How To Debug Reporting Anomalies
Have you ever tried to check the numbers for a campaign your team ran and the campaign code is nowhere to be found? Or check the KPIs for a website only to realize that not even basic visits and page views data has been recorded for the past two months. Still worse, you can pull some data, but right away you can tell that there is no way this data can be correct?
Having seen our share of data anomalies, we have put together a cheat sheet classifying the most common issues into five categories. The cheat sheet also provides debugging techniques, explanation of the possible causes, and recommendations on the type of automated audit steps you can take to prevent such anomalies from occurring in the future.
Download a print-friendly PDF
REPORTS CONTAIN NO DATA (NOT EVEN VISITS AND PAGE VIEWS)
|
||
ANOMALY TYPE |
DESCRIPTION AND POSSIBLE CAUSES |
RECOMMENDED AUDIT |
Generic page tag missing |
The tracking library is not loading on page view. |
Scan the website to identify any pages where the generic tag is not firing. |
Generic page tag not firing |
Other errors are preventing proper execution of the tracking JavaScript. |
Perform additional Scans to catch any fatal JavaScript errors. |
Wrong reporting bucket |
Generic page tag is firing and data is getting collected, but it is getting sent to a different reporting property or collection server. |
Scan the website with checks configured to look for the reporting bucket identifier. |
Browser-specific issues |
While your site might be tagged and sending data to the correct reporting bucket, browser-specific errors may be causing data to be missing from your reports. |
Configure Scans to be ran from different user agents. |
Vendor-specific requirements |
Analytics vendors might have rules for acceptable values for certain built-in parameters. A good example is Adobe Analytics, where the currency parameter needs to follow a particular format, i.e. "USD" vs. "$". Passing a wrongly-formatted value will result in data getting discarded from your reports. |
When setting up Scans for such parameters, enable Regular Expression-based rules for the list of acceptable values. |
Environment-specific issues |
In some government or other highly-controlled environments, network or anti-virus, anti-spam restrictions might result in data not reaching the collection servers. Depending on the type of web property you are responsible for, environment-specific issues can be negligible or really extreme. |
Your only recourse in such cases is to lobby for lifting or making exceptions for some of these restrictions. |
KPIS / METRICS / MEASURES / SUCCESS EVENTS ARE TOO HIGH
|
||
ANOMALY TYPE |
DESCRIPTION AND POSSIBLE CAUSES |
RECOMMENDED AUDIT |
Double-tagging |
The tracking library is loaded twice on the page resulting in more than one tracking request. Another explanation might include tracking functions being called more than once when the users interact with elements on your web property. |
Use occurrence checks inside Scan or Flow tasks to make sure that a specific action is only tracked once. |
Parameter is set at the wrong time |
The intended tracking tag has been moved to a place where it no longer correctly tracks the intended user interaction. A common example for this category is tracking form submissions on form load. In some more difficult to debug scenarios the tag gets hijacked altogether to track unrelated interactions. |
Configure Flow tasks to automate the validation of such scenarios. |
Parameter has wrong value |
In addition to tracking counts, some vendors allow you to assign numeric values associated with KPIs or various transactions. Watch out for commas, dots and other punctuation that may wreak havoc. |
Configure Flow tasks to automate the validation of such scenarios along with rules for acceptable values of specific parameters. |
Bots |
For easily accessible pages (pages that do not require user input) beware of bots hitting these pages and inflating counts. In most cases bots will not be able to executive JavaScript, but as technology keeps evolving, keep an eye out for unexpected increases in traffic that can be attributed to bots. Reports based on IP or User Agent-related information can be a life-saver. |
Perform investigation of this type of traffic using segmentation based on IP, Geography, or User-Agent related info. |
KPIS / METRICS / MEASURES / SUCCESS EVENTS ARE TOO LOW
|
||
ANOMALY TYPE |
DESCRIPTION AND POSSIBLE CAUSES |
RECOMMENDED AUDIT |
Generic tag is firing, but specific parameters are not set |
||
DIMENSION ANOMALIES
|
||
ANOMALY TYPE |
DESCRIPTION AND POSSIBLE CAUSES |
RECOMMENDED AUDIT |
Generic tag is firing, but specific parameters are not set |
||
CAMPAIGN ANOMALIES
|
||
ANOMALY TYPE |
DESCRIPTION AND POSSIBLE CAUSES |
RECOMMENDED AUDIT |
Generic tag is firing but campaign query parameter is missing or misspelled |
||
Tags: Data Quality Whitepapers