Analytics for independent publishers is always a tricky thing. We need the same high-end data-rich insights that the biggest publishers have, but we don't have the resources, the scale, or the data to work with. So what do we do? Well, in this article, I'm going to take you through my focus on analytics here on this site, and how that has changed over the years.
I have talked about independent publisher analytics before. For instance, in "A guide to analytics for independent journalists" I talked about many of the more specific details of how to think about analytics. All of the things in that article are obviously still true, but I wanted to talk about the change itself. How has this changed over the years?
In the beginning, I was doing analytics the same way as everyone else. I had added an analytics script to my site (I used Google Analytics, but you can use whatever), and I then simply checked my analytics dashboard. It would then tell me how many unique visitors I had over the past month, how many pageviews I had, which articles had the most views, what the bounce rate was, and a bit about frequency and recency.
However, for anyone who has ever worked with analytics, you will know that these types of metrics don't really tell you that much. Sure, that one article was the most popular this month, but did it really help you as a publisher? You have no idea.
So, instead of using that, I became obsessed with custom reports. Reports where I had put together custom parameters that could better help me understand what was really going on. And after I started using that, I rarely looked at the normal dashboard.
However, even this was very limiting. Normal analytics doesn't necessarily measure what you need to know, and it's often organized to fit into the normal data structure of normal analytics. For instance, you could get a report of which articles were the most popular per month, but what if one article had been published on June 6 and another on June 28 then clearly looking at that per month makes no sense.
It was the same with other metrics. Two key metrics are read rate and read completion, but normal analytics don't tell you that and you often have to do weird workarounds to even add it.
Then we had the problem of not being able to update data back in time. For instance, most of my Plus reports are 30+ pages, so it's quite common that people see them multiple times before they actually read the whole thing.
The result was I would have two pageviews for the same person with zero read completion, and one article with a read completion. But when you then put that together in a custom report, it will tell you that you have a 33% read completion rate ... which is not true, because for that person it's actually 100%. That's useless, and so I got to the point where I couldn't use custom reports anymore.
To fix this, I started building my own analytics system where I could measure things exactly the way I wanted to. It obviously didn't have nearly the same features as the big analytics service, but it did what I needed it to do.
Then I started focusing more and more on 'score analytics'. This is a type of analytics where instead of looking at a single metric, you look at the combined value of all the metrics associated with a person over time. And you then score those signals depending on how valuable you think they are in getting people to do what you want them to do (like subscribing).
This was great and, for big publishers, I highly recommend it. But, for me, I found that I just didn't have enough data. I write 25 Plus reports per year, and the average subscriber reads about 60% of them, so the total number of pageviews per person is not actually that high.
Then I also started looking at learning analytics (here is an old article about that from 2015), which, again, if you are a big publisher I highly recommend you focus on. This method uses machine learning (or simpler equivalents) to try to learn what behaviors or patterns lead to a certain result. And many publishers have found that not only can they use this to increase subscription rates and lower churn, they can also use it to create dynamic paywalls.
It's really fascinating. But again, here on this site, I just don't have enough data to do anything meaningful with it.
Then I also started looking at editorial analytics (article here). Editorial analytics is really important because it defines your newsroom focus, and it measures your relevance and your impact on the public. Again, for big publishers, this is an essential tool for your newsroom.
For this site, I do look at this, but I don't measure this using a tool. I measure it depending on the feedback I get.
Let me give you an example.
Back in May 2019, I published a podcast episode/article called 'Episode 011: The trends around news fatigue and avoidance'. As a podcast episode, it didn't really do that well. Only about 200 people listened to it, and the pageviews weren't that spectacular either. In fact, if I look at my 'top articles', it's not even in the top 100.
However, it had a massive impact. So many people in the industry started talking about it. I was hired to write a more detailed report about it for the European Broadcast Union, and I did presentations about it at several conferences.
I was also hired to do internal lectures about it for two big publishing groups, and I have been interviewed numerous times by journalists who were writing about the problems of news fatigue. In other words, even though this article's metrics were not that impressive, it had a super-high level of impact.
And this is not something I can measure using normal analytics. It's not about pageviews or clicks. It's about "what happened afterwards?"
And a number of my articles are like this. My recent article "Paid-for strategies: Defining what people pay for" is the same thing. In terms of pageviews, it doesn't really stand out that much (it's above average, but not by much). But in terms of impact, it has resulted in several extra things.
So, to me, my most important articles are not the ones with the most views or even those with the highest read rates. It's the ones who have the most impact on people afterwards. And this is a metric that I don't have an automated system for.
Another factor is amplification. As any publisher knows, in order to be profitable, you need to have two elements working for you. You need acquisition (new subscribers), and you need retention (existing audience resubscribing).
To get new subscribers, you need to get the word out that you exist and that you have something interesting to offer. You can do this in three ways.
First of all, you can do it via search (and measure this as search referrals), but to make that work requires that you have a focus that is searchable. In other words, you have to write your articles with search in mind.
Secondly, you can do it with advertising (and measure it as clicks), but again, how well that is going to work really depends on what type of publishers you are. For some publishers this will be very easy to do. For others, like on this site, it's not that useful.
And finally, you can focus on word of mouth, or what we call amplification. In the past, this was really easy to do because the biggest source of amplification was always the social channels. They brought us tons of traffic, and this was also how it started for me here on this site.
However, things have changed and, today, social traffic is not the most important source of amplification. Instead, direct word-of-mouth (people sending links to their friends directly via email, via private chat messages, or simply by speaking) has now become the main driver for a lot of sites. The problem with this, however, is that it has no referrer, so it's incredibly difficult to identify.
So what I do instead today is look at patterns. I will look at key metrics defining value, like read completions, new trial sign-ups, and weaker signals like people generally talking about my articles or journalists wanting to do interviews because of something I recently wrote, and I will use that as my amplification rate.
In other words, I'm no longer measuring amplification rate in terms of referrals. I measure it instead by looking at the (perceived) amplitude of non-subscriber activity.
Now, so far, all I have talked about is analytics. But let me tell you a secret. None of the things I just explained are my most important metrics. Instead, those metrics are from my accounting system.
One of the things you learn extremely early on when you become an independent publisher is that pageviews don't matter. What matters is how much money you make. Without money, you can't pay your bills.
Because of this, my most important metric is my financial dashboard, and I actually have three of these:
I have the overall financial look, which is my bank account, where I simply check whether the amount of money I make over time keeps going up (is it cash flow positive?).
I have another simple overview of additional revenue. This is revenue from articles I write for other publishers, lectures, speaking gigs, and client projects. Keep in mind, this is not my main focus and only represents a small part of my annual revenue, so this is less important for me. But, it's still very interesting if more people are asking for these things because it means that the articles and reports that I write had an impact.
And then I have my big financial dashboard, which is Baekdal Plus. This dashboard represents the revenue I make from my subscribers, and since this is my main focus area and revenue stream, it's vital that I have the details about this.
Luckily, I use Stripe as my payment platform, and Stripe has created a great dashboard for that (located under reports -> billing if you are a Stripe user).
This dashboard tells me:
This is brilliant, and it's exactly what I need to truly understand how my business is doing. So these are the metrics I look at first. Before looking at my analytics, I want to have a clear understanding of my financial performance and patterns.
So if I, for instance, see a huge spike on lost revenue to churn one month, then I want to go look at when that happened, and then find what articles caused that in my analytics.
This, of course, is never as simple as it sounds, but you need this focus on finance first, analytics second.
The only exception to this is when you are doing something new and you don't yet have any financial data to work with. In this case, you need to experiment with your analytics to see what might create valuable activity. And then you audit this by looking at your finances when that starts to come in a few months later.
And again, this is much harder than it sounds.
But, as you can see from this article, my definition of analytics has changed dramatically over the years, and in the next 10 years, new ideas and new approaches will come along.
So, if you are an independent publisher, my advice is to figure out how you make a living, and then define your analytics around that. Find the elements that cause your audience to do what you need, and focus on those finances and metrics.
In other words, analytics is a tool that you work with. It's not a dashboard.
Founder, media analyst, author, and publisher. Follow on Twitter
"Thomas Baekdal is one of Scandinavia's most sought-after experts in the digitization of media companies. He has made himself known for his analysis of how digitization has changed the way we consume media."
Swedish business magazine, Resumé