The research team

When to call this team

Research is the team for any claim about the world outside your building, the claims you are otherwise tempted to take on faith. Reach for it when:

• A new product idea feels obvious and nobody has yet tried to kill it.
• You want to know what your market's customers complain about, in their own words.
• Timing is the question: is this space heating up, seasonal, or already regulated against you?
• Two AI assistants handed you confident opposite answers and you need an adjudicated one.
• You keep generating ideas from the same two angles and want the other ten.

The cost of getting these wrong is paid later, in a build nobody wanted or a window you missed. This team moves that cost forward, to a day of research instead of a quarter of regret.

Who shows up

The research orchestrator runs a pipeline rather than a pool, so work flows through stages: generation, then validation, then signal scouting, then tool selection. The specialists along it:

• Idea generator, working a 12-angle ideation framework with each idea attributed to its angle.
• Demand validator, who runs candidates through a 7-platform protocol (Wikipedia, Reddit, Google, YouTube, TikTok, app stores, Amazon) and returns traffic-light scorecards.
• Audience miner, who reads middling reviews and community language for unmet needs.
• Signal scout, who reads timing through four lenses, including seasonality and regulation.
• Multi-model synthesizer, who cross-checks answers across Claude, OpenAI, and Gemini with per-claim provenance.
• Tool curator, who matches each question to the right sources and grades them by quality tier.

Three worked examples

"Validate this before we build it." The demand validator runs the idea through all seven platforms, measuring whether real people already search for, complain about, watch, and buy around the problem. The scorecard comes back green, yellow, or red per platform, with the evidence under each rating. A red scorecard delivered in a day is one of the cheapest gifts a company can get, because it kills a quarter of wasted building before it starts.

"What do these customers actually want?" The audience miner reads the 2-3 star reviews of incumbent products, where honest disappointment lives, and returns the unmet needs in the customers' own vocabulary rather than yours. That captured language feeds positioning and copy directly, which is why this engagement so often runs just before the copywriting work in chapter 20.

"Is now the right time?" The signal scout examines the timing of a move through four lenses, geography, seasonality, patent and IP activity, and regulation, then reports which windows are opening and which are closing. Strategy that ignores timing is just strategy that arrives late, and this is the check against that.

What they need from you

The decision the research is meant to serve, stated plainly, plus any prior beliefs you want stress-tested rather than confirmed. The team's search and platform connections handle the gathering, and the multi-model synthesizer needs nothing from you beyond the question itself.

Hand-offs

Research feeds the teams that act on what it finds. Validated demand goes to advisory for the business case (chapter 9) and to product for discovery (chapter 12). Mined audience language goes to marketing and copywriting (chapters 10 and 20). And anything the team concludes can be written into the hypothesis ledger (chapter 21), where a conclusion stays accountable instead of fading into a slide nobody revisits.