Jun 3, 2025
8 data principles you should apply for marketing
5-MINUTE READ | By Zach Bricker
[ Updated Jun 3, 2025 ]
Marketers know that better data leads to better results—98% of them said so. But our recent study revealed that 56% of marketers don't have enough time to make sense of their data. My guess is that it's less about the lack of tools, but more about the mindset of how marketers are approaching data.
This is where adopting some basic data science thinking could help. Before you back away, no, I'm not trying to turn you into a statistician or build neural networks overnight. Rather, in this article, I'll share how you could prioritize, question, iterate, and optimize your data more effectively, like a data scientist.
You can't make sense of your data, and it's not your fault
One of the biggest challenges marketers face is what Dr. Barry Schwartz calls the "paradox of choice." With so many data points, platforms, dashboards, and tools at their disposal, marketers often fall into "analysis paralysis."
For example, a single Google Ads connector can contain over 1,200 data points. Very few organizations need all of them. Yet many marketing teams pull massive datasets by default, creating a swamp of granular information that's either too dense to analyze or too fragmented to yield coherent insights.
If you try to collect every data point available to you, you're not setting yourself up for success—you're setting yourself up for overload. As someone who's spent years in the trenches of marketing analytics, I can tell you firsthand: more data doesn't mean more clarity. It often means more confusion, slower decisions, and far less impact.

Data science principle #1: Start with what you know
Data science starts with prioritization. Instead of chasing every possible metric, begin by identifying the ones that matter most to your business today. For most marketers, these include:
- Sessions
- Clicks
- Conversions
- Cost-per-click (CPC)
- Return on ad spend (ROAS)
- Engagement rates
Once you establish this core, use it as a foundation to build curiosity. Ask: What else might be affecting these metrics? Can we segment by geography? Device? Creative type?
Treat your dashboard like a conversation. Every dimension in your data should represent a question you're trying to answer. When I sit down with a dashboard, I treat it like a conversation. I ask: What is this telling me? What else could it reveal if I look at it differently? A well-structured dashboard isn't just a collection of numbers—it's a tool that encourages curiosity and helps you approach your data like a scientist would, probing and exploring to uncover insights that matter.

See how your campaigns are performing with this paid channel mix template for Looker Studio >>
Data science principle #2: Avoid the trap of too much data
Rather than wondering "do I have enough data?" ask "do I have too much?" The goal isn't to gather data for the sake of it, but to understand the causality behind trends and behavior.
When you can't accurately describe the effect a dimension or metric has on your daily processes, you've reached a point of too much. I've seen this time and time again: marketers buried in dashboards, chasing micro-metrics that add no real value. If a number doesn't inform a decision or highlight a problem, it's probably just noise.
My recommendation: Start with 5 key dimensions and 5 metrics. If you can confidently explain how each one connects to a business outcome, only then consider adding more.

In the 2025 Marketing Data Report, we asked the respondents what metrics matter to them.
Data science principle #3: Become data-informed, not just data-driven
There's a subtle but crucial difference between being data-driven and data-informed. Data-driven suggests decisions are made solely by the data. Data-informed means human expertise and judgment are layered into the decision-making process.
- Data: Raw numbers, often meaningless without context.
- Information: Data that has been processed or contextualized.
- Insight: The union of subject matter expertise with meaningful measurement.
Marketers should strive for the third stage—insight. This means making decisions where both the data and your expertise agree.

Data science principle #4: Be curious
Good data science is grounded in continuous curiosity. You don't need to know everything up front. What matters is the willingness to ask new questions based on what the data shows.
Start with a baseline dashboard. Then iterate. For instance:
- "How does ROAS vary by region?"
- "Are video ads outperforming image ads?"
- "Is seasonality playing a role in engagement?"
This approach, known in data science as Exploratory Data Analysis (EDA), is about uncovering patterns through iterative questioning.
Data science principle #5: Be familiar with different types of measurement
Everyone said that better measurement leads to better results. But what does "better measurement" really mean?
It starts with right-sizing. Not every company needs advanced techniques like marketing mix modeling (MMM) or multi-touch attribution. In many cases, simple, accurate measurement of existing data (e.g., ROAS, customer lifetime value) is more impactful than expensive, complex models.
Tools like post-purchase surveys, zero-party data collection, and platform-native reports can go a long way when used thoughtfully.

The most common marketing measurement methods according to the 2025 Marketing Data Report
Data science principle #6: Don't be afraid of predictive analytics
Predictive analytics and AI are underutilized in marketing—and for good reason. People don't necessarily trust predictive analytics, and I get it—most marketers haven't been trained in statistical modeling or probability theory. When we talk about forecasting outcomes or projecting trends, it's easy to feel like you're relying on a black box. But understanding prediction requires understanding the basics of statistics. If you've never studied confidence intervals or probabilistic outcomes, it's hard to know whether to trust what a model is telling you. That's why many marketers hesitate. They aren't rejecting the insights—just the uncertainty that comes with not fully grasping the methods behind them.
Instead of jumping into predictive tools without context, start by building confidence in basic models like trend analysis or regression. These provide directional insights with fewer assumptions.
Adopting AI should be a gradual process. You're not looking for AI to replace your intuition but to act as a "force multiplier" for your decisions.
Data science principle #7: Keep data privacy and security in mind
Privacy concerns and the deprecation of third-party cookies are reshaping digital marketing. As marketers shift toward first-party and zero-party data, it's essential to understand how to collect, store, and analyze this information responsibly.
For example:
- First-party data and CRMs help you build sustainable audience intelligence
- Post-purchase surveys can generate zero-party data
- GA4 events and privacy-friendly measurement techniques can capture engagement while staying compliant
While tools like Google's Privacy Sandbox may still be under the radar, marketers must begin embracing "signal loss" and redesigning their analytics with consent and control in mind.
The Power of B2B Signal
Read this guide to learn how B2B companies are dealing with signal loss and drive better outcomes with Conversions API.
Data science principle #8: Don't run before you can walk with measurement
It's tempting to think that adopting advanced techniques like MMM will instantly solve your measurement problems. But not every business is ready.
If you're spending less than $2 million on marketing annually, MMM probably isn't right for you. I've seen too many teams get excited about marketing mix modeling without realizing how data-intensive and computationally heavy it really is. Unless your media mix is complex and diversified enough to provide the necessary inputs, you're better off focusing on foundational techniques like attribution and incrementality testing before diving into MMM.
Instead, begin with:
- Attribution modeling
- Incrementality testing
- Basic forecasting with ARIMA models
Once these are in place and you're resourced accordingly, then you can explore MMM, ideally as part of a combined measurement strategy alongside attribution and lift testing.

Final thoughts
You don’t need a PhD in statistics to get better at marketing measurement. But you do need to change how you think about data. That means prioritizing the right metrics, questioning what you see, reducing noise, and staying curious. It means resisting the urge to hoard data and instead focusing on insights that move the needle.
When you adopt a data science mindset—even at the most basic level—you'll gain more clarity, make faster decisions, and ultimately, drive better results.
The 2025 Marketing Data Report
Read this guide to learn the trends, challenges, and opportunities for marketing measurement.
About the author

Zach Bricker
Zach leads the Solutions Engineer team at Supermetrics in the US. In his role, he's helping brands improve their marketing analytics, MTA, and MMM capabilities.
Stay in the loop with our newsletter
Be the first to hear about product updates and marketing data tips