Why do I often miss 20-25% of Google Analytics conversions or transactions?
I have noticed several times for myself that goals that are set up, again and again, are not tracked in Google Analytics, although other systems (own customer systems) clearly show the achievement of goals. This article is about how such discrepancies occur.
Missing conversions: The Google Analytics tracking code is not integrated on the landing pages
Of course, this is more of a fundamental problem. If the Google Analytics tracking code is not installed, no conversions can, of course, be measured. It is possible that the integration of the tracking code is overlooked even if the page is revised or changed, and differences then arise in the further course.
A tool like Screamingfrog can be used to check that it has been installed correctly. You select the custom filter configuration in the Screamingfrog and then, in a second step, display all pages that do not contain a Google Analytics tracking code. You can identify the UA number in the administration under property settings.
Missing conversions: Only the first goal counts
Depending on how you configured conversions in the Google Analytics data view, there may be deviations in reality. There are fundamentally two variants of conversion measurement. The first variant is via the definition of target projects, the second via the implementation of e-commerce or enhanced e-commerce tracking.
If you have integrated goals, a measurement is triggered once per session and configured goal. Since you can configure a maximum of 20 goals per data view, a user can only achieve a maximum of 20 goals per session.
Example PDF downloads: A user loads 6 PDFs onto his computer in one session. You have defined the download of a PDF in Google Analytics as a goal; then you would only log the first goal achievement in later evaluations.
Example of multiple seminar registrations in one session: Let’s assume in our case that someone and their colleagues register for a seminar with us, then only the first target achievement is measured, although I would actually charge two seminars.
By the way: Should you ever plan to take the Google Analytics Qualified Individual (GAIQ) exam – I’ve been running the Google Analytics Bootcamp for 3 years – and this question almost always comes up in the exam.
Measurement via e-commerce tracking
When measuring via (enhanced) e-commerce tracking, several transactions can be measured within one session. That means, should you book a seminar with us for you and a colleague within a session, this would be counted as a target achievement for target projects, but as 2 transactions in e-commerce. Since we have integrated both in our Google Analytics account, we came to 9.94% more transactions (e-commerce seminar registrations) than counted target projects (seminar registrations) for the second half of 2017, for example.
Missing conversions: manual entry of transactions
Sometimes you have to enter generated contacts/leads/registrations manually depending on your own business model. Example: As a seminar provider, we occasionally receive registrations by fax or letter. So that these are then transferred to our other systems, we have to enter these registrations into the system manually. Since we exclude our own IP address, these transactions are also not measured in our case that makes a difference of about 5%.
Missing conversions: Users use browser add-ons to deactivate Google Analytics
Google Analytics offers users a browser add-on or an opt-out cookie that allows users to better control which data is recorded by Analytics from the pages they visit. This is what it literally says on Google:
Furthermore, your users via AdBlocker or tools such as “block yourself from analytics” are not counted by your Google Analytics implementation.
Missing conversions: Tracking code is implemented incorrectly
The problem is as follows: Old recommendations for the correct installation of the Google Analytics Tracking Code (GATC) often recommend installing the tracker at the end of the source code. The reason here was an optimization of the page loading speed. The result of this implementation was that due to the late execution of the tracking code, some user transactions or target attainments were not even measured by Analytics because users left the page before the GATC was executed. Even if this happens less and less, it can still be a possible reason why you are logging fewer target achievements.
With the new asynchronous tracking code from Google, the source code’s late positioning is no longer necessary. Google itself suggests a split implementation of the asynchronous code (see also the following screenshot).
One part of the code goes into the head; the rest of the tracker goes into the body. Overall, according to estimates by LunaMetrics, the change from the old tracker to the new asynchronous code resulted in “data improvements” of 5%, in some cases, even up to 20%.
11 recommendations for handling data & tools
In sum, all of these different influencing variables can lead to a reduced number of actual conversions compared to measured conversions. In several discussions about this, the deviations are usually 20% -25% fewer measured target attainments. My recommendation here:
- Don’t trust any numbers without knowing how they came about.
- Know what effects different measurement methods have on the analysis (e.g. logfile analyzes versus Google Analytics).
- Data from the various systems are not a representation of reality.
- Data is interesting when put in context, not absolute.
- Context can be, for example, a comparison with the previous year or a comparison of different sources.
- Data must be seen as indicators and always checked against reality.
- Initially invest a lot of work in data quality (shit in -> shit out).
- Always try to reduce sources of error.
- Verify key performance indicators via other systems and merge them with various data sources.
- No debates about which analysis tools are “right”. The technical background usually explains the differences.
- Commit to a tool within the company that depicts reality.
Do you also have these effects? How high are the deviations for you? Do you know more reasons why there are deviations here? Then we look forward to hints in the comments.