Most AI-assisted market research looks like this: open a chat window, type "what are the main trends in [your industry]," read a plausible-sounding but generic response, and close the tab. This is the worst way to use AI for research — you get surface-level synthesis of whatever was in the training data, with no connection to your specific market, your specific customers, or your specific positioning problem.

The autoresearch approach to market research is different. Instead of asking open-ended questions, you give the AI a structured program: a specific research question, a set of inputs to analyze, and a defined output format. You run the program, review the output, update the inputs, and run again. The loop doesn't converge on "better content" — it converges on a clearer model of your market.

Research Loops vs. Optimization Loops

The cold email and landing page loops in the other guides are optimization loops: you have a metric, you test variants, you keep what works. Market research uses a different loop structure — a discovery loop. The goal isn't to maximize a number; it's to build a model of a market that's accurate enough to inform your optimization decisions.

A discovery loop runs like this:

  1. Define the research question. Not "tell me about my market" but a specific, falsifiable question: "What are the top three objections buyers raise when evaluating [category], based on the customer reviews and forum posts I provide?"
  2. Provide structured inputs. The AI doesn't know your market — you do. Feed it real data: customer interview transcripts, competitor review pages, support ticket themes, sales call notes. The program.md format has a dedicated section for this.
  3. Specify the output format. What do you want out of this analysis? A ranked list of objections? A positioning matrix? A summary of the language customers use to describe the problem? The more specific your output format, the more useful the result.
  4. Review and update. Run the program, read the output, and ask: what's missing? What's surprising? What does this suggest about what to research next? Update your inputs and run again.

Competitor Positioning Analysis

One of the highest-value market research tasks for a lean team is understanding how your competitors are positioning — not what features they have, but what jobs-to-be-done they're claiming to solve, and for whom.

To run this as an autoresearch loop:

  1. Collect the homepage headline, value proposition, and one "how it works" section from three to five competitors
  2. Feed those as structured inputs into the research program.md template
  3. Ask the agent to: identify what customer segment each competitor is targeting, what problem they claim to solve, and what makes their claim distinct from the others
  4. Ask a second research question in the same program: where are the gaps? What segments are under-served? What problems are everyone claiming to solve (and therefore not differentiated on)?

The output won't be a strategy — it will be a structured view of the competitive landscape that you use to make positioning decisions. The AI's job is synthesis and pattern recognition, not strategy. Your judgment enters at the interpretation step.

The Autoresearch Playbook includes a competitor positioning template that structures this exact analysis. It ships with a worked example from the email marketing software category so you can see what a useful output looks like.

Customer Interview Synthesis

If you have customer interviews, sales call recordings, or support ticket logs, the autoresearch research loop is particularly valuable for synthesis. The problem with raw interview data is that it's too rich to absorb directly — you end up with a mental model that over-indexes on the most memorable quotes and under-indexes on the patterns that appear across multiple interviews.

The research program.md solves this by turning synthesis into a structured task:

  • Paste three to five interview transcripts or call notes into the inputs block
  • Ask the agent to: identify the three most common problems customers mentioned (with frequency counts), identify the language customers used to describe those problems (actual phrases, not paraphrases), and identify one unexpected finding
  • Review the output and run a second cycle on the unexpected finding: what additional inputs would confirm or refute it?

The key discipline is in the output specification. Asking for "a summary of customer feedback" produces a narrative that's easy to read but hard to act on. Asking for "three problems with frequency counts and customer-verbatim language" produces structured data you can use directly in your copy and positioning.

How Research Feeds Into Optimization

Market research loops and optimization loops are complementary. Research loops tell you what to optimize — which customer objections to address, which benefits to lead with, which market segment to focus on. Optimization loops tell you how well a specific version is performing on those dimensions.

A well-structured research workflow looks like this:

  1. Run a competitor positioning analysis to identify where you can differentiate
  2. Run a customer interview synthesis to identify the language buyers use to describe their problem
  3. Use those outputs to write a positioning hypothesis ("our target buyer is X, their problem is Y, our differentiated solution is Z")
  4. Write a landing page headline using that positioning hypothesis
  5. Run a landing page optimization loop to test whether the hypothesis converts

The research is upstream of the optimization. Without it, you're optimizing the wrong message — making the wrong headline more efficient. With it, you're testing whether your positioning hypothesis is correct, which is a much more valuable signal.

What AI Market Research Cannot Do

It's worth being clear about what this approach won't produce. AI synthesis of market data will not give you a breakthrough insight that no one else has — it can only pattern-match on inputs you provide. It will not tell you whether your market is large enough to build a business on. And it will not replace talking to customers directly.

What it will do is process more inputs faster, surface patterns more consistently, and structure your findings in a format you can act on — without the three-hour analysis session that usually precedes a positioning decision.

The free 12-point assessment includes a section on market research maturity — it will show you whether your current research process is strong enough to feed useful inputs into the optimization loop.