Have you ever wondered how reliable are the data that you see in a Google Analytics account rather than on an advertising platform like Facebook?
Do you trust 100% of the sessions, clicks or conversion rates reported?
Well, think about that web analytics systems are flawless is Utopian, but how can you stay calm and serene take decisions based on reliable data?
In this article I will try to share with you the reasons why those who have an approach data driven should have both eyes open and activate all its receptors to identify and avoid the pitfalls that the digital world – the web analytics – a day today hides.
Google Analytics: Free, but with its limits
Let’s start with the web analytics tool most widely used and known, especially thanks to its accessibility and ease of installation and use.
As you may know, the data provided by Google Analytics free version is sampled if they exceed a certain volume. In other words, for sites that generate a lot of traffic it is likely that the reported numbers are not precise, but the statistical estimates.
In particular, if you want to analyze a period in which statistics on the site more than 500 thousand sessions, the data will be sampled they were recorded. In these cases, the tool also tells you the level of accuracy of the data, bringing a message such as “This report is based on XX% of the sessions.”
The Google servers will be many and powerful but not unlimited!
In the free version there is indeed a hit limit (i.e. iterations that are drawn from the instrument) beyond which the data is sampled.
This threshold is, to be honest, very high (up to 500 hits per session and 10 million hits per month for properties) and consumption as you can control from the account administration panel.
What can those lucky enough to have a website that generates more traffic than what Google allows you to properly track to have reliable data? Pay, going to the premium version of the tool, or Analytics 360°. Or, become aware of this limit.
Users or bots? The problem of spam
As already told some time ago in this my other post, the reliability of Google Analytics data has been “affected” by various forms of very subtle spam.
The referral spam, the “ghost tours” or the newest “language spam” (the one with the message “Vote for Trump!” That appears in the report on languages), are the best-known examples of how it is not all gold that glitters.
In this case, however, waiting for Google engineers definitely pose a remedy, the solutions to make the most reliable site data and clean them from spam there.
It is the responsibility of those who analyze the stats make sure that the right countermeasures are taken.
You may also like to read another article on BSOinvest: How To Choose A Good CRM For My Company?
Which tool is telling the truth?
You often also run into the same action data recorded by two different platforms that provide widely divergent numbers between them.
I give you an example, maybe it happened to you.
Advertising campaign on Facebook: banner with links pointing to a website landing page in which were used the UTM.
Google Analytics shows X accesses the landing page from that banner, Facebook says there were Y-click on the ad. With X far from Y. Not to mention the possible conversions.
Who has more right? Who is reliable?
Questions to which it is difficult, if not impossible, to give an answer that is shared by everyone. You have to know first of all how the different platforms record and report data, or which attribution model is used, or the length of time of the cookie.
In short, also in this case, the knowledge and skills of the web analyst make the difference.
AdBlocker, these (s) known
Research by May 2016 has brought to light some very interesting results on use of AdBlocker systems.
These are applications that block the browser commercials of a site or a search engine and they seem to be used by 13% of users who navigate to your desktop and 7.6% of those sailing from Mobile.
You think, yes, but what to do with the data?
Well, there are evolved AdBlocker that even block the tracking of the website that the user is navigating.
So you can imagine yourself that since this is a phenomenon not yet widespread (more abroad), but evolving, It can really undermine the consistency of the data being collected and analyzed in the web analytics tools.
In other words, the risk is to lose because of some visitors and make the feedback on underestimated data.
Unfortunately, the remedies in this case are very complex and technologically difficult to implement.
But we can monitor more or less easily how users navigate the site with this system activated and take note.
E-commerce transactions: When the numbers do not add up
Finally, a last case very common when it comes to e-commerce (both online shops, but also booking), concerns the issue on transaction numbers and revenue that will never return with what the company records in its management.
In fact, it happens frequently that the web analytics tool adopted less transactions than actual records, with a direct result on the reliability of performance metrics such as conversion rate and ROI of a campaign.
If it turns out such that only half of the purchase of an online shop are properly collected by Google Analytics, how you can make sensible assessments on campaign performance rather than on the site’s ability to convert?
For this type of problem the cause is almost always due to the way in which the payment systems are integrated into the web site , when in fact the user leaves the site to go to PayPal or the banking gateway site to complete the purchase , it may be that he does not return to the site where it shows the final confirmation of the order and the data is recorded on Google Analytics.
Especially for the payments to be made by PayPal users “most experienced”, it is very frequent.
Even in this case, it is the good analyst of digital data to be aware of this problem, evaluate the extent of the problem with cross-checks and suggest suitable remedies.
In short, as you can imagine, there are often no solutions for possible pitfalls of consistency and reliability of digital data. But already become aware of and make all possible precautions to the tools we use for analysis can make a difference.