Learning the Language of Data

I want to go back to basics for a second and ask a fundamental question:

Would you say you understand marketing data?

Tough question, right? Marketing data means something different to each person, each role, each organization. It could mean campaign performance, brand affinity, or provider location data. At one large academic medical center, the data team defines “marketing data” as simply defined “website visits.”

What?

Clearly, that’s not what a marketing team member might consider as “marketing data.” So, how do you know what to ask for or use to build revenue models or return on marketing investment formulas? It’s a bit of a mystery.

On top of that, I’m seeing that marketing teams also need more access to data points that are more than “website visits.” And who owns that data? The data and/or IT team within your organization does. To gain access to that data, marketers need to learn how to speak the language of enterprise data. They need to know exactly what to ask for and how to use the data they receive in return to drive better outcomes – better campaigns, better targeting, better online presence, better agentic connections, etc.

Now is the time to flip the script: think data first, marketing second.

When I joined Snowflake, I thought I understood this “data language.” I knew I was embarking on a new side of technology and healthcare and because I used to regularly talk about knowledge graphs, I thought I was good to go.

Ha! I was dead wrong.

It took me at least six months to truly understand what people were talking about on calls with data teams, and even longer to actually speak their language fluently.

In my former roles, I talked about marketing data as the information that surfaces when someone looks for a doctor: provider name, specialty, insurance accepted. I talked about connecting entities (e.g., “this doctor works at this location and treats this condition”). I truly believed this was the be-all, end-all data we needed to organize to help patients find care online because if you didn’t have this information, how could you facilitate a conversion (e.g., appointment booking)?

But now that I’ve stepped into the actual data side of the world? Boy, does provider and location data pale in comparison to what else exists within a health system and ways it can be used to drive better marketing outcomes. I was naive to think marketers only needed provider and location data to drive consumer discovery and patient acquisition.

Data teams are sitting on massive amounts of data. There is such a robust ecosystem of data within healthcare organizations that if you are only focused on website schema tags, branding campaigns and maybe website visits, you are playing in the kiddie pool while everyone else is swimming in the ocean. (I’m not saying they aren’t important. They are. But they are just a tiny bit of what is needed now to successfully market in today’s day and age.)

Lately, I’ve been getting a lot of questions about how marketers can better communicate with their health system’s data and IT teams. Here is the problem: I’ve spoken to numerous data teams at massive health systems, and when I talk to them about a particular initiative or platform decision and ask: “What does the marketing team think about that?” I almost always get the exact same response:

“We didn’t ask. Marketing will just have to deal with it.”

Bottom line: the data team isn’t considering marketing when making decisions.

Maybe they don’t understand why you need to connect with them, or maybe they think marketing shouldn’t touch enterprise data (this is actually the most common sentiment).

So, it’s already an uphill battle to fight.

Regardless, there is immense value in the data for marketers and your organization holds it, but it is entirely up to you to build the relationship with the teams who own it. By building trust, you can then hopefully unlock that white-knuckle hold the data team has on it and use it for powerful good.

So, back to the swimming analogy: if you want to get out of the kiddie pool and swim in the ocean, here are a few things you will want to consider (and do):

1. Build the Relationship First

Your health system’s data team is just as busy and strung out as you are. Just like you do, they serve the entire organization (so it’s definitely one immediate thing you have in common)! Setting up a “get to know you” meeting might take a while, but it needs to be your top priority. There’s usually at least one person on the data team who serves as the liaison to marketing and/or patient experience (many times they are the same person). Find that person. If you’re VP-level or above, find your data team leader equivalent. Try to set up a recurring meeting, starting with a “get to know you” 1:1 and, before you meet, make sure you do your homework (see below).

2. Understand Your Health System’s Core Tech Stack

Find out what technologies are dominating the data team’s time. I’m almost positive it is Epic, Microsoft, data platforms like Snowflake, and maybe Salesforce (though I’m seeing this one disappear). Epic and Microsoft discussions literally take up so much of their time. So already, you’ll be competing with time they have to spend in half- and all-day sessions these companies regularly schedule alone. However, most of the time, the data you eventually want to access will likely live in a centralized data platform. This makes it a bit easier on the data team because you don’t have to ask for Epic data sourced from Epic directly (they wouldn’t give it to you that way anyway). If your organization has a data platform, it’s likely you can get access to it from the data platform instead. But know that regardless, there are platforms that the health system has invested in and paid millions of dollars for, have staffed entire teams to work on, and that will likely be the first option for most requests you make in the future. Which leads me to…

3. Prepare for the “Consolidation” Pushback

When you evaluate a new technology vendor that isn’t already part of the core stack mentioned above, expect a long, drawn-out fight and a likely “no” (you’ve probably already encountered this, need I say more?). Platform consolidation picked up steam during the pandemic and never really relented. IT teams will look at your request, determine that an existing tool (like Epic or Microsoft) can do 80% of the job, and deny your request, all in an effort to save money and consolidate platforms. This is a “live with it” scenario and you have to accept the fact that you might just, in fact, have to live with it.

4. Stop Asking for Tools. Start Bringing “Use Cases.”

Historically, I’ve watched marketing teams go to data and IT teams to ask for an evaluation of a tool. Over 90% of the time, the evaluation is delayed or the IT team provides a flat out “no” to purchasing the tool. Why? Because every time you approach IT with a request to buy a new marketing tool, they hear two words: technical debt. They immediately dread the integration work. It’s an automatic “no” not because it won’t work, but because it’s too much work (and may be duplicative work, too). Take it from me, listening to data teams talk about integrations is painful. If it’s painful for me as a non-technical person, it’s likely 10x more painful for them.

So, you need to work on getting to a yes. And after over a year talking to data teams at massive health systems, I’ve now learned that to get a “yes,” you have to speak their language: the language of data.

For example, here is what usually happens:

The Marketing Ask (“I need a tool.”): “We need an email tool to send appointment reminders to patients who haven’t been seen in a year.” (IT groans: “We have one already. It’s called Epic and the functionality exists somewhere with Epic already.”)

Instead, what needs to happen is:

The Data Language Ask (“I have a use case.”): “We have a patient engagement use case that requires us to drive better patient appointment adherence. We need to leverage data in our unified data foundation (or “enterprise data warehouse” or “enterprise data zone” two commonly-referred to places where data might live) to analyze lapsed patient appointments and also see if we can identify behavioral triggers on our website (like patients or agents searching for specific providers, specialties or insurance accepted) so we can automatically route them to scheduling with the right provider.”

Data teams understand this. They love solving problems (I’ve never seen more people who just love data puzzles as much as these data folks!). Bring them your use case, and let them help you map out a solution using the infrastructure they’ve already built. Sometimes this means using an existing tool and existing data sets (they like to talk about data in tables a lot!). Sometimes it means actually finding a tool or a platform that can solve your particular problem. But the data team owns this decision so you have to present them with a use case that makes sense to them.

That use case doesn’t start with a tool. It starts with the problem you’re looking to solve.

Marketers also need to start thinking differently, because the world is different and moves much faster than it did 5, 10, even 15 years ago. When I was at Google, organizations used to run campaigns and they would do so for a particular timeframe. That’s not something we can do anymore. The world moves so fast right now that it’s almost impossible to actually run a campaign because data access can now allow campaigns to be living, breathing entities.

For example, with access to data, you can literally pivot on a dime when you see inpatient volumes decline and need to fill beds (wouldn’t that be nice?) or when you see a provider’s panel is lighter than normal and you can then couple that deficit with data from Epic about patients who have lapsed appointments and quickly spin up a digital marketing campaign (and email outreach) in those markets to drive patient acquisition.

This leads to precision marketing rather than campaign-based marketing. It’s a huge shift and it starts with understanding the data you may be able to access today. But you can’t access that data if you don’t know how to first speak the language of data.

Adapting to these changes won’t be solved by point solutions or by burying our heads in the sand. This is a slow process, but a worthwhile one at that. Once you learn the language of data, a whole world opens up to understand how to use the data to do a better, more precise job, and allows you to better adapt to changes that are happening with technology and consumer behavior. You won’t need to understand what “marketing data” is because you’ll be able to customize the right data for your use case because you’ll have unlocked access to it within your organization.