Bringing Order to Disorder – Understanding Data Discrepancies


August 18, 2020

At Adswerve, we work hard to solve problems for our clients. Really hard. We like to humanize data, support technology with personality and empower our clients to do their best work by providing best-in-class tech with the support to match. The technical issues that our clients bring to us can be straightforward or complex, reactive or proactive, and range from generally benign to downright sinister. Of all the technical issues we see, data discrepancies stand out as one of the few issues that warrant an actual paradigm shift. 

Because discrepancies tend to have more variables and more data sources involved than lesser issues, they can’t be solved in a silo. You’ll need to stay in regular communication with your client, technology partners, publishers and internal team, proactively communicating the issue, sharing your next steps and providing regular updates on progress. 

In addition to working with multiple teams, effectively troubleshooting a discrepancy between sources requires fluency across several platforms. You’ll also need some serious patience. A bad discrepancy is more like a marathon than a sprint and will involve elements of preparation, a thoughtful approach and adequate pacing to push you toward the finish line. Here’s what that entails: 

Phase One: Preparation (Build a solid foundation)

The preparation phase of solving a data discrepancy is all about understanding the problem and building a solid foundation to work from. 

In this phase, you’ll define the scope of the problem, the platforms involved, the severity of the issue and the timeline of when it began. You’ll also determine whether or not the issue is ongoing. This is a crucial phase and it’s important that you:

  • Connect with strong resources that can help across the key platforms where you may not have full access or fluency. 
  • Make sure that your data all has the same scope in terms of dates.
  • Align your conversion sources.
  • Define your key metrics for each platform. 

Remember, you’ll be building on your work and an oversight here can significantly shift your future trajectory, so be methodical and take your time. 

Phase Two: Approach (Start your investigation)

Once you’ve built your foundation, it’s time to plan your approach. In general, you’ll want to keep your investigation broad before going deep. Take a high-level view of all your platforms and see if an early outlier begins to emerge. If your Floodlight data matches up well with Analytics but your Google Ads data is way off, that could be a good place to make a note for a future deep dive. 

This phase is also when you will also want to look into the high-level settings across the platforms like lookback windows, counting methodologies and whether or not conversions are being de-duped across channels. These exercises can help you determine if your settings might be contributing to your issues. 

During this phase, you should flip over every rock and check every box. Are your conversions being scaled and is cross-device tracking enabled? Is your organic traffic getting credit along with paid? Don’t stop looking once you find a good answer; keep investigating and organize your findings. 

Once you’ve completed this phase, you’ll start to see a much clearer picture of which factors are contributing to your issue and which ones you can rule out. You’ll also notice that data sources rarely align 100% and you should begin to develop an understanding of what can be considered a ‘normal’ range and pattern for discrepancies between platforms.

Phase Three: Pacing (Take a closer look)

By this point, you should already be in regular communication with your client as well as the team assisting you across your other platforms. You’ve vetted your reporting and aligned high-level settings across your platforms as best as possible. You should also have a couple of solid suspects for the discrepancy. Your broad investigation is complete and now it’s time to go deep. 

Your deep dive might involve a content tagging audit, conversion testing, more in-depth reporting or spying out tag implementation. Whatever you do during this phase will be more complicated and technical than what came before. This phase is also time-intensive, which is why it’s so important to be thorough in the first two phases. If you’ve laid the right groundwork, you can be confident in your findings and laser-focused on the problem area when you attempt a deep fix. Don’t rush; stay the course and pace yourself.

Phase Four: The Finish Line (Resolve the issue)

By the time the troubleshooting process is in its final stages, you’ve likely learned a few new things and reinforced your relationships with your client as well as with the team that was bold enough to take the journey with you. 

You won’t always find 100% resolution, but by following a thoughtful and thorough path, refusing to cut corners and testing every assumption, you’ll know that you took the problem as close to completion as humanly possible.

Keep in mind that you’re not alone in this marathon. We’re here with best-in-class digital tools and the skills to back them up. If you’re a current client with Adswerve and have encountered a technical issue across the GMP (discrepancies included), reach out to our talented support team and we’ll be happy to help. 

Leave a reply

Your email address will not be published.