Testing third party data can be a daunting task. Follow these practical steps and increase the value of the time and effort spent on testing.
Why does testing matter?
Selecting a third party data provider can be a critical strategic decision. The costs of incomplete and inaccurate data is well understood and easily quantified. Making the right decisions will increase effectiveness and profitability.
What are some aspects of data testing that everyone should pay attention to?
- Time is not a friend of data testing. Get current data, perform the testing and do the analysis quickly. As data naturally decays, the more time spent in testing degrades the actual results
- Get a statistically significant sample size for testing. Too often, a data test is a very small number of test records. This will inevitably lead to results that will not reflect reality. We all know that if we flip a coin twice and it comes up “heads” twice we can not assume that every coin flip will result in “heads”.
- Diversify the test data. While the majority of the test data should be production data, it may include known data – both correct and incorrect to provide insights beyond match rates
- Evaluate multiple sources. Multi-sourcing data enables a first pass – second pass methodology that can be used to optimize both results and profitability.
- Look beyond just the data. Good data that is difficult to access or integrate into a process flow can be less valuable than average quality data that can be seamlessly incorporated into the tech stack.
- Pay attention to pricing models. A licensing agreement will have different cost / benefit implications than per-match or per-query models.
- Make a single resource accountable for testing.
In summary, data testing is critical, and it needs careful attention, resources and analysis to make it effective. Please reach out to us to discuss how to properly conduct a test and how our data services can be of assistance!
