5 Common Roadblocks to Great Data
As a Digital Analyst in an organization with no enterprise digital analytics solution, you can easily feel left up a creek without a paddle. This article will explore five of those common issues that leave you with untrustworthy data and hampered capability. In the next post (stay tuned!) we’ll review how Google Analytics 360 allows us to overcome those issues.
#1: Incomplete Data
Free versions of analytics platforms often leave you with sampled data or, otherwise, incomplete sets of your users’ clickstream behavior. These holes are filled using sampling algorithms that can lead to big discrepancies between reality and your measurement. This is particularly problematic as you begin to try attributing on-site conversions, like purchases.
There are techniques you can use to try to reduce sampling—by using the GA reporting API. There are tools you can buy that use these tricks in the API if you’re not a developer. This approach has the analyst operating in a very mechanical way, pulling reports and culling through XLS/CSV output files.
When exactly does “analysis” occur when all the time is spent managing a Rube Goldberg-esque ETL to get your teams the data they need? The sad truth is we often put aside our analyst hat, roll up our sleeves and switch to a data engineer role. Both are important functions of our job but can be disparaging if the balance is heavily skewed in one direction.
#2: Data Silos
Many of us inherited data and legacy platforms that are disjointed and not unified. We can be asked to answer questions that require these datasets to be joined, only to find out we have no reliable join key. This unwinds the trust in your data organization because the complexity of join-keys and data synchronization are not a concern for the business and product teams!
There are techniques that can done to try to join data together in aggregate, perhaps using the date as a join key, this is largely making approximations and correlative attribution of metrics; and will sometimes not be enough to make assured proclamations in terms of business impact.
#3: Updates Require Code Push
Measurement is always incomplete because your business is always changing. This forward progress means your data measurement framework needs to move at the speed of business.
If you are having to update your measurement using code pushes then it can be difficult to keep up. Development cycles and product sprints can take weeks, if not months, to go live. If your stakeholders aren’t getting you measurement requirements early, then there can be campaigns that flight without the proper tagging.
This is another form of missing data, or incomplete data. There are holes that form as the demand for measurement exceeds the ability to instrument it into the website. These holes are front loaded, and in marketing that is bad because the acquisition strategy is also front loaded. This is a compounding effect that leaves the most important segment of campaign traffic unmeasured while we are waiting on dev and sprint cycles to go live.
#4: Limited eCommerce Data & Attributes
If your website generates revenue directly, through eCommerce or publishing, then it’s critical to capture fine-grained measurement to be able to answer increasingly sophisticated business questions coming from your team like, “What product categories are viewed most often on mobile devices?” Without enhanced tooling and data attributes in your measurement framework, you wouldn’t be able to answer those questions.
Analysis of eCommerce, or revenue generating conversions, are becoming increasingly personal and granular, with more and more demand to make micro decisions to optimize conversion paths that have gone through countless iterations of conversion optimization over the years. The demand for this data is outpacing the free tier analytics capabilities.
#5: Limited Access to Data
It wouldn’t be surprising if you’re having issues with accessing the data you need to answer the questions that are being asked of you. This is a problem that exists universally, and is compounded when an organization has not executed a formal data access-controls and governance initiative.
We already know that inheriting legacy systems is a common struggle, unfortunately these ownership transitions often do not include proper administrative provisioning!
If your organization has financial and legal obligations to data stewardship, the absence of a data policy can cause big problems in a discovery or audit process.
If you find yourself nodding in agreement, or shaking your head in solidarity, then you are probably very interested in migrating to an enterprise analytics solution. Analytics Pros wants to help you make that decision, and give you and your team the information and context needed to frame the importance of the investment in measurement and analytics.
Please reach out to us at @AnalyticsPros and let us know what kind of Rube-Goldberg processes and platforms you’ve had to overcome during your digital analytics journey!