Is MI Too Early Without Ready Data? — A Reverse Approach to Accelerate Materials R&D

‍Introduction

For many organizations, the interest in Materials Informatics (MI) is already there—but progress often stalls when it comes to actually getting started.

“We’re interested in MI, but our data is still on paper or scattered across Excel files…”
“Do we need to build a company-wide data infrastructure before we can even think about using AI?”

These concerns are common, and they point to what is often perceived as a “data readiness barrier.”
When large volumes of historical data remain unstructured or analog, the idea of digitizing everything first can feel overwhelming—and time-consuming.

While it’s true that data is essential for AI, waiting until everything is perfectly organized can slow down innovation. In practice, the assumption that MI must come after complete data preparation often becomes a bottleneck rather than a prerequisite.

In our previous article, we introduced a six-step framework (CRISP-DM) for executing MI projects.
However, in environments where data is still evolving, trying to follow this process rigidly can leave teams stuck in the early stages.
In this article, we take a different approach. By looking at the three stages of DX, we introduce a practical “reverse approach” that allows organizations to move forward faster—even with imperfect data.

‍

The Three Stages of Materials R&D DX

DX is often described as a three-step progression.At Polymerize, we adapt this framework specifically for materials R&D

‍

Step 1: Digitization

Converting analog data into digital assets‍

• Converting handwritten experimental notes into Excel or electronic lab notebooks
• Converting paper or PDF outputs from instruments into numerical data (e.g., CSV)

Goal: Transform physical information into a format that computers can handle, and establish it as data assets.

Step 2: Digitalization

Structuring and standardizing data‍

• Organizing scattered data across teams
• Standardizing formats and naming conventions
• Preparing data for analysis

Goal: Enable consistent and scalable data usage

Step 3: Digital Transformation (DX)

Driving innovation through data and AI

• Applying machine learning to predict material properties
• Optimizing formulations and processes
• Integrating AI into R&D workflows

Goal: Create business value and accelerate innovation

The Common Mistake: Trying to Do Everything in Order

Many organizations attempt to move through these steps sequentially:
Digitization → Digitalization → DX‍

However, in reality, this often leads to:

• Spending years on data preparation
• Losing momentum before reaching value creation
• Delayed ROI from DX initiatives

A Better Approach: Start with MI First

Instead of waiting for perfect data, a more effective approach is to start with MI early—even with limited or imperfect data.

Why the MI-First Approach Works

1. Clarifies what data is actually needed→ You don’t over-invest in unnecessary data preparation
2. Creates immediate value→ Early insights help build internal momentum
3. Accelerates learning cycles→ Trial → feedback → improvement happens faster

From “Data First” to “Use First”

Rather than asking:  ❌ “Do we have enough data?”

Shift the question to: ✅ “What can we learn from the data we have today?”

Practical Workflow

Instead of fully completing Step 1 and 2 first, try this:

1. Start with available data (even if incomplete)
2. Run a quick modeling cycle
3. Evaluate results
4. Identify missing data
5. Improve data and repeat

This iterative loop is far more effective than waiting for perfection

Conclusion

Start small. Learn fast. Scale gradually.

You don’t need perfect data to begin MI.In fact, starting early is often the fastest path to meaningful results.
By adopting a reverse approach—starting with MI and refining data along the way— organizations can accelerate Materials R&D DX and unlock value much sooner.

How Polymerize Supports This Approach

Polymerize’s platform is designed to support this iterative workflow:

• Data structuring: Standardized templates for faster organization
• Modeling: Easy-to-use machine learning tools
• Application: Built-in forward and inverse analysis

Whether you are just getting started or scaling your DX efforts, the platform helps you move forward without losing direction.

Get Started

Why not begin with your existing data?Start your first cycle today—and experience how MI can drive real progress.

‍