Your Adobe CJA Implementation Is Complete. But Do You Trust the Reporting?

A completed Adobe Customer Journey Analytics (CJA) implementation doesn't necessarily mean you have reliable analytics reporting.  Heavey Digital Consulting recently worked with a major healthcare system where that distinction became painfully clear.

The client had migrated from Google Analytics 4 (GA4) to Adobe Customer Journey Analytics through another agency roughly a year earlier.  We were brought in through an agency partner to assess the existing implementation, recommend improvements for a minimum viable product (MVP), and implement the approved changes.

My initial assessment wasn't particularly encouraging.  Based on what I found, I would have given the CJA setup a grade of 75%.  Passing, but barely.

After getting further into the data and working through the MVP improvements, I'd lower that grade to no better than 60%.

Here's why.

The Data Was There.  The Reporting Wasn't.

Part of my initial grade stemmed from the fact that the underlying Adobe Experience Platform (AEP) data collection implementation was fairly good - the client didn’t implement a data layer as recommended, so there was a good deal of DOM scraping and use of URL-based conditions in Adobe Tags Rules.  There was a substantial amount of usable data being collected through Adobe Web SDK.

The problems were largely in how that data was being used within CJA.

The client had hundreds of dimensions and metrics available in its data views, many of which were irrelevant, redundant, or had no data whatsoever (over 300 in this category).  Each production data view contained more than 400 components, making it unnecessarily difficult for analysts to find what they actually needed.

Marketing Channel attribution was another major concern.  The existing derived fields didn't adequately account for the client's marketing activities, internal referrals, or the wide variety of UTM values being collected.

Page names and URLs also needed normalization to make reporting more useful, and several important conversions weren't defined as reusable metrics.

These were substantial problems, but the deeper investigation uncovered more.

Nearly One in Five Page Views Was Bot Traffic

This was probably the most striking discovery.

The implementation was already collecting Adobe bot detection data, but that information wasn't being used to exclude bots from normal reporting.

Looking at approximately 53 weeks of data, roughly 17% of all page views were identified as bot traffic.

Think about that.  Nearly one in five page views in the reporting environment wasn't attributable to a human visitor.

The solution wasn't to discard the bot data altogether.  We configured the reporting environment to exclude identified bots from the primary data views while maintaining a separate view where that traffic could still be analyzed.

Campaign Attribution Had More Problems Than Initially Apparent

The original Marketing Channel configuration needed substantial rework, but we also found issues with the campaign data itself.

For example, numerous landing page URLs contained parameters such as `campaignId`, `adGroupId`, `rid`, and Google Ads-related parameters, but lacked standard UTM campaign values.  Our partner agency, which runs most media for the client, was notified and is looking into the issue.

Were these actual marketing campaign visits that weren't being properly attributed?  In many cases, that appeared likely.

Complicating matters, the AEP schema didn't explicitly capture common advertising click IDs such as `gclid` and `fbclid`.  This became a potential issue since the client is implementing Salesforce/Data 360 and this data is not readily available to port from AEP to those platforms.

We updated the Marketing Channel logic, standardized the classification of UTM values, and added derived fields to help identify campaign-related traffic that might otherwise fall through the cracks.

This doesn't magically correct missing campaign tagging, but it does provide considerably more visibility into the problem.

Session Configuration Was Another Issue

Healthcare websites present some interesting challenges for digital analytics, particularly when visitors move back and forth between the marketing site and patient portals that are not tracked by the analytics platform.  There is also a significant amount of traffic that navigates from the marketing site to an external career site and back again.  

We found that the existing session timeout could fragment some of those journeys, potentially affecting session-based reporting and Marketing Channel attribution.

Our analysis showed that increasing the timeout to 60 minutes would reduce these reported “return session starts” by about 30%, while reducing overall Session Starts by only about 2%.

Based on that analysis, we recommended and implemented the longer timeout.

Cleaning Up the Reporting Environment

Not every improvement involved identifying a major data quality problem.

Some were about making CJA useful to the people who actually work with it.

The MVP included:

- Cleaning up data views by removing unnecessary components and improving the remaining reporting dimensions and metrics.

- Reworking Marketing Channel attribution to better reflect the client's marketing programs.

- Creating and refining derived fields for page names, URLs, campaign information, and other reporting needs.

- Establishing calculated metrics for important conversion activities.

- Developing a way to identify and classify form submissions using confirmation pages, since the existing Form Name field wasn't populated.

- Separating identified bot traffic from normal production reporting.

- Adjusting session configuration to better accommodate actual visitor behavior.

None of these improvements required starting over with AEP data collection, and that's an important point.

The Takeaway

Organizations can spend considerable time and money implementing an enterprise analytics platform and still end up with reporting that doesn't adequately support their business needs.

Sometimes the problem isn't the data being collected.  It's the configuration, attribution logic, reporting structure, or simply a lack of attention to how the organization actually uses the platform.

In this case, the underlying data provided a solid foundation for making significant improvements without rebuilding the implementation.

At Heavey Digital Consulting, we specialize in Adobe Experience Cloud technologies, including AEP, CJA, Adobe Analytics, and Adobe Journey Optimizer.

If your organization has implemented CJA but isn't confident in the reporting, it may be worth taking a closer look at what's happening between data collection and analysis.

A completed implementation isn't necessarily a successful implementation. If you’re unsure of whether your CJA implementation is where you need it to be, click the button below and choose the option Assessment/Cleanup of Existing CJA Implementation in the What kind of help are you looking for? field.