Strategic insights
Debunking The Unique Visitor: Finding Your Real Readers

If you are publisher of content online, either as a newspaper, magazine or just a brand publishing a blog, you are likely to look at two measurements above any other. Those are "absolute unique visitors" and most popular content based on "unique page views."

But those two numbers are pretty much the worst thing you can look at. While the data is useful for further exploration of your statistics, the numbers alone are likely to lead you down the wrong path.

Many people are mistakenly assuming that unique readers are the same as a value audience. In December 2011, The Guardian wrote a very interesting article about the shift in media, but then made the same mistake as most people do when they wrote:

The knock-on impact of the decline [in print sales] has been a push for digital readers that have seen newspapers like the Daily Mail win five million unique visitors a day - compared with its printed sale of two million - but struggle to generate revenues to match.

The reason digital is struggling is simple. "Unique visitors" is not comparable to "print readers". A print reader is a paying reader. A unique visitor is just anyone who happens to come across the website for any reason. Those are two completely different things.

In this article we are going to debunk the myth of the unique visitors. We are going to take it apart and see if we can find the real valuable reader somewhere inside the data. How many of your unique visitors are actually real readers? And, how many of those are likely to buy your product (or buy a subscription)?

The data is ...sketchy

Here is a sad fact about web analytics. It's based on a world that no longer exists. When the internet was first invented, all traffic was assumed to be the result of this simple formula:

As a result, all analytic services track unique visitors by setting a cookie that identifies one browser on one device and then assumes that equals one unique visitor. That might have been true in 1995, but today we live a widely different world.

Because of the social world, we no longer think of "browsing the web" as an activity we do on a specific device. If someone shares a link with you on FB, you are not going to wait until you get home to click on it. You just use whatever browser, app and device you happen near at that moment. But this wrecks the old model for unique visitors. Instead of being one browser and one device for every person. It is now one person using any number of browsers on any number of devices.

And you cannot measure this with cookies, because cookies are based on the old world and limited to one device and one browser.

If you were to visit the Guardian using Chrome on your home computer, IE from your work computer, Safari from your iPhone, within the iOS Facebook app, the Twitter app, Safari on your iPad, Zite on your iPad, and your newsreader. You will be recorded as 8 "absolute" unique visitors.

One area in which this is heavily distorting our analytics is in how we measure our mobile market share. Every single analytics package on the market measures mobile as a separate visitor, meaning that total unique visitors equals "web visitors + mobile visitors".

If you have 150,000 unique web visitors, and 50,000 unique mobile visitors, your analytics will tell you that you have a total of 200,000 unique visitors. Like this:

At first glance, this seems perfectly reasonable and mobile traffic makes up 25% of your total. But you then realize that this is not how people browse the web. People are not using one thing or another, we use both. Consider this graph instead:

Now instead of a total of 200,000 unique visitors, you have 100,000 desktop-only users, and 50,000 user who use both desktop and mobile when visiting your site. Mobile now makes up 33% of your total traffic instead of 25%

Your analytic will still report them as being 200,000, but in reality one quarter of them are "fake" due to over-inflated numbers. That is a big difference both in term of total traffic, but also in mobile market share.

Ask yourself; is the 15% increase in unique visitors you got in 2011 because more people are coming to your site? Or is it simply because more people now own an iPad or iPhone and are thus counted twice?

Of course, these are extremes. Your analytics will always assume the former, but it is likely closer to the latter. In reality it is somewhere in between - like this:

To put this problem into perspective, here is an example from the real world:

In Denmark the largest newspaper (in terms of online traffic) is "Extra Bladet" and these are the numbers they reported for October 2011.

• Unique visitors: 6.238.848 (Danish IP's only)
• Visits: 32.264.986
• Page views: 167.523.457

The problem is just that the total population of Denmark is 5.5 million people, almost 1 million people less than their reported unique visitor.

Google Analytics, Webtrends etc are all doing a great job trying to make it as accurate as possible. But the fact still remains, the data is based on one person equals one device using one browser. If one person uses two devices your data will be way off.

You just don't know how bad it is. You have no way to verify the accuracy of your unique visitors, or even know the data's percentage of uncertainty.

I would love to be able to tell you that you just need to do this, or use that to solve this problem, but I can't. This is how the internet works. The only way to really solve this is when you ask people to register (or subscribe), because then you can forget about cookies and IP addresses and instead track unique *people* based on what they signed in as.

How do you define a reader?

The next problem is a bit more complicated. The mistake the Guardian made when comparing print readers to unique visitors was that they never defined what "a reader" was.

Consider this: If you were to ask a person, "Are you a reader of the Guardian?" What would it take for that person to answer "Yes, I am!"?

There are five criteria that have to be achieved:

1: To be a reader, you actually have to read. A person who just clicks on a link on Twitter is not a reader until that same person also reads the content. Visiting a web page is not the same as reading. You need to eliminate all the unique visitors that don't result in the content being read. On this site, only about 36% of the unique visitors also read an article.

2: To be a reader, you need to read more than one article. If you just come to a site, read one article and never read anything again, then you are not a real reader. You are just a stranger. There is no lasting connection.

3: To be a reader, you need to come back another day. Just because you pick up a copy of the Guardian one day, while waiting for a friend to arrive, doesn't make you a reader. It isn't until you decide to continue to read the content that you would consider yourself a real reader.

4: To be a reader, you need to come back because you *want* to come back. This is a relatively new problem largely created by the news aggregators. If you are using Zite for example and you follow a link to the Guardian two days in a row, it doesn't mean that you are now a reader of the Guardian. You are a reader of Zite. To be a reader of a publication, you need to come back to it because you want to, not because a news aggregator happens to link to it more than once.

Note: Flipboard is different (and so is Google Reader), because then you have chosen to follow a specific publication, which makes you a reader.

5: And finally, you need to be a paying subscriber! If you want to compare print readers with digital ones, paying real money is part of the connection. If you want to read a print magazine, you first have to pay for it. This also means that your content needs to be something people are willing to pay for.

You might argue that if you are a paying subscriber, you don't need any of the above steps - and you would be right. But in the digital world you have to be a reader first before you decide to subscribe, unlike the "buy first, discover later" business model of print.

6: (Optional) you might also argue that there is a sixth element to being a reader and that is that you follow the publication. And there is a strong correlation between being a follower and being a reader but it is not "required". One thing I discovered with Baekdal Plus is that several subscribers are both loyal readers and paying customers, but they are not necessarily following me on Twitter or signed-up for the email update.

Having people sign-up for an email, follow you on Twitter, Facebook or Google+, or subscribe to a RSS feed are all great things. And you should look at that as an indication of loyalty. But following is not necessarily considered to be a real reader.

Putting it into practice

How do you make all this work? How do you identify your real readers?

I wish I could tell you to go into Google Analytics or another system and just "click that button". And I wish I could give you a link to a custom report that would magically put it all together. But I can't do that. The data you need is not something that your analytics is measuring. At least not yet (are you listening Google?!).

You could just look at the fifth criteria of "how many subscribers you have" and calculate your conversion rate by comparing that to your unique visitors. And then you could use Google Analytics' very fancy conversion path features.

But the problem with this approach is that you will never learn the real difference between a random visitor and a real reader. You will know what had happened, but not why.

So let's go through them one-by-one.

1: Measuring readers instead of page views. The way I measure this is somewhat simple. As you read this article, I'm measuring how long it takes for you to scroll from the top of the page to the bottom. Based on that number I make a determination if you are reading the article or if you are just skimming it (or, never reached the end for that matter). The result of this is then stored in the database.

It's not perfect. But it is pretty good. As I wrote above, the result is that only 36% of my unique page views are articles being read. The rest is just people clicking and leaving (or bouncing).

2: Measuring number of articles read per *reader*. This is a bit more complicated. Essentially, you set a cookie for each visitor and then you start counting how many articles each unique visitor actually reads. Of course, the question is, "how many articles does a person have to read to be considered a reader?" If you publish 500 articles per month, you might say that if people read more than 50 of those, then they can be considered "a reader". On this site, I'm publishing 2-5 articles per week, so if you read three or more articles per month, then you are a reader.

Note: Can't you just use "Page debt" and then compare that to your read-rate? No. Page depth is determined by looking at how many pages you read per visit, and that is irrelevant for you as a publisher. It won't tell you how many articles that people read, nor will it measure it over time.

3: How often do people come back? Returning visitor rate is a fairly common feature in all analytic services. But they all assume that a returning visitor is someone who visits more than once. For us, that is not enough.

A person visiting your site twice is not enough to make that person a reader. They have to come back continually. If you are publishing something every day, they need to come back (on average) more than once *every week*. You need to combine recency and frequency of visits.

But more to the point, you need to compare it to your reader. There is a huge difference in returning visitor rate when you compare people who never read and people who do. The average of the total that you have today doesn't really tell you anything.

Look at your readers, how often do they come back and how many times per month do they visit? Then you set a threshold, say: "people have to come back at least 3 times per month, and continue to do that every month."

4: How many readers wanted to come back? This one requires you to look at your referrers. All the traffic that is coming from external sites (as referrers) is not coming back because of you. They came because someone else told them to.

In terms of marketing this "sharing" is really important. But in terms of judging how many real readers you have, it's not. These people might turn into readers (if they come back and read more than one article), but until they do, they are not readers.

So look at your referrers and identify your direct visitors and the people who are coming from places where they can follow you. On this site, that number is around 24% of the total traffic.

5: How many readers paid? Finally, you look at your payment system and you count the number of people who own a subscription, or in some other way have paid for the content.

The result of all this is that you get these nifty graphs that, when stacked together, can give you a much better picture of your real audience. You will find that a huge chunk of your audience isnot worth anything. It is just people who have clicked on a link somewhere, but they didn't read the article, nor come back. They are great for selling ad impressions, but if you want to compare them to the print world of the past, they are meaningless.

This "worthless traffic" is like the people who are walking by your store, looking through your windows as they pass by, but they never stop. They just keep on walking.

Now you need to ask yourself a number of questions:

The first question is, "why are so many visitors worthless traffic?"

Part of the answer might simply be the point I explained in the first part of this article - that your total unique visitors are over-inflated because of the way they are measured. It is entirely likely that the real effect that you have is much higher than you think, but you just cannot see it because the data is wrong.

But also look at other things: How many people start to read an article but never finish it? Is that because you are promising one thing in the headline, but delivering another? Are you employing link-bait tactics and is that actually hurting your business (even though it does generate high amounts of clicks)?

Next is the really interesting difference in conversion rates between your one-time readers, your loyal readers, and your paying readers.

In this example, one critical element is that people are coming back, but they are not reading enough of the content. Why is this? Is it because you are posting too much? Is the content too long? Is the content too complicated or too hard to read? Is it a problem with the design? Is the text too small? Or do you use a strange kind of scrolling that makes it uncomfortable to use?

We could also have a different scenario, in which people are reading many articles, but they are just not coming back often enough. Why is that? Is it because you make it hard to follow the content, and people forget to revisit your site because of it? Is it because you are locking your content inside an app that people have to "go to" and "download"?

When you start to look beyond your total unique visitors and instead look at your real reader, you get this amazing explosion of data that helps you understand your real audience. Not just in how many reader vs. loyal readers you have, but also where the problems might be.

Look at your reader's behavior and dig into the reason behind the numbers. The data within it is absolutely fascinating.

And of course, the *big* factor is how many people pay. The path to a subscription is to get people to read, then turn them into loyal readers and that gives you your potential number of subscriber. But how do you get people to subscribe?

Part of the problem, obviously, is the culture of the internet. Somehow people believe that content online should be free. People have forgotten that valuable content is often the result of a lot of work.

A newspaper that just rewrites a press release and posts that online is not valuable (and good luck trying to get people to pay for that). But valuable content, analysis, insights, expert opinions, experience - that is worth paying for.

This article, for instance, took 4 days to write, it involved a lot of data crunching and is partly the result of an ongoing analysis over the past 4-6 months. But the culture of the internet just thinks "it's text, so it should be free", completely neglecting the value within.

We need to change the culture of free. It's favoring crappy content, and punishing quality.

But the reason people might not pay could also be due to a number of other factors. Is it too hard to do? Is there a problem with trust? Does the payment system involve too many steps, or are you using a payment provider with a bad reputation?

Maybe it is because people can't try it out and see what it is all about before they pay. Or maybe it is the design. Maybe your payment system looks like a banner ad and is thus being ignored by people with banner blindness.

There are so many fascinating aspects to this.

The point is two things:

1. If you want to compare print readers with you digital audience, forget about total unique visitors. They are two completely different things.
2. If you want to understand what is really going on with your site, look at your readers. Look at their user behavior and usage patterns. Identify that threshold for a successful conversion and match that against your audience.

You will likely be surprised by what you find.