How to Turn Any Unstructured Text into Structured Data Using GenAI
For years, there was an unspoken rule in the analytics world: if your data isn't structured, it's useless. Freeform text, lengthy descriptions, and unformatted content were essentially dead ends for anyone trying to extract meaningful insights or build automated workflows.
That era is over.
The Old Problem with Unstructured Data
When I started my career in analytics and automation nearly a decade ago, the advice was clear: if you want to analyze data, you need to collect it in a format designed for analysis from day one. Unstructured data - emails, job descriptions, customer feedback, meeting notes - was considered a non-starter. The workaround? Painstakingly design forms, surveys, and data entry systems that forced information into neat little boxes.
This approach worked, but it had serious limitations. It constrained how people could communicate, created friction in data collection, and meant that valuable information locked away in unstructured formats simply couldn't be leveraged.
The GenAI Game Changer
Generative AI has fundamentally changed this equation. Today, virtually any unstructured content can be transformed into properly structured data tailored to your specific use case. The implications are massive.
A Practical Example: Parsing Job Descriptions
Let me walk you through a simple but powerful example I've been experimenting with. Imagine you want to analyze job postings at scale - extracting specific information like required skills, salary ranges, experience levels, and company details from lengthy, unformatted job descriptions.
Here's the workflow I built using Zapier:
Step 1: Capture the Unstructured Data
First, I take a large chunk of freeform text - in this case, a complete job posting from LinkedIn. This is the kind of content that would have been unusable for automated analysis just a few years ago.
Step 2: AI Converts to JSON
The raw text gets fed into an AI step that I've configured with specific instructions: convert this unstructured content into a precise JSON format. I define exactly what fields I need and how they should be structured.
Step 3: Code Processes the Structure
Once the AI has converted the text to JSON, a Python code step takes over. This code knows exactly what to expect (because we defined the JSON structure) and can reliably parse out the specific fields we care about.
Step 4: Output to Your System
Finally, those extracted fields get written back to a Google Sheet (or whatever destination makes sense for your workflow), where they're ready for analysis, visualization, or further automation.
Why This Matters
This isn't just a neat party trick. This approach opens up entirely new possibilities:
Legacy data becomes usable: All those old documents, emails, and notes you've accumulated? They can now be systematically analyzed.
Reduced data collection friction: You no longer need to force people into rigid data entry forms. Let them communicate naturally, then structure it afterwards.
Scale previously manual processes: Tasks that required human review of unstructured content can now be partially or fully automated.
Connect disparate systems: Information from one system can be extracted and reformatted to fit another system's requirements.
The Technical Sweet Spot
What excites me most is combining AI steps with traditional code. The AI handles the messy, ambiguous work of understanding and restructuring unstructured content. The code then provides reliability and precision for the structured processing that follows.
This combination gives you the best of both worlds: the flexibility and intelligence of AI with the predictability and speed of traditional programming.
What's Your Use Case?
I've shown you job description parsing, but the possibilities are endless. What unstructured data has been sitting in your systems, untapped because it wasn't in the right format? Customer feedback? Meeting notes? Email threads? Research documents?
The tools to unlock that value are available right now.
—
Evan is a consultant at Quantify Consulting, specializing in analytics and automation solutions. Connect with him to discuss how these approaches might work for your organization.