Is it time to ditch your data lake dreams and get real about your data strategy? Learn how normalizing, accessing, and ensuring data accuracy can turn your publishing organization into a truly data-driven powerhouse. Discover the steps to make data work for everyone, from your ops team to your business users.
Media and ad tech conferences have been dominated by discussions about AI and cookie deprecation over the past couple of years. These are important topics, but one equally important topic gets less attention: data strategy.
Everyone wants the mystical data lake that will solve all their data needs and finally make them “data-driven,” the thing everyone claims to be but few actually are.
A data lake can be a great thing but, not unlike a normal lake, it can also be filled with toxic waste and be more like a dump than a beautiful lake anyone wants to touch. Just putting your data in a data lake doesn’t actually fix anything. A data lake is just a fancy marketing term for a database.
The key to enabling your organization to make data-driven decisions is to make the data accessible to the whole organization and different stakeholders, including those who don’t have a computer science or data science background.
For example, your ops team may want to know the latency of ad loading or be able to see how many impressions an ad unit generated for a certain audience. They shouldn’t need to know SQL to achieve that.
A SQL prompt (even though it is powerful and one of my favorite tools) won’t help, and a static dashboard won’t help either because you can only think of and design so many things ahead of time. You need something else.
3 Steps to Unlock Data for the Entire Publisher Organization
So, how do you make your data truly accessible — and understandable — to every relevant person within your organization?
- Ensure you have a solid ETL pipe. You want all the data in one place, but more importantly, you want it normalized across your sources so you are actually comparing apples to apples when reporting. A business user shouldn’t need to know how Magnite or Index Exchange defines their ad types. Their tools should account for these differences.
- Make the data accessible. Enable the data to be queried with easy-to-use tools that take care of the logic in the background. People are strapped for time, and if it is a hassle to get to the data — maybe they have to submit a ticket to the data science team and wait two weeks to hear back — they are probably not going to do it.
- Monitor the data for accuracy. One thing that will definitely kill a data strategy is inaccurate or out-of-date data. If users can’t trust the data, they will not use it, instead retreating to manual Excel spreadsheets or other less effective methods.
A data lake won’t make publishers data-driven. But getting all their data in one place is indeed the first step to more efficient, data-driven decisions.
Normalizing the data, making it easy to query, and shoring up its accuracy will help publishers get the rest of the way so that “data-driven” is a real way of doing business and not just a nice-sounding slogan.