The Problem with Page Views

Photo by Lukas Blazek on Unsplash

Recently, the difference between page views and page loads when tracking analytics has come up often so I thought I would write out some thoughts to address this common concern. Realistically, tracking either one of these is important and better than no tracking at all and in my experience, most companies are lucky to have an analytics framework in place let alone one where they understand the differences and nuances in the way things are tracked. That being said, as I help companies set up their tracking — this nuance is proving to be more important as time goes on, especially as things like bots or regression testing can really bog down ideas of how we think about tracking on websites and where we can actually garner valuable information.

The way we tend to think about traffic is by unique users and we tend to build funnels around what we want unique users to do and whether or not they are completing the funnel. When they do not complete the funnel, and “drop off”, we tend to look at aspects of the funnel to improve on to see if we can move the numbers in a way that directly affects revenue. Ideally, you work somewhere where all of your metrics lead to clear goals and the KPIs you are measured against make sense in a larger business context. If this is the case, you will likely be looking at user behaviours and, in the context of something like a marketing website, looking at number of visitors versus how many clicks on a CTA and then compare those to a specific revenue-driving event such as a sign up or a purchase of a product.The big issue here is that most places are not comparing apples to apples in funnels like this, and since we tend to rarely have an idea of “uniqueness” in our initial Page View event, there is a huge likelihood that the denominator is not accurately reflecting a user’s journey through a funnel and instead is being skewed by different factors that contribute to abnormally large Page Views. Further, metrics like bounce rate tend to be skewed since they are based on Page View counts. There are two key elements at play here: How we understand a page visit and how we understand a unique visitor.

Since we don’t tend to have information on visitors when they first arrive on a page, understanding them as unique typically considers each user sessions. There are several gaps in the assumption that a new user session = a unique user. Yes, it is possible to store information in a cookie so we are not considering every new visit from the same IP address or device as new, but this is still an imperfect way of tracking. This approach can and does consider the same user multiple times throughout their time using your product. Further, it tends to over represent anything programmatically hitting your site or platform. For example, a bot might hit your site 1k times a day and that will dilute the number of unique users you think you have per day. Further, there might be regression testing or QA that visits your site multiple and sometimes hundreds of times a day which will inflate your Page View count. Ultimately, digging into Page Views here makes it clear that this might not be the best route to go down if you are looking at a specific funnel and trying to find opportunities to optimize within that funnel. If you’re wondering why 95% of the users dropped off from visiting the site to clicking the CTA, the problem might not be the CTA, it might actually be that 50% of that traffic are not unique users intending on looking at your page but rather something programmatically affecting your numbers. Page Views should only really be considered directional unless you have safeguards in place like preventing bots or the same IP address from hitting your site multiple times.

Conversely, Page Loads can work to thwart some, but not all bot traffic, since many bots tend to hit a page before a full page load. This is still an imperfect approach to be sure, but better than a simple Page View and is much more used in a Conversion Funnel. Page Loads are a better way to not only understand whether or not a user has hit your site but also how long it takes for that page load event to happen. Here, you can track 500 errors to understand whether or not a user is being dropped due to an error rather than a lack of interest. This allows you to build out Dashboards and Charts not only on User Behaviour but also on performance. That being said, Page Loads still can be victims of bots and the numbers still can be bogged down by web crawlers but at least you can rule out anything programmatically hitting your page before it loads. Further, if you are working on a conversion funnel like a checkout funnel that includes multiple items, you are in a better position to compare apples to apples. For example, if a user had multiple items in a cart, you can track the checkout page load event as the top of funnel rather than the product page view. There is certainly more to be said around conversion funnels, so I will address this in a future post.

While Page Loads and Page Views are two different things, the main takeaway is to uncover what is affecting your numbers rather than taking them at face value when you see them. Page Views in general should be thought of more directionally, and when you are thinking about metrics such as bounce rate, there could be any number of things weighing down the number you see. This is where a good analytics platform where you can dig into your data more is essential. One of the best things you can do for your Growth Strategy is to be critical of your numbers and curious about where they are coming from. This approach tends to uncover more and opens up possibilities you might not have thought of. Importantly, make sure your numbers are true which will get you closer to what your users are actually doing which in turn lets you get closer to fully addressing their pain points. Ultimately, a great Growth Strategy is one that has a full understanding where the data is coming from so we can more accurately predict where we can go.

Data and Analytics Strategist and Consultant. Product Manager at TWG