Home / Analytics  / When Good Analytics Goes Bad
analytics line graph

Try as we might, sometimes data tracking doesn’t work the way we expect it to. I sat down with our technical account manager extraordinaires, Roberto Reyes and Abhiram Bhagwat, to ask about the most common mistakes they see with pixels, data tracking, attribution and what they recommend doing to avoid errors in tracking that could result in inaccurate reporting.


Problems with Pixels


Pixels are the backbone of analytics, but they are not as simple to implement as it seems. Multiple vendors use different formats for their pixels, and require different implementations, even if they’re tracking the same event. When implemented improperly, pixels may not fire, misfire or cause a conflict with another pixel.

The knowledge of the person implementing the pixel is often the most important factor in making sure it works properly. Sometimes a developer will assume one pixel is the same as another pixel, and put it in the wrong place. Or sometimes the person implementing the pixel will assume one pixel has similar installation requirements as another, and not install it properly. Pixels aren’t cookie cutters, so we see the best outcomes when the account manager and tech team work together to make sure the pixel is installed properly in the right place.


Deciding What Data to Track


Pixels are one piece of the entire data-tracking system, but they are not the most important element. The most important part of data analytics and pixel placement is figuring out which events to track in the first place. Some clients want to track everything, which results in data overload. Other clients only want to track the sale, but that doesn’t provide the full picture of how customers move through the conversion process.

We recommend finding a middle ground between tracking every micro-conversion and tracking only the final conversion. Part of that decision stems from strategy. Abhiram and Rob prefer to see a strategy for each channel that will be tracked to make sure there is consistency between naming conventions and tracking processes for each.


Attribution Varies Across Platforms


In the old days, clicks were attributed only to the last source of the conversion. Now there are multiple attribution models, and multiple platforms on which to view them. Further compounding the challenge is the fact that each platform sees the data slightly differently.

Each person viewing the data needs to understand how to read it and how to compare two different data sources. Don’t try to compare exact numbers, because they won’t always match. But you should be able to see similar trend lines on each platform with the same attribution model. If one is out of whack, then you know there’s a problem with the tracking — which brings us to our next point.


Marketing Automation Headaches


Marketing automation is the hot topic in analytics circles, but it’s easier said than done. The key to marketing automation is consistency. For example, if the team can identify patterns in URLs, then they can create rules in Google Sheets to manage your tracking. If each URL follows a different pattern, then everything must be manually created.

Obviously, marketing automation saves time, money, and effort, but you must consider your data tracking needs from the beginning of the development process to achieve that goal. Pixels should not be the last piece you insert. If you build in scalable hooks, everyone will be happier.


Communication and Planning Save the Day


Communication is the biggest area of failure that Abhiram and Rob encounter. Always maintain communication with the technical account management team. They can help you figure out whether changing something in the middle of the path will break the tracking at the end. They can also help you ensure that what seems like a minor change won’t disrupt the whole process and make troubleshooting easier. For example, if a developer changes a function, but the same function is used in other areas of a site, it could affect how other pixels fire and the analytics team may not be able to quickly identify that as the source of the error. However, if you tell them the plan before making the change, they could tell you what to watch for.

Planning is the second component of effective communication. Use a project plan for every campaign and make sure analytics is a part of that plan. Talk early and often about the plan and what elements need to be considered.


Back-up Systems Save the Data


Each client needs multiple data-tracking systems as a back-up. Back-up systems help in two ways:


  • Ensuring data doesn’t get lost
  • Identifying the source of the problem if one system is different from the others

    By having a backup, you can quickly see if a pixel is misfiring on a single system. You can also verify reporting to ensure the rest of the data tracks properly, or fill in the errors produced by that misfiring pixel.


    Above All: Be Vigilant


    Abhiram and Rob’s top tip is to remain vigilant. Analytics is not a set-it-and-forget-it enterprise. One small change can change everything. Avoid errors by:


  • Conducting regular audits, quarterly at a minimum
  • Setting up rules and alerts so you will know the second an error occurs

    In a perfect world, analytics would always work seamlessly, but as long as people are managing the process, there will be errors. We hope these above tips will help your team avoid some of the simplest mistakes that make good pixels go bad.

    Aryn is a Sr. Director of Earned Media and Creative Services at The Search Agency, where she’s worked for eleven years. Her previous experience includes managing the Creative teams, developing new business, creating client strategies, developing new Earned products and tools, being the lead project manager, overseeing workflow for Paid and Earned, implementing new strategic processes, leading client intake process, and developing internal efficiency projects.

    1 Comment
    • Kate

      Thanks for posting! It seems that you need a good plan and strategy in order for things to work. Patterns definitely make it easier to recognize what’s going on. We’ve recently started using GetResponse for marketing automation and they also have analytics there.

      October 16, 2017 at 9:15 am Reply
    Post a Comment