The Globe and Mail’s ‘Sophi Dynamic Paywall Engine’ Increases Digital Subscriptions 51%

The Globe and Mail launched a hybrid paywall in 2012 — applying a simple meter for most articles, with editors deciding to place about 10% of articles behind a hard paywall. Like most news organisations, we were heavily reliant on advertising revenue at the time and did not want to risk killing the golden goose. However, we knew that the future of a sustainable journalism business depended on how aggressively we could acquire and retain subscribers. We rapidly started to see a bottleneck for both new subscription growth and registration growth from our static paywall. We tried to tighten the paywall by increasing the total number of hard-paywalled articles and experimenting with the number of articles one could read via the meter — but that either depressed user engagement significantly or we missed out on potential subscriptions because we asked the wrong people at the wrong time. To break through and create sustainable growth, we needed to redefine how we decided when to present a paywall or a registration wall and when to let the reader keep reading uninterrupted. In other words, we needed a fully dynamic paywall that offers one-to-one personalisation in real time to ensure we were not leaving any money on the table. The dynamic paywall knows when to let users keep reading and when to ask them to subscribe. We were finally in a position to undertake this ambitious project. We had one data scientist in 2012; the team of data scientists and engineers at the Globe now numbers 50. How we did it We leveraged the customer journey to train our data-science model to study readers’ sequential online behaviour and then predict when and on which article to show a paywall or registration wall (or no wall at all) to each specific reader. We also used sophisticated natural language processing techniques to ensure that we were picking appropriate articles for each reader. Because the new paywall directly affected our new subscriber acquisition, we wanted to be careful and gradual in introducing it, using our test-and-learn methodology. We started by limiting our A/B testing to audiences in a few small cities. After seeing encouraging results, we then removed our geo limitation and switched to testing it on small segments of our entire audience and gradually increased those segments. During our transition to the new paywall, we encountered COVID. We needed to balance competing needs — driving subscriptions while ensuring that we fulfilled our mission in providing the public with vital information on a poorly understood, swiftly spreading disease. So we implemented various constraints within our dynamic paywall, which actually diminished its performance in favour of ensuring that Canadians understood how to stay safe and healthy in unprecedented times. We also ensured the newsroom retained the power to manually override any algorithmic paywall decisions for cases where it believed the content must be free to all readers because it was a matter of public interest. Despite having many complicated rules, we were still pleased to see stunning results. Our goals were to increase the subscription and registration conversion rates without diminishing reader engagement. The personalisation paid off, with increases across the board. The results: We also encountered another unexpected success: We typically run a few “flash sales” a year and usually see a drop in subscription gains soon after each one. But with the fully dynamic paywall algorithm, the diminishing subscription trend that followed flash sales has disappeared.

*First published on

How to harness human bias in algorithmic content curation

How to Harness Human Bias in Algorithmic Content Curation

Algorithms that curate content are a popular villain. We know now how Facebook has repeatedly used them, both in politics and personal life. They are well known for amplifying human biases – because algorithms run on data, and it is humans that pick the data that is used and also how it is applied. But what if we could harness human bias to run an algorithm that actually made humans’ lives better? What if we could find places where it’s a good thing to have human views and to ensure that they actually matter more than what the algorithm considers important – while the algorithm takes on tasks that rely less on the power of human discernment? This is what we set out to do when we created Sophi Site Automation, an AI system that incorporates the best of human judgment as well as machine learning, with safeguards built in to prevent the machine from descending into clickbait hell. Let me explain. Sophi Site Automation places content on publishers’ pages according to what is most valuable to them (where “value” incorporates subscriber acquisition and retention as well as reach on search, social and other platforms). But most importantly, Sophi will only place content on a page within constraints specified by editors. In other words, the humans entrusted with safeguarding the brand and mission of a newspaper are the ones who decide the parameters for what is eligible for placement on a digital page. This makes logical sense. Editors are the stewards of the brand. They work for your publication because they care about what you do -- and about what they do. They are detail-oriented and mission-oriented professionals. That is why you hire them. If you want your website to look as though it is being run by thoughtful humans who understand your brand, it is important that you take their thought process into account when setting up an algorithm to do the work of content curation. So, editors decide what kind of content can go where – the mix of topics, how old an article can be, and so on. We encode this in the algorithm. Then, we let Sophi do its work. We let it optimize the placement of content so that the most valuable articles are getting exactly the right amount of promotion on the right spots of the page and being served to the right segments of the audience. We are now experimenting with true one-to-one personalization. Does it work? As a journalist, I was surprised to see how well it works. We have been using Sophi Site Automation to place 99 per cent of the content on The Globe and Mail’s digital pages for more than two years now. Our click-through rates and subscriber acquisition rates are dramatically higher. But the most remarkable result is that not a single reader has noticed that it is an algorithm placing content on the page. We have received not one complaint or question from our readers (and I can tell you that they are generally not shy about complaining or questioning our news judgment!). By running our digital pages for us, Sophi has given our editors time to work as journalists, to find news and use their superior story-telling abilities and news judgment to shape coverage. They are back to doing what humans do far better than any machine can, bringing about change by holding the rich and powerful to account. But does the algorithm ever make questionable decisions? Yes. Our editors don’t love every single decision that it makes. In those rare cases, editors can override Sophi’s decisions by using a simple Slack command that takes a few second to execute. This is generally used about one to four times a day, which is impressive, considering that we produce hundreds of articles a day. As an aside, what has been really interesting is to see what the algorithm has taught us. We had assumed that our news content had a short shelf life. We had assumed that readers did not value our service journalism as highly as our news content. We had assumed that wire content would never be as valuable as our proprietorial journalism. And we were wrong on all three counts. There was a lot more value in our content than we had imagined, and it took an algorithm that channeled human intuition to unlock it.

Organizations need a strong first party data strategy

The Best Way to Activate Your First Party Data Strategy

Google deferred the death of the third party cookie until 2023, but that only means you have a little extra time to take ownership of your relationships with your audience and get to know your customers within your own site. If you go to many sites today, within a few seconds you‘re immediately presented with a giant popup window asking you to sign up for a newsletter you’re not currently interested in given the value you haven’t yet received from that site. It’s ultimately an irritating barrier to what you’re trying to accomplish on that website. Now there is hopefully some A/B testing behind that registration popup, but it’s definitely not the most effective way to get signups and build relationships – by asking (bothering, really) everyone, all the time. “That’s a common first party data strategy these days, but the likelihood of success just isn’t very high,” said Gordon Edall, Co-Founder and CRO of The newsletter bar may be lower than when you’re trying to drive registrations or logins or an account-type relationship, so it’s easier to be more… should I say persistent, and care less about scaring visitors away. But this is all part of a chain of actions you take to get your users and customers to tell you more. You need them to login so you can develop a first party relationship. Gordon explains: “Sophi for First Party Data, very similar to Sophi Dynamic Paywall Engine, treats this as a problem of progressive disclosure – how do you get from one piece of data about your visitor, to two pieces of data, to a completely different relationship with higher loyalty, retention and engagement.” The key here is that Sophi understands when and where some of these steps can be skipped in order to get a consumer to your end goal faster. Companies usually follow the ordered steps of first asking a user to register, then asking for their credit card for a sale or subscription, and so on. All these steps are great opportunities to activate your first party data strategy and learn about your customer, but not every consumer needs to follow those steps in that progressive order. Making everyone follow the same path regardless of who they are and where they are in the buying cycle is a slow, ineffective process that can be optimized using artificial intelligence and machine learning-powered solutions like Sophi for First Party Data, which has visible advantages over more progressive data strategies. Sophi’s technology system is designed to get you to the highest value state as quickly as possible. Don’t ask someone to sign up for a low value newsletter when you actually want them to take out their credit card and make a purchase. Sophi for First Party Data is fully dynamic and personalized in real-time. Sophi can tell you – through your offer management system – how often to ask for an account signup first and bypass the newsletter all together. Or for people who signed up for a newsletter and haven’t given you their name or income yet for example, Sophi can tell you as soon as there’s a chance to trigger ‘tell me more about you’. Once Sophi understand the tiers of your offers (for example: do nothing, ask for a registration, get a credit card number), it understands the goal and automatically goes after as many victories as possible, looking at each individual user and their propensity to say yes to a given tier of offers at that particular moment in time. “Sophi takes your value proposition and gets users to the higher levels of value in their relationship with you, as quickly as possible.” – Mike O’Neill, Co-Founder and CEO of It’s highly portable too as it’s a lightweight trigger engine. It doesn’t replace your marketing technology stack; it integrates into it. It enhances the systems you already have in place, making them work better to achieve your goals more effectively. Sophi for First Party Data can also be used to drive ecommerce sales directly to grow revenue, but that’s a whole other conversation. To recap, Google has postponed the death of the third party cookie, but it is still dying. If you want that invaluable relationship with your customers, you need to start building it now by developing a robust first party data strategy within your own platforms. This inevitably means moving to a registration-based relationship where users tell you what you want to know. And they aren’t just going to do that on their own. You need to convince them to do that, to tell you their gender and their income and so on. Your competition is working on this now, so don’t handicap yourself by not taking those steps today. Feel free to reach out to discuss enabling and optimizing your first party data strategy.

The 'aha moment' with Sophi Social

The “What Planet Are You From?” Aha Moment

What happens when the seventh biggest company in the world finds out that you’ve been using an artificial-intelligence system to maximize your performance on their platform? At, our mission is to help organizations identify their most valuable content and ensure that it’s shared at the right time and place with the right audience, leveraging AI. Facebook found out we were doing this, enabling The Globe and Mail to better engage Facebook readers by selecting which stories should be shared on the site and when. It turns out AI is really good at this. Facebook was impressed enough to publish a case study on their website last month. The highlights are that through Sophi’s artificial intelligence, we posted 3 times as much content, with no drop off in engagement, and digital subscriptions were up 27%. Why does this work? We recognized that simply posting stories as they’re published is a poor way to engage users. We began to leverage the key strengths of AI: it learns remarkably quickly, understands your audience extremely well, and can make more lightning-quick decisions than people ever can. These small optimizations add up to huge differences at scale. So, how does it work? 1. The Globe and Mail produces a lot of content, up to a couple hundred stories every day. 2. Throughout the day, Sophi is constantly posting content to our Facebook page. 3. As users click on articles, they arrive on article pages that Sophi is constantly optimizing to drive recirculation and engagement. 4. Sophi’s fully dynamic real-time personalized paywall analyzes both reader behaviour and content to decide when to present the option to register or pay for Globe content – and when to let the reader continue uninterrupted. 5. As new stories are published throughout the day and new readers engage with our content, Sophi Site Automation optimizes everything again, continuously. From this workflow, there are millions of combinations of what content to put in front of readers at different times and on different platforms. Sophi is also continually evaluating how much of any particular topic can be posted at any time. It’s also making sure that as new stories become available, the most valuable stories are posted at the right time, so our Facebook page is more than just a duplicate feed of the stories posted to our website. And when we drive traffic to our site, Sophi is working under the hood to ensure those readers are more likely to go deeper into our content by showing them what they consider relevant. A few weeks ago, a potential client understood the value of constant optimization, and exclaimed: “What planet are you guys from!?” In my 11 years in enterprise sales, it was one of the most memorable aha moments I’ve witnessed. And what was Facebook’s reaction when they found out that we were using Sophi to optimize our social media posting? They decided to join us - and we’re now working together to help other publishers with their social media posts and subscriptions and expect to have another case study out within a few months.

Content publisher

4 content lessons learned from journalism that can help brands

In the two and a half decades that I have worked in newsrooms, I have constantly been surprised by our readers. They are willing to pay for what? They pay us more than what they pay Netflix every month – and then spent most of their time reading what? Here are four lessons I’ve learned that anyone who produces content should be aware of: 1. What you think is your most valuable content really isn’t what your readers think is valuable. I am amazed at what our readers take out their credit cards for. Policy-changing investigations that we spent 20 months on? Sometimes. How to sleep better at night? Often. A story on something marginal, obscure or secondary? Good god, a couple of hundred times a month! We only realized this to be true after we let an intelligent algorithm that uses Natural Language Processing, which we call Sophi, scrutinize every article that we publish and predict what we should paywall. It turned out that we were leaving millions of dollars a year on the table because we, as editors, didn’t realize the value of all of the content that we put in front of our readers. We were too busy looking only at the biggest stories that we, as journalists, took the most pride in. But readers don’t always share our sense of pride in holding the rich and powerful to account – they just want to be smarter or richer or healthier. The moral here is that content editors should not be the ones trying to figure out what readers consider valuable. Let a machine that doesn’t have your biases do that work objectively and intelligently to best understand your readers. 2. What brings them in the door isn’t what keeps them as paying customers. Ok, so now you’ve committed to paying us 25 bucks a month. Surely you’re spending your time on our site reading thoughtful journalism that helps build a stronger nation. That’s what we’d like to believe, anyway. Actually, our subscribers like to read a lot of frivolous stuff. Our tv critic’s views on three new thrillers on Netflix are right up there at the top of the list, next to headlines on grim COVID news and fretting about the runaway real estate market. Once we understood this at the Globe, we decided to experiment with serving two different versions of our section pages and article pages – one that offers anonymous readers a mix of content that is optimized for driving conversions, and another that serves paying subscribers a mix of content that is optimized for retaining them as paying subscribers. We were pleasantly surprised to find that this kind of segmented optimization was very successful. 3. Perfectly polished copy is not essential. When I first started at the Globe, our writers’ copy would go through four pairs of eyes before it made it to print. Then we got into blogs and breaking news – two pairs of eyes were plenty. And now, you can read star writers’ raw, unedited copy through media such as Substack – and readers don’t seem to mind one bit. 4. Fresher is not always better. If you’ve worked in the news business, chances are that you automatically assume your readers always want fresh news. If it ran yesterday, then why would anyone read it tomorrow? When we automated our digital content placement using Sophi, we were surprised by how often the algorithm would surface content that was more than a few hours old. Surely it was broken? But, no: many of our readers actually do want yesterday’s news tomorrow – and sometimes even next week. It could be because they weren’t glued to their computer screens for eight hours a day reading the news the way we were. They may have only just heard of something (such as the mystery of why butter is harder in COVID times) and come to our site looking to understand why. Or maybe they stumbled upon it serendipitously while browsing altogether different content, and it’s still interesting to them. This is why pages where content is in reverse-chronological order are full of lost opportunities. A smart algorithm sees this, and can find your buried gems for you.

Automation helps publishers

4 ways automation can help publishers

I work for a serious, 177-year-old national newspaper that has embraced radical automation: 99 per cent of the content that you see on The Globe and Mail’s digital pages is placed there by a clever artificial intelligence algorithm called Sophi. Every article on our Facebook page is also placed there by Sophi. How did this happen? I believe it is because we could clearly see the value of harnessing machine learning– and reallocating our talented editors to the tasks of gathering news and shaping it into absolutely first-rate journalism. Here is a look at four actions that we have automated because we believe that our talented staff are better suited for other work: What we realized over time is that Sophi can accurately predict both the amount of subscription revenue and the amount of advertising revenue expected by every article that we produce – and it is astonishingly good at it. So, our editors went back to assigning and editing content, and Sophi took over hard paywalling. (Of course, our editors still retain the right to manually override any decisions that they disagree with but they rarely use this veto power.) As it turns out, we ended up switching off our meter as well. Sophi now runs a fully dynamic, personalized, real-time paywall that knows when to ask for money, when to ask for an email address and when to simply let the reader read uninterrupted. Sophi looks at all the different ways in which every article that we produce is valuable, taking into account how much advertising revenue and subscription revenue it generates, as well as understanding the role it played in retaining existing subscribers. It can serve many different versions of the same page, optimized for different outcomes. So, for example, it can populate one version of the page with articles that are optimal for driving subscriptions and offer this to visitors who have yet to subscribe. It can offer another version of the page to existing subscribers, this one optimized for retaining existing subscribers. Sophi understands that different visitors may value pieces of content differently and can optimize the page depending on who is visiting. This means interpreting and autonomously acting upon thousands of signals simultaneously, a task that any human would struggle to do efficiently. Intelligent automation really can move the needle on revenue – and it frees up our brilliant staff to focus on work that humans can do better, like producing award-winning journalism.

See for Yourself

Get in touch today to learn how Sophi can transform your business.

Get Started