“How many transactions did we have last week? "Let's see… Our back office says 12,000, our data warehouse says 15,000 and Google Analytics shows 11,000. "How is that possible? If different systems register and report transactions (such as back office systems, data warehouses and Web Analytics tools), you would expect the same number of orders to be registered everywhere. In practice, the numbers often do not match. Where do these differences come from?
Deviations in measuring the number of orders can have various causes. Most of them make sense, some are due to limitations in measuring instruments and occasionally a measurement error can be made.
When numbers differ per system, you may well lose confidence in your systems. But don't worry: the solution often lies in checking the definitions of an order in different systems.
In addition, depending on the system you use, a small difference is normal. Moreover, even with a small deviation, the numbers of orders still provide insight into trends.
What is deviation?
Deviation means that system A and system B report deviating numbers. Questions we regularly hear are:
- "Web analytics says there were 10,000 orders, but our back office says there are 15,000. How is that possible? "
- "We see a deviation between different systems of more than 20%. Is that normal?"
Where do deviating transaction numbers come from?
Deviations in transaction numbers can have various causes. We divide these into 4 categories:
1. Differences in transaction methods
Transactions from other channels: web analytics only measures transactions on the website, while another system also records call center transactions or transactions in physical stores .
Cancelled transactions: Users place an order, but don't meet the acceptance criteria or cancel their orders. This order will be removed from one system, but not from the other.
Test transactions: test transactions can be registered in one system but not in another.
Unclear funnels or definitions: sometimes it is not clear what does and does not count as a transaction. Think, for example, of placing an order on behalf of a customer, a sub-order or a repeat order.
2. Storage settings (from web analytics)
Cookie settings (especially since the introduction of the GDPR): when websites use an 'all-or-nothing cookie offer', it is possible that the behaviour of users in a website is not registered at all. Note: this doesn't happen often.
Data blocking/redirection: web analytics data is sometimes blocked, redirected or tagged for certain IP addresses, for example those of one's own organisation. This is usually visible in the user interface in an analytics tool. It might go undetected like in Adobe's VISTA.
3. User Preferences and Settings
Adblocker: some users install adblockers in their internet browsers. Commonly used ad blockers sometimes block tag managers, which means that tracking and/or web analytics software can no longer be loaded. This must be set manually by users.
In addition, it may be that adblockers block commonly used functionalities of web analytics tools by themselves.
Blocked by browsers: Intelligent Tracking Prevention (ITP) by Safari blocks certain cookies and easy referrer access.In addition, other browsers now also offer privacy functionalities, such as anonymous browser windows.
Mobile browsers with data saving features: in order to limit data usage on mobile phones, some mobile browsers block tracking functionalities.
Slow internet connection: if a user has a slow internet connection, some tags will not load or scripts will be blocked.
4. Functional deviations due to technical errors
Implementation errors: when you build new functionalities, they must be seamlessly implemented within the existing system. This can be a challenge, because a small mistake can have major consequences. For example, if there is an error in the funnel a conversion page will not be loaded and therefore a transaction will not be saved.
When you see more transactions in one system than in the other
Is the transaction data in the different systems you use inconsistent? Then take another look at your definition of a transaction. Is it the same in every system?
Based onthe above four categories you can explain almost all deviations. Do you think there should be another category or are you unable to figure it out? Contact us: we are happy to screen the issue together!
Former colleague Shengnan Lu was a Data Solution consultant at Digital Power
Shengnan has extensive experience with both the technical and psychological sides of web analytics. He is dedicated to the the emotional and functional connection between data-driven insights and business strategies.
Receive data insights, use cases and behind-the-scenes peeks once a month?
Sign up for our email list and stay 'up to data':