Market Sizing
How does FutureSearch derive its answers?
Or go straight to a sample report.
Methods
Step 1: Primary and secondary research
FutureSearch conducts far broader research of public information than other AI tools, akin to what top-tier consulting firms do. Some key information won’t be in highly-ranked search results. For this report, FutureSearch found:
Personal anecdotes emphasizing the need for facilities with fermentation capabilities to scale; through industry interviews
A PDF report that consolidated and analyzed numerous databases, survey responses, and interviews with alternative protein companies to infer the share of production of fermentation-based products done by contract manufacturing organizations (CMOs) rather than in house
Just as important is what FutureSearch rejected. Its research involves fact-checking web sources when they conflict with other web sources. This results in removing data points that other AI tools will parrot as correct.
Other AI systems hallucinate factually incorrect information and make misleading assumptions to arrive at an answer without offering any explanation for their reasoning.
Step 2: Model the domain
The facts gathered always tell an incomplete story. There are many methods of inferring the answer.
Do some math. FutureSearch alternates gathering facts with proposing ways of calculating the answer. When those calculations don’t work, it constructs new equations that draw from different data points.
Give the AI time to explore. It’s common on hard questions for multiple lines of research to lead to dead ends. FutureSearch keeps working until it finds one that is sufficient to answer the question.
If given just 5 minutes to answer, human analysts would have to report the results from their first approach. Good answers take time, whether from a human or from an AI.
Make a tree. FutureSearch reports the top level answer as simply as possible.
The full justification is shown across a linked set of pages, each of which answers just one question.
If you just want the answer, you need only read the top node. But if you’re curious, say, about how much revenue companies selling plant-based meat products make, you can click through to that question.
The FutureSearch dropdown subquestion menu allows users to research the share of plant-based meat products that utilize fermentation in production or estimated profit margins for companies selling plant-based meat products— whatever components of the question that the user cares about most.
Step 3: Estimate the unknown
FutureSearch was built to handle uncertainty. It handles questions involving forecasts, where by definition the answer is uncertain and judgment is required.
FutureSearch applies a few techniques to handle uncertainty:
Fermi estimates from known data points: In this report, FutureSearch had to reason about the total market size of alternative meat products that use fermentation.
Accounting for (1 - P) — one minus the average profit margin for such products — and the production capacity for fermentation-based CMOs (as opposed to in-house capabilities), it inferred the market growth rate between June 2023 and July 2024, and then estimated the global size of the market for fermentation-derived ingredients of alternative meat products.
FutureSearch uses known datapoints for the share of production done by CMOs and the CAGR for the fermentation-derived ingredients of alt. meat products market to estimate the global market size.
Reasoning from analogous data points: There are no credible public data on the share of plant-based meat products that utilize fermentation in their production. Here, FutureSearch inferred what it needed with data it gathered on other, comparable plant-based meat products:
Beyond Meat’s Beyond Burger
Impossible Foods’ Impossible Burger
Good Catch’s Plant-Based Tuna
FutureSearch identifies plant-based meat products that are likely to utilize fermentation in their production, then uses this information to create a base rate modeling the share of plant-based meat products that use fermentation. Note that this is a subset of the full list.
The challenge here is identifying comparable offerings. We have honed FutureSearch’s reasoning by analogy by tuning it to forecast accurately from base rates of similar events.
That’s FutureSearch! View other case studies or request your own custom report.