Conversion Tracking Errors That Are Ruining Your Marketing Budget
In 2024, the Silesia Performance Lab team checked exactly 47 Google Analytics 4 accounts belonging to companies in the IT and professional services sector. Our data shows that in 39 cases, the configuration was incorrect, leading to business decisions based on false numbers. This article shows where your money is leaking.
Event Duplication: Why you see twice as many leads as you have in CRM
During an audit conducted in May 2024 for a software house in Gliwice, we discovered that their system counted the submission of a contact form twice. The problem was that the event was generated both when the 'Send' button was clicked and after the automatic reload of the thank you page. As a result, the marketer reported 84 leads per month, while salespeople saw only 42 real inquiries in the system. This is a classic error that makes the cost per acquisition (CPA) seem half as high as it actually is.
At Silesia Performance Lab, we check facts and eliminate such technical errors within the first 4 hours of an audit. Event duplication most often results from old tags in Google Tag Manager overlapping with new scripts implemented directly in the site code. The average measurement error in the companies we studied in Katowice was 22%. This means that every fifth zloty spent on advertising was allocated based on incorrect data. Fixing this problem requires precise settings for exclusion rules and unique transaction or event identifiers.
It also happens that users refresh the 'Thank You' page after submitting a form, which GA4 interprets as another conversion. In one technology company we have worked with since September 2016, this single error inflated results by 14% for over seven months. We implemented a session-based repeat event block, which immediately straightened out the statistics. The data speaks for itself: without clean tracking, you don't know which campaign is really paying the rent on Chorzowska Street and which one just looks good in the tables.
As many as 39 out of 47 audited GA4 accounts in 2024 had critical conversion counting errors.
Payment Gateways and the 'Referral Exclusion' Error
If you sell subscription services or IT training, your biggest enemy in analytics are systems like Stripe, PayU, or PayPal. Without the correct configuration of the referral exclusion list, GA4 attributes sales to the payment gateway rather than a Google ad or a blog post. In July 2024, we analyzed the case of a training company where 68% of conversions were attributed to the source 'payu.pl'. Because of this, the owner thought SEO wasn't working and wanted to cut the budget for content that actually generated most of the traffic.
We solve this problem by adding the payment operators' domains to the 'Unwanted Referrals' list in the data stream settings. This is a task that takes exactly 12 minutes and can completely change the picture of a company's profitability. At Silesia Performance Lab, we do technical SEO that works, but without correct analytics, we are unable to prove to the client exactly how much they earned thanks to our actions. We always start a specific audit by checking if your purchasing path breaks when the customer leaves for the bank.
It's worth remembering that every new payment method added to the service requires updating these settings. We often encounter situations where, after implementing Apple Pay, the statistics suddenly 'go crazy' because no one added the new domain to exclusions. Our team in Katowice watches these details during every implementation. The average response time for such anomalies in our fixed contracts is 2h 14min. Thanks to this, our clients do not lose data continuity even during large changes on the site.
Ignoring Internal Traffic and Employee Visits
In small and medium IT companies, where the team counts from 10 to 35 people, traffic generated by one's own employees can destroy data reliability. Developers, testers, and salespeople enter the site dozens of times a day, testing new functions or checking prices during a conversation with a client. If their IP addresses are not excluded in GA4, the bounce rate and average session duration become useless metrics. For example, in March 2024, a company from Sosnowiec had a session duration of 12 minutes, but after excluding the office, it turned out that real customers spend only 48 seconds on the site.
Excluding internal traffic is a foundation. At Silesia Performance Lab, we use filters based on fixed office IP addresses and special URL parameters for remote employees. This allows separating technical tests from real user behaviors. During an audit in Q2 2024, we noticed that for one of our clients, as much as 19% of all 'events' came from the QA department, which was checking the service calculator's functionality. After cleaning the data, the real conversion dropped from 4% to 2.1%, allowing the owner to understand that the site needed usability improvements.
Data must be clean because you plan next quarter's expenses based on it. If you see artificially inflated traffic, you might get the impression that your brand is more recognizable than it actually is. We do technical SEO that focuses on acquiring new customers, not counting your admin's visits. We check facts and configure filters so that every session in the report comes from someone outside your organization. This is a simple change that drastically increases trust in monthly reports.
Excluding office traffic allowed one of our clients to discover that real customer session time is 48 seconds, not 12 minutes.
Custom Event Configuration Errors
In GA4, everything is an event, but not every event is configured correctly. Technology companies often rely on automatic tracking ('Enhanced Measurement'), which doesn't catch specific interactions, such as clicks on 'Request a Demo' buttons based on AJAX technology. In August 2024, we helped a SaaS company from Katowice that didn't see 40% of signups for the test version because the analytics script didn't wait for the server response. A specific audit allowed us to track down this error and implement a dedicated event triggered by the DataLayer.
Using raw data without first organizing it is a direct path to chaos. At Silesia Performance Lab, we emphasize correct event and parameter naming. Instead of a generic 'click', we configure parameters like 'button_location' or 'service_type'. Thanks to this, in the reports, you see exactly whether the customer clicked on the hosting offer or the programming services price list. Since September 2016, we have implemented over 1,200 such dedicated tracking schemes for clients across Poland.
We don't promise miracles, but we guarantee that after our intervention, you will know exactly what users are doing on your site. A common mistake is also not tracking errors in forms. If a customer enters the wrong phone number and the form doesn't send, GA4 won't register this by default. We set up tracking for such messages, allowing us to quickly catch places where customers 'drop out' of the process. These are the specific profits we talk about – reducing losses in the sales funnel by a few percentage points can bring tens of thousands of zlotys in savings annually.
Lack of Consent Mode v2 Implementation and Marketing Losses
Since March 2024, Google has required the implementation of Consent Mode v2 for all companies using advertising systems. Many IT companies ignored this topic, resulting in their remarketing lists in Google Ads stopping building. During an audit for a client in the cybersecurity sector in September 2024, we noticed that their bidding algorithms were working 'blind' because they weren't receiving signals about user consents. This led to a 37% increase in cost per click in just two months.
Correct implementation of consent mode allows GA4 to model data from people who did not agree to cookies. This way, you don't lose 100% of information about these users, only their personal identifiers. At Silesia Performance Lab, we integrate tools like Cookiebot or OneTrust with your GTM in a way that doesn't slow down page loading. Our tests on 14 different installations in 2024 showed that correct modeling allows recovering an average of 65% of 'lost' conversion data.
Ignoring legal and technical aspects of privacy is today's biggest risk for marketing budgets. A specific analytical audit must include checking if tags trigger only after consent is given, but also if they send the appropriate parameters (ad_user_data, ad_personalization). If your competition did it and you didn't – their campaigns will be optimized for real results, and yours for random clicks. In Katowice, we have already helped 18 companies go through this process without losing a single day of campaign continuity.
