In light of the economic downturn, companies want to integrate internal data science teams into their marketing data platforms to prove the value of their investment. While the platforms are due for a ROI performance report, we suggest a better solution: Bring Your Own Model.
This post is part of our 2023 Data Transformation Outlook series
Data Marketing Platform Performance Anxiety, or, As Jimmy Spicer Posits, “Dollar Bill, Y'all”
Businesses are tightening their belts, and making hay has become more critical than ever. Companies are starting to ask more about what they are getting back from all their investment in marketing data platforms, including Adobe Experience Platform, Salesforce Marketing Cloud, and Google Marketing Platform.
Descriptive vs. Prescriptive Analytics
These ROI-focused questions aimed at marketing platforms have arisen not just because of the economic downturn, however, but because of an unfortunate lack of realized business impact. The analytics components of the platforms deliver data that’s descriptive of what happened, but they don’t deliver on the promise for perspective support. In other words, stakeholders don’t know what they’re getting out of the platform, and they don’t know how to apply the data to their business decisions.
Low Adoption by Marketing Teams
The platforms engender a “no code” philosophy that seems like a great idea on the surface: Drag-and-drop, point-and-click data science solutions that anyone can use? Yes, please! But marketers have said no, thank you. Marketers are especially busy, and they aren’t able to confidently activate on the results of model outputs. It’s not what they do. But the pressure on marketers to prove the value of these platforms is fierce.
How to Prove the Value of Marketing Data Platforms
Organizations are attempting to prove the value of their marketing data platform to varying degrees of success by doing the following:
- Hiring a platform specialist to learn how to use the tools within the platform: This results in a limited deliverable for the marketing team and limits ROI as a result.
- Integrating Data Science teams with the marketing team’s data platform: Marketing teams desperate for answers turn to their Data Science teams for help proving the value of their platforms. But these platforms aren’t designed for custom or complex modeling. Again, they’re designed simply, for marketers, and sold as such. Compared to what effective DS/ML teams can do, the solutions are limited. A “no-code” first workflow is NOT the way effective DS/ML teams function. It hinders speed to production as well as model accuracy. These are essential for successful teams, so most simply refuse, when asked, to work in the marketing data platforms.
- Implementing a hybrid approach that we call “Bring Your Own Model.” This is where the platform developers create an API that allows data scientists to pull the data out, do the creative stuff they need to do to answer business questions in a cloud or local environment, and then push their model outputs back into the platform for marketing teams to activate on
Benefits of Bring Your Own Model
Once I did a really dumb stunt on my buddy’s bike. I crashed it. It wasn’t my bike, after all. It was weird. Using another team’s workflow to deliver projects is kind of like that epic recipe for failure. In this case, the platform’s models are your buddy’s bike: they’re not what you’re used to; they’re a weird workflow; they lead to unsought outcomes and broken things.
On the other hand, when a team can integrate with their native workflow, they achieve faster more tangible progress, the ability to containerize functions, accessibility and flexibility to handle a variety of use cases, and version control to be able to move faster and break less stuff.
Showing value of platform investment along with optimizing internal resource investment is a big deal in solving for the economy downturn challenge. As we continue to see an increase in the number of companies attempting to integrate their DS and ML teams into the marketing analytics platforms, we caution against it. There’s a better way! Bring your own model! We are helping teams do this by utilizing platform APIs and delivering industry-leading platform expertise. If you are considering this type of solution to help your company address these challenges, we can help.