← Video Library · Workshops
How to Do Market Research for a New Business (DARB DEMO)
Published January 20, 2025 · 93 views on YouTube
Key takeaways
- Feed the AI a reference competitor's website plus a few plain-language sentences about the idea before asking for a strategy.
- Rerun the same prompt with small refinements (like narrowing to one sub-segment) to see how much the output shifts.
- Use an initial AI-generated report as a discovery step before spending a full blueprint credit on a polished report.
- Turn identified pain points and symptoms into short outreach messages, survey questions, or a nine-word email campaign.
- When targeting more than one audience (like busy professionals vs. busy parents), run the process separately for each.
How do you do AI-assisted market research for a brand-new business idea?
In this MarketSauce weekly office hours call, Darby demonstrates the process live with two members: paste a link to a similar existing product or service into the “Quantum Mentalist” tool inside MarketSauce Studio, add a few sentences describing the new idea in plain language, and submit. The tool scans the reference site and generates an initial executive summary, SWOT analysis, market sizing, and persona draft that the member can then refine by re-running the prompt with more specific context.
What’s a real example of the process, start to finish?
One member, Derek, pitched a rough idea he’d had two nights earlier: a Costco delivery service for people living outside delivery range in Mexico who don’t have a car. Darby walked through it step by step:
- Identify an existing comparable service. Derek named three local food-delivery apps (Uber Eats, Rappi, and DiDi Food) as reference points.
- Copy the URL of the closest comparable site and paste it into the Quantum Mentalist tool for scanning.
- Type out a plain-language summary of the idea: a Costco delivery service, twice a month, in bulk, for people 80+ miles from a store or without a car.
- Submit and review the first-draft report (executive summary, SWOT, market size, and an initial persona).
- Narrow the input based on what looked promising in the first draft. Derek and Darby noticed the “elderly, can’t drive, remote area” sub-segment stood out, so they reran the prompt focused specifically on that group.
- Review the refined persona (“elderly population… typically over 65, often retired, mobility issues”) and use the report’s chat interface to keep asking follow-up questions, such as generating ad copy for that specific segment.
Darby was clear that this scan-and-chat step is meant as an ideation and discovery phase, not the final output. The next step, when a member wants a fully organized deliverable, is spending a blueprint credit inside the MarketSauce portal to generate the complete report.
How do you go from a persona to actual marketing content?
The second half of the call covers a different member, Re, who already had a client and a defined audience (busy professionals and busy parents dealing with gut-health symptoms like bloating, low energy, and excessive gas) and wanted help turning that into outreach content for Flowchat and social media. Darby’s approach:
- State the audience and their symptoms directly (age range, income, specific pain points) rather than a vague “we help everyone” pitch.
- Ask the AI to identify top pain points first, then generate five related symptoms, to keep the language aligned with what the client already knows about their audience.
- Turn those symptoms into short “raise your hand” outreach messages that could be dropped into a Flowchat sequence, following the pattern of asking a question, confirming the symptom, then offering next steps.
- Rewrite the same messages into other formats on request, including compressing one into a nine-word email in the style popularized by Dean Jackson.
- Submit a full blueprint request in parallel so a more complete, ten-page-style report (topics, offer targets, secondary goals) is ready to reference once the initial content ideas are drafted.
Do you need a full blueprint credit every time?
No. Darby’s guidance on the call was to use the free Quantum Mentalist scan-and-chat step first to get clear on who the audience is and to sanity-check the AI’s first attempt as “the human in the loop.” A blueprint credit is worth spending once the target segment is genuinely settled, since the blueprint expands into a fuller report (including offer targets, customer journey mapping, and roughly ten pain points and symptoms per topic) that can be uploaded elsewhere for continued chat-based work.
What tool were they actually using, and how does it relate to ChatGPT?
The call also touches on “DAR,” an early internal GPT Darby was building on top of the buyer brief book and MarketSauce’s frameworks, distinct from “The Mentalist” GPT that members were already using inside ChatGPT. As of the call (January 2025), both were running on OpenAI’s GPT-4o, and Darby said there wasn’t yet a meaningful output difference between them, recommending members stick with The Mentalist for speed while DAR was still in development and intended to eventually be built directly into the MarketSauce platform.
Full video transcript
Hey everyone, Darby here, founder of GenAIUniversity.com, and in today’s call we’re going to be diving into some MarketSauce weekly office hours. We’ll be covering different questions and connecting with the community of folks out there using MarketSauce and integrating it into their business. Today marks week one of our weekly office hours for 2025. The format of today’s call: we’re joined by Ivy, good to see you. It’s really just answering your questions about what you’re doing with MarketSauce, if you have questions about how to utilize it in your business, or any particular thing related on that note that we can address here live on these calls and through the comments coming in through our community.
Ivy, tuning in from the Philippines at two in the morning, was the first to join but didn’t have a specific question. Derek joined next and shared an idea he’d been working on: marketing a Costco delivery service for people who live around 80 miles from a Costco or Sam’s Club and don’t have a car. Derek, currently in Mexico, described it as his own version of a delivery service like DoorDash, using his car to shop and deliver goods, an idea he and a business partner had come up with two nights earlier.
Darby suggested starting by identifying an existing product or service with parallels to the idea. Derek named three delivery providers common in his area: Rappi, DiDi Food, and Uber Eats. Darby then shared his screen, copied the URL for one of those services, and opened MarketSauce Studio’s new scanning tool, explaining that with any generative AI, the quality of inputs and context provided has a large impact on the quality of outputs.
Darby typed out a plain-language description of Derek’s idea: a Costco delivery service similar to DiDi Food and Uber Eats, but focused on Costco delivery for people too far away or without a car, offered twice a month in bulk. After confirming the details with Derek, he clicked submit. The tool first scans the reference website, then factors in the additional context provided, and generates an initial report.
The first-draft report included an executive summary describing a Costco delivery service in Mexico for people without car access, targeting rural and suburban households, an expanding middle class, and increasing internet penetration; a SWOT analysis; and initial market size and persona information. Derek noted the sub-segment that stood out most to him was the elderly who can’t drive, in a remote area. Darby suggested rerunning the report with that additional context to see how much the output would change.
On the second run, the persona development section named the primary persona as the elderly population residing in remote areas, typically over 65, often retired, with fixed income and mobility issues, which Derek confirmed was accurate. The report also included customer journey mapping, problem/solution framing, and a competitor comparison matrix. Derek said the output was very useful this early in the idea, giving him more direction than he’d had going in.
Darby explained that the next step to build this out further would be spending a credit inside the MarketSauce portal on a full custom blueprint, which would expand to roughly ten pages plus ten different micro-segments the member could target within the broader elderly, remote segment. He noted the process is iterative and conversational, and requires clear thinking up front about what output you’re looking for.
Darby then spoke with another member, Re, who explained her background: she runs outreach and social selling campaigns using Flowchat, and works with rebrand clients by scanning their existing websites. Her challenge was that after strategizing with clients, target audiences often change, and she needed a way to quickly generate messaging, hooks, and content ideas for a specific audience rather than a full long-term strategy document.
Re’s example: a client selling a six-week gut-health program combining Ayurvedic medicine, kitchari, supplements, and community support, targeting busy professionals and busy parents dealing with bloating, low energy, and excessive gas. Following an approach inspired by Traviso’s teaching, the goal was to speak to symptoms rather than the underlying problem, in a “raise your hand” pre-sell campaign for a not-yet-announced VIP program.
Darby walked Re through the same avatar/niche/goal/problem framework, asking for a quick sentence about what makes the client’s product different (a well-known advisor in the functional-medicine and gut-health space, and about 500 people who had already been through the program). After submitting that context for a full blueprint report, Darby switched to demonstrating the secondary approach: MarketSauce Studio’s DAR GPT, trained on the buyer brief book and MarketSauce’s frameworks.
Using DAR, Darby fed in demographics (ages 34 to 45, household income $100K+), then asked it to identify top pain points, and finally to generate five related symptoms, arriving at phrasing like “afternoon energy crash” and “uncomfortable bloating after meals.” From there, he generated outreach message examples usable in a Flowchat sequence, and, at Re’s request, condensed one into a nine-word email in the style used by marketers like Dean Jackson.
While that conversation continued, the full blueprint report that had been submitted earlier finished generating, giving a complete write-up including topic ideas, secondary goals, and roughly ten offer targets such as marketing executive, software developer, freelance writer, sales director, and nutritionist, each of which could warrant its own nuanced messaging.
Re asked how DAR compared to using The Mentalist GPT or ChatGPT directly. Darby explained DAR was being built on Pickaxe as an early, more customizable alternative, but that as of the call there wasn’t a major output difference from The Mentalist, which was already trained on the buyer brief book and MarketSauce’s frameworks and ran on GPT-4o. He recommended members continue using MarketSauce’s Mentalist GPT for now, since DAR was still in development and not yet built for team access, with the long-term plan being to integrate DAR directly into the MarketSauce platform so members wouldn’t need to switch into ChatGPT at all.
The call closed with Darby thanking Ivy and Re for joining, noting the office hours would continue on a weekly basis through at least mid-March, with recordings planned for a separate YouTube channel for members who want to catch the replays.