revenue, we’re reducing inventory, but also in the long term, are we limiting our brand awareness to new people who maybe aren’t willing to pay, no longer discover our content?”
Fast-forward to early 2019, and The Economist is now behind a hard paywall. Readers can see just two paragraphs before being asked to register or log in. So how did data help the team make this decision?
Using data to determine the risks
Changing a paywall involves more than just tweaking the technology. Davison made the point that the whole culture of how data was used in the organisation had to change in order to get actionable insight, and make changes that would be beneficial to a long-term strategy.
“Historically, we’ve not been very data-driven when evaluating these trade offs basically,” he admitted. “It’s been very much sort of…business strategy gut feel, maybe a little bit of data here and there, but probably not used anything like as effectively as it could have been. So I think with this latest transition, I really wanted to try to do this the right way, use data to be as informed as possible when we made this decision.”
This involved making some changes to the way The Economist dealt with data across the business. The company had talented data analysts, but they worked in independent, siloed teams. It was a similar story with the data, which was collected with individual tools across separate parts of the business.
Part of the journey over the past few years has been unifying data collection across the whole organisation, and bringing together a centralised data science team for proper analysis. Once the right data and teams were in place, testing could begin on the effects of changing the paywall.
To get to the current evolution where only the first few paragraphs of an article are available to read, the team did extensive A/B testing to evaluate the impact a harder paywall would have. The findings may not be entirely surprising, but they are encouraging:
“We definitely saw a big increase in registrations as a result of forcing people to make this decision to sign up. We also saw an increase in conversion rates, and this signalling to us that, yes, there is an upside to doing this, at least a short term upside to doing this.
“There clearly were an audience of people out there who were happy to read the content for free, but actually valued it enough that they were willing to go further than that if we forced them.
Additionally, once people had registered, they went on to read and engage more on the site, which offset the bounce rate from those who chose not to register and left the site.
“We actually found a relatively small negative impact on engagement. I think there are two reasons behind this. One is that once people had registered, they tended to then become more engaged with that content than they would have been otherwise. So actually, although there was a trade off in kind of some, maybe increased bounce rates from a hard paywalls, what we found was that actually the increased registrations offset that to some extent.
“But also if we’re honest with ourselves as a business, before, anonymous readers can only read one article. In fact, on top of that one article, we’ve added a load of messaging, CTAs to try to explain to them why they should sign up to newsletters, or be aware that we have a 12 for £12 offer.
“Actually, in some ways, just saying to people ‘Just pay or go’ somehow simplifies the message.”
A/B testing alone is not the solution
One of the key findings Davison was keen to highlight is that A/B testing on its own was not exclusively the solution.
“So the A/B test gave us a really good measurement of the short term impact on conversion, registrations, advertising, inventory, of making this hard paywall change. But there were some things that we couldn’t test for tech reasons, specifically, some ad delivery metrics, we just decided the technical work to get those into the test just wasn’t worth it. And also the way external platforms like Facebook are going to respond to changing engagement on your site, you just have to monitor that and try to build in some some estimates.
“I think the most important thing that we sort of knew in the back of our heads but we didn’t take seriously enough up front was the fact that the A/B tests can only tell you about short term impact of these changes. And some of these risks and opportunities to do with brand perception are really long term things.”
Their approach therefore was to run the A/B test to reach conclusions about things that could be of short term benefit, and then once they were aware that there was a potential upside, they had to consider what the long-term strategic business goals were to see whether there would be a wider benefit to these changes.
“It ended up being aligned, because we’ve been really clear up to now that our strategic direction as a business is to move towards revenue from paid subscribers, and away from a focus on advertising. We believe The Economist is a premium brand. So it just made sense if the short term things are positive, and your long term direction is aligned, it’s actually an easier decision than you might think.”
Key lessons from changing the paywall
Davison closed by highlighting some of the key lessons he’d apply if he was to do it again, and the advice he would give to others who were looking to make similar changes.
“I don’t think it is controversial to say that A/B testing this kind of stuff is is really, really powerful. Right? It’s a pretty powerful tool, you should definitely be doing it if you’ve got the tech and the capability.
“But some people in the business expected A/B testing to be a magic bullet, it’s going to tell you what to do. And it doesn’t, it tells you some things, it tells you something about how your users behave in response to your paywall, and changes to that. But it doesn’t answer every question, it doesn’t tell you what you should do.
“And ultimately, you need to be able to combine your quantitative data that tells you about the short term impact, with business strategy, what you think the right thing to do is.”
Davison also highlighted the role of their in-house data scientists and experts in making the initiative work, pointing out that in-house people make it much easier to interpret the results, and the impact they would have on the strategic goals of The Economist.
Finally, he concluded by recommending that businesses think early on about how those test results are going to be turned into an understanding of the financial impact.
“At the start we thought, we’ll do the tests, we’ll have the results over to the finance team, they’ll do some calculations, it’ll all work. And of course, it turns out the context that they think in terms of is totally different.”
“I think if we’d started that conversation up front, we would have been able to marry those two together much more easily.”