How I Built an Automated Market-Research Engine in n8n (and Solved a Tricky LLM Merge Problem)

Over the past few days, I've been exploring how far we can push modern workflow automation for market research. Traditionally, building a structured research report requires hours, sometimes days, of manual work: defining scope, gathering data, comparing insights, synthesizing findings, and formatting everything into a business-case-ready document.

I wanted to see if I could automate that entire pipeline.

The result: an n8n workflow that completes a full market-research cycle in 5–6 minutes, reducing the equivalent manual work by dozens,if not hundreds, of hours. Not only does it drastically cut research time and cost, but it also creates structured, repeatable insights, ready to plug directly into business strategy.

The Goal: Fast, Structured, Repeatable Market Research

My idea was simple:

Can I build a workflow that takes a single form submission and returns a full business case, including problem statements, market scans, customer pain points, and an opportunity summary?

To make that happen, I needed a multi-agent approach, where different LLMs handle different parts of the workflow:

Scope Definition Agent – defines the research scope.
Deep Market Research Agent (LLM #1) – gathers strategic, quantitative, or trend-based insights.
Customer Problems Research Agent (LLM #2) – uncovers user frustrations, pain points, and unmet needs.
Business Case Writer – synthesizes everything into a clean, business-ready document.

Everything is triggered automatically from a Form, with the final output landing in a Google Doc, ready for use.

The Workflow Architecture

High-level flow:

Google Form Submission A user submits a problem statement or research question.
Scope Definition Agent An LLM interprets the input and structures the research scope.
Deep Market Research (LLM #1) One agent gathers strategic insights.
Customer Problems Research (LLM #2) Another agent runs in parallel to capture user pain points.
Merging the Research Streams The two outputs are merged into a single, unified structure.
Business Case Writer An LLM transforms the combined research into a complete business case.
Google Docs Output The workflow automatically creates or updates a formatted document.

In essence, it's a small automated "research team" operating entirely inside n8n.

Workflow Diagram:

n8n Automated Market-Research Engine Workflow

The Unexpected Technical Challenge: Merging Asynchronous LLM Outputs

Because n8n runs nodes asynchronously, the two large LLM calls (deep research + customer research) return at different times.

At first, I assumed I could simply merge the outputs using a standard Merge node. Instead, I ran into a frustrating issue:

Whichever LLM returned first would overwrite the other output—meaning I only ever received half of the research.

This is a classic asynchronous pipeline problem: parallel workers don't wait for each other unless you explicitly manage their outputs.

My Solution: Controlled Restructuring with Edit Fields + Merge

After experimenting, I found a reliable pattern:

Edit Fields Node (for LLM #1 output) Wrap its content into a predictable JSON structure.
Edit Fields Node (for LLM #2 output) Enforce the same schema consistency.
Merge Node (append mode) Ensures both outputs are included.
Final Edit Fields Node Transform the merged array back into a single clean JSON object for downstream consumption.

This reconstruction step is critical because the Business Case Writer needs a unified structure—not an array of parallel results.

Result: Both research streams reliably flow into a single combined payload, every time, regardless of timing.

Business Metrics and Use Cases

This workflow isn't just a technical curiosity, it produces measurable business value:

Time Savings: A manual research cycle that usually takes 10–20 hours can now be completed in 5–6 minutes. That's a ~95–97% reduction in research time.
Scalability: You can run multiple workflows in parallel without extra headcount. One "research team" can handle dozens of topics per day.
Consistency: Structured outputs reduce human error and ensure repeatable insights.
Immediate Use Cases:
- Market opportunity evaluation for product managers
- Competitive landscape analysis for strategists
- Customer pain point discovery for UX teams
- Pre-sales or investor briefing preparation

In effect, a single form submission now produces a fully fleshed-out business case faster than most people can even schedule a research meeting.