tal analytics accuracy: What marketers must know //

Digital analytics reports can be misunderstood as inaccurate. They are accurate, but not exact, in reality. Users don’t understand what analytics is or how it’s gathered. This is the problem. Even worse is the fact that different tools measure things differently, but they are still called the same thing.

This article will examine the nuances of data measurement and how different analytics software work.

Analyzing the nuances of data measurement

The purpose of digital analytics tools was not to be used as sales registers or accounting systems. They are designed to quantify user interaction data and create reports and insights. These tools’ data collection methods over the years have changed. The way data points are measured has also changed.

Let’s suppose that you have changed the imperial measurement of your tape measure to metric. This is a measurement in centimeters. A desk’s length might be reported as being 39.4 inches in one hand and 100 in the second. However, the measurement of the desk has changed.

Switch between analytic programs. You’ll often find that although your numbers might be different, the trend lines are similar. Every tool counts things differently. This is often the case when you upgrade software.

One time, unique users could be counted by adding the total number unique IP addresses that visited a website during a particular period. Organizations eventually started to use proxy servers/firewalls, which required all internal users to access internet using a single IP address. While the number of unique IP addresses didn’t change, it was dramatically reduced in terms of the number of unique users.

To count unique users, you need to combine IP address, browser type and version. Then add a persistent cookie to get a better idea of the number of users. No analytics tool can provide an exact number no matter how many unique users were counted. Tools now take into account other factors when counting unique users.

Get deeper: Data analysis: Your stack’s past, limitations

How to think about your analytics data

Many factors outside of your control can cause analytics software to be imperfect. Users may be blocking cookies and other tracking methods. Internet blips could prevent data reaching the data collection server. Your analytics data can be viewed as a poll of user behavior.

Everybody is familiar with polls during election times. A typical U.S. presidential poll surveys about 10,000 people (or less) of the 150+ million eligible voters (0.006%). When news media report on poll results, they often say that it is within 4 percentage point 4 out of 5 times. This means that it is off by more 4 percentage points 20%.

Most analytics professionals believe that data loss is minimal at 10% for digital analytics tools and less than 5% for most. This translates into data accuracy.

If you had 10,000 visits to your site in a reporting period, but could only collect data on 9,000, your data would still be accurate 99 percent of the time.

This means that 99 percent of your data is correct, and only 1 percent of 100 times is it off by more then 1%. Your data may be accurate but not exact (precise), and it will not match sales records.

This data can be used to identify which marketing strategies — paid ads, sponsored posts and social media marketing, email marketing etc. – are most effective. — which are effective and which ones drive traffic, versus sales.

Go deeper: Don’t let your data be influenced by wishful thinking

Analytics put into action

Analytics data can be very accurate but even a tiny error in precision could cause your analysis to be questioned. This is especially true if the data source differs.

It is important to keep track of the data and compare it where possible. You should investigate any sudden changes in accuracy. Is your website updated recently? Did you properly tag the change to capture the data?

One client added a popup to Shopify after placing an order but before the thank-you page was generated. The analytics tool tracks sales only after the thank-you page is generated.

Although the pop-up was in place, the order went through. However, many users didn’t click through to the message. A large number of sales didn’t get captured because there was no thank-you page. It wouldn’t have been a problem if the popup appeared after the thank-you page.

Here is an example of how Shopify monitors sales and orders. It is easy to see how much data has been lost due to various factors. Shopify’s analytics can be used to track true sales. We can then compare it with data from GA4 and see the following:

Daily variations in total revenue, orders and sales varied from almost 0% to close to 13%. GA4 saw a 5.6% drop in revenue and 5.7% less orders over the 24-day period. These numbers are accurate, especially when they are applied to marketing efforts to determine what brought the user to the site.

Is GA4 allowed to be used by this company for reporting sales? 100% no! Accounting software is designed to do just that.

There are many ways to send data directly to analytics tools (server-side) if your organization requires more precise data. This eliminates the need for cookies and user browser issues.

Although sales data is more precise, some soft metrics of user interaction (e.g. scroll tracking) may not be as accurate. This method is difficult and takes time to implement in most companies.

Ask yourself: “Is this extra effort necessary to capture an additional 2-5% sales revenue in my analytics reporting?”

Understanding your analytics data

Everybody needs to believe in their analytics data. It is important to ensure that your analytics software is properly installed and configured. It can’t capture all the information.

The analytics software simply creates a poll with an over 90% sample size. The results are highly accurate and on target, if not exact (actual numbers).

The post Accuracy In Digital Analytics: What Marketers Need to Know appeared first at MarTech.