Whether you are a small business with one dedicated “web guy” to do all things related to the website or you are a larger business with a team of dedicated analysts who manage analytics, you need to have a web analytics strategy in place to leverage web analytics effectively.
Here is the fact – web analytics is hard. And it keeps getting harder as more and more devices and web properties (websites, mobile sites, and mobile apps) can send usage data that can be measured. Without a robust plan around how to use analytics, you could run into inefficiency issues, accuracy issues, and the basic inability to make the most of your web analytics data.
Why is Web Analytics Important?
Web analytics can help you understand what is happening on your website. You can use it to understand the behavior of visitors to your site, the actions they take on the site, the source of these visitors, and much, much more.
And then there are other advanced reasons to use analytics – to understand and optimize ecommerce/subscription funnels, to determine ROI of marketing spend, to measure results of A/B tests, and so on.
Regardless of what you use web analytics for, there are a few problems that are commonly seen across companies.
Typical Problems in Using Web Analytics
Here are a few common web analytics related behaviors found across tiny, small, large, and massive organizations:
- Web analytics (e.g. Google Analytics) is implemented but not used
- Web analytics isn’t really a priority; it is more of an afterthought
- Every time a report is sent out, no real action results from it
- Web analysts are very busy generating reports and have no time for ad hoc investigative questions
- Big budget decisions are often made without web analytics data
If any of these statements ring true for you, please tweet us and read on!
So what can one do to use web analytics effectively without needing a web analytics degree? Read on!
The Truth about Web Analytics
There are only three things in web analytics that are important to understand. Once you understand these three things, then using web analytics becomes more of a mechanical exercise. And this applies to businesses of any size. Here are the three simple truths:
- Knowing what metrics to measure is half the battle
- Understanding the context to apply when measuring is the other half of the battle
- Ensuring your web analytics tool is recording correct data accurately is paramount
Figuring out Web Analytics Metrics to Use
Any popular web analytics tool (Google Analytics, Piwik, Adobe Analytics, etc.) comes with a standard set of reports. Once your web analytics tags are implemented, then data starts getting recorded in your tool. At this point, you can open up the interface and pull a bunch of reports to get a read on site performance.
However, you will get the best bang for your buck and time by knowing what metrics to use when measuring performance of your website. However simple your website, you need metrics that make sense for your business and not just the ones that are common across all websites (visitors, sessions, pageviews, etc.).
So let’s walk through a quick guide on how to identify metrics that make sense for your business:
- Write down how your business makes money (i.e. services rendered, products sold, subscriptions acquired, etc.)
- Write out how your website contributes to that business goal (i.e. it drives phone calls, it captures potential customers’ interest, it processes transactions, it provides information about your business, etc.)
- Now write out how a change in the website’s contribution changes how much money your business generates. For example, fewer phone calls means fewer appointments and fewer subscriptions means less ad revenue
- Figure out which of these changes drives 80% of the impact to your bottom line. Select at most three such drivers
These will be the three metrics you can use as Key Performance Indicators for tracking the performance of your site. Let’s get just a bit fancier – The three drivers you have identified are probably absolute metrics such as visitors, subscriptions, sales, etc.
The problem with absolute metrics is that they hide the root cause for changes. A good metric is some sort of a comparison to another metric. So it is relative (a ratio or a percentage). Why? Because there is almost no scenario where a single absolute metric tells the entire story. Let’s use the example of a site that generates revenue by selling a widget. For this simple example, a potential KPI could be total revenue. And you could look at total revenue per day. A week over week change in revenue could tell whether the site is selling enough. However, it doesn’t tell you whether the change was because you sold more widgets to the same number of visitors or because more visitors came to your site and you sold fewer widgets per visitor! The point of a good metric is that you need to make it actionable. So instead of looking at just visitors or just revenue, you could create a new metric called Revenue per Visitor and track that over time. Now you are tracking two metrics with a single report! If that metric changes, then you can dig into the report to understand which of the two core components – revenue or visitors changed. Below are two charts that show the contrast between using three metrics to tell a story vs using a single actionable metric. Note that there are other issues with using RPV, but I won’t get into those now.
This is just a simple example of how to build a KPI. The key takeaways here are that a great metric is
- Easy – It is simple to communicate and simple to understand
- Contextual – It is relative to another metric
- Actionable – It can drive further investigation or immediate adjustments
It isn’t easy to create good metrics. So start simple and generate reports using these metrics. Once you see these reports on a daily basis, ask yourself – so what? So what if my report shows an improvement or decline in a metric over time? Why should I care? Once you start poking holes in your own reports and metrics, creating a great metric will be just a matter of time.
How to Apply Context to Web Analytics
A famous web analytics guru once said “all data in aggregate is crap.” Perhaps I am paraphrasing, but he is right. The whole point of web analytics is to understand what is happening on a site and then to find the reason that it is happening. And that is only possible by slicing the data and looking at it from various perspectives. And that is what I mean by “applying context to the data.”
Context needs to be applied to standard web analytics reports as well as to ad hoc reports that are run to investigate website issues. I am going to use the example of a typical web analytics investigation to explain how context is created.
Let’s say you have seen a drop in transactions in the past week. And you’ve seen this to be a 5-7 day trend. It is very tempting for even the most seasoned analysts to just dive into analytics to solve a problem rather than talk through the approach. Instead of opening up the web analytics interface to investigate, the first step actually is to create a plan of how to approach the problem. In this particular example, we would first make a list of possible reasons why transactions dropped. Here are some things we would want to check:
- Did the conversion rate change (transactions/visitors)?
- Did site visitor counts drop?
- Did transaction counts for a type of product change?
- Was there a change in marketing tactics recently?
- Was there an update to the site recently?
The purpose of such high level questions is to help understand the areas that we need to dive into. So it is just a start. Let’s say we found that there was an update to the site. Note that in the real world, it is highly improbable that only one change happened. The issue is typically a result of half a dozen things happening at the same time. But for the sake of simplicity, let’s say the only change was a site update. Then the next step will be to check whether the update to the site caused an unexpected problem. Here are five things you would want to look into:
- Is this a real problem? In other words, has the presence of a problem been validated by any other means such as an automated analytics audit or a backup web analytics solution?
- Does the problem show across all platforms (i.e. mobile devices, desktop, various popular browsers, etc.)?
- Is it a sitewide technical problem such as slower than normal response times?
- Is there a particular page in the eCommerce funnel that is now showing higher exit rates?
- Is the problem limited to a particular channel (e.g. Paid Search)?
Once such questions have been formulated, then and only then should you even bother to open up a web analytics interface for investigation. Using such context, reports can now be run to see if the data trends can reveal the root cause of the problem.
As I said before, creating context is not just for investigating problems. This similar exercise needs to be executed even when putting together the daily dashboard or set of reports that you will use to track the performance of your website.
Ensuring Your Web Analytics Tool is Accurate
Ensuring that the web analytics tool continues to report reliable information is probably the most overlooked area in web analytics. Accuracy of data should be a fundamental concern for every analyst. Yet so many CMOs, CEOs, and other executives ask every day “How confident are you about this data?” Or worse, they are the ones that point out obvious problems with the data and then they lose complete confidence in the data, the analyst, and the analytics tool!
So let’s step back and ask ourselves “How do we ensure the data in the web analytics tool is accurate?” To be clear – the issue is not with the tool itself! The root causes are bad data was fed into the tool, the tool was improperly configured, or bad code broke analytics tags on the site.
So there are three approaches to addressing this problem:
- Pre-Implementation Processes – Implement a dozen processes on the front side to ensure incorrect data is not fed into the tool, to ensure changes to web analytics configurations are done correctly, etc. The problem with this approach is that there are a dozen points of failure that need to be monitored and covered, so this should be used in conjunction with the other two options.
- A backup analytics tool – A second analytics tool could be used to compare and confirm that the data being reported by the primary tool is directionally right. Here is a great write up about why a backup analytics tool makes sense. While this is a good solution, it still falls short in the scenario where a technical team could make identical incorrect changes in both tools. So now you end up with two tools that don’t have the correct data!
- Post-Implementation Checks – A third approach that addresses shortcomings of the first two is to monitor changes made to the configuration and to web analytics tags by running automated audits every day and reviewing the changes. The problem is that no tool on the market does a good job of monitoring such changes in a cost-effective manner. We hope to change that. If you are interested in hearing more, please sign up to get notified when Amplytics launches.
Web Analytics as a Whole
Web Analytics is an essential component of any organization’s tool chest today. It is also one of the harder components to use effectively. In a nutshell, we recommend measuring what can be actioned, keeping the measurements contextual, and maintaining data accuracy. We hope that our review of web analytics strategies has provided you with enough information on how to best leverage a web analytics tool to benefit your organization. We would love to hear from you! Please drop us a line at ThinkingAnalyst@amplytics.com