This is part three in my introductory series about web analytics and digital analytics. You can find part one about data driven decision making here and part two about terms and definitions here.
How does it work behind the scenes?
The most commonly used method to track data in web analytics is via javascript (“page tagging”). Though, some other methods exist like logfile parsing, which is often used to turn machine-generated data into visualized information. In addition to that, KISSmetrics is a good example of an analytics tool which offers libraries to directly connect to their servers via the backend of your software (e.g. a web shop system).
This is how the most widespread solution via javascript generally works:
Once a visitor views your website in a browser, the embedded javascript tracking code triggers the call to the analytics servers. For the geekier people it is worth mentioning, that it is generally not the javascript itself, but a so-called tracking pixel generated by the javascript which is transmitting the data. Afterwards, all data is processed and aggregated, before an analyst can retrieve the gathered information for producing reports, insights and recommendations. This way, pageviews and visits get recorded.
As stated in part two of my series about web analytics, pageviews do not represent a good metric for modern web applications full of javascript / ajax features, which modify parts of the website without reloading the browser. This behavior does not trigger the standard tracking code, hence these pageviews do not get tracked. Therefore, many web analytics tools offer additional tracking abilities which are not tied to a page loading but events like clicks on some certain element of a website (e.g. called event tracking in Google Analytics). I will cover more about this issue in further posts.
Data accuracy in web analytics
When making analytical decisions, it is no surprise that you should trust your available data. Although there will be situations when you need to reconsider data integrity on a regular basis, you should have made all necessary (setup) tests in advance. This way, you can be sure to trust the numbers in your tools. You waste a lot of money with non-reliable data. But even worse, you might take wrong decisions based on incorrect representations of reality.
However, here is a lesson you should take: most of the times, web analytics is about trends. It is not uncommon to experience deviations of around 10% if you compare numbers with other data sources like a transaction database of a web shop. There are plenty of reasons, from technical issues to different underlying fundamental assumptions between distinct tools. Often, it is not necessary to have exact numbers for comparing different conversion flows / versions / channels / events / … / as long as the basic population is sufficiently large. Follow the saying: “It is better to be approximately right than precisely wrong”. Though, when data discrepancy gets too high, you should double check all installed tracking codes used for web analytics. Again, I need to point out that a thorough pre-examination of the whole conversion flow, including its tracking, is obligatory.
Marketing analytics
Even though marketing analytics is not the focus of this introduction, let me emphasize why it is one of the most important topics once you start (paid) customer acquisition. From a technical perspective, it is not much different to web analytics, but the metrics, findings and interpretations can get more complex. In today’s multi-channel marketing, it is absolutely normal that different channels will differ in results. Not a big surprise, when we keep the distinctions of users’ intentions on different websites or devices in mind.
KISSmetrics timeline feature, © Space Pencil, Inc.
For instance, traffic from social networks is usually less likely to convert into paying customers when selling goods. Many big companies use social media for brand building and customer interaction rather than direct selling. On the contrary, a really smart social campaign might still outperform other channels if done correctly. Another example: even if a specific campaign leads to more initial conversions than another, the latter might still best in terms of retention, e.g. repurchases or engagement. Calculating the effective return on invest is worth a book on its own. Marketing analytics and controlling is both, invaluable and indispensable if you do not want to throw money down the online drain. If you fail completely, you might never reach a positive return on invest. This is what marketing analytics is all about. As stated in my first post in this series, web analytics and marketing analytics share quite some similarities and activities performed in both – many people, even analysts, do hardly differentiate at all.
What’s next?
This was part three of our introductory series about web analytics. You can find part one in the series here: Introduction to Web Analytics and Digital Analytics, and part two here: Web Analytics: Basic Terms and Definitions. Let us keep in touch for more posts about this topic.
If you liked what you read, subscribe to this blog via RSS or email (see below) to keep updated when the series continues. A comment below would be fantastic as well. Upcoming posts will include basic step-by-step setup-guides to well-known tools, information about advanced web analytics and marketing analytics topics like cohort analyses and KPI reporting automation. By following these posts, you will learn how to use analytics to improve your business, generate more conversions and make more profit. Additionally, we give away free tips and online-marketing secrets for you. Do not miss it!
Subscribe to our free newsletter:
The post Web Analytics: Data Accuracy and Technology appeared first on aboutVentures.com.
