When I look at a company, I want to understand why people pay for it and what it costs to deliver the product. My background is in private equity, so those are the questions I bring to AI businesses too. This guide walks you through that research, with Suno (the AI music app) as the worked example.
You don't need a finance degree. You need six questions, a good research prompt, and the habit of keeping three things separate: what you know, what you're assuming, and what you think. By the end you'll have filled in a real memo, and you'll have the blank template to do it again for any company.
What you'll have at the end
A short research memo on a company of your choice, with sources and dates, covering what it sells, who pays, why they stay, and what could break it. Below is a worked example on Suno so you can see the shape before you build your own.
Write down what's known and what you're assuming, and keep them visibly apart. You'll quickly see which parts of your argument still need evidence. That habit is most of the skill.
Is this for you?
The six questions
Run every company through the same six. Keep the answers plain.
What does it sell?
In one sentence, what does the product do for someone? Describe it the way a customer would.
Who pays?
Who's the customer, and who signs the cheque? Selling to individuals, to teams, and to enterprises are three different businesses, even with the same product.
Why do they need it?
What painful, expensive, or repetitive thing does it remove? Would the customer genuinely miss it tomorrow, or is it a nice-to-have?
How does it find customers?
Word of mouth, free-to-paid, enterprise sales, paid ads? A company that can reach customers cheaply has more room to make money, so this often matters as much as the product.
Why would customers keep choosing it?
Why might it still be winning in three years? Switching costs, data, brand, owning more of its own stack. Be honest about which actually apply.
What could stop it growing?
For AI companies especially: does it depend on someone else's model? What does it cost to serve a dollar of revenue? Is there a legal or regulatory question hanging over it?
The research prompt
Paste this into your AI assistant, swap in the company name, and let it pull a first draft. Use the first draft to organize your research, then open the sources and check the claims yourself. Every claim needs a source before it earns the word "fact."
You are helping me research a company like an investor would.
Company: [COMPANY NAME]
Give me a sourced first draft, and for EVERY claim, tag it as
[FACT] (with a source + date), [COMPANY CLAIM] (from the company
itself), or [ESTIMATE] (reported or estimated, not confirmed).
If a field is unknown, write "unknown" rather than guessing.
If sources disagree, show the range and both sources. Do not
smooth over gaps. Tell me what you couldn't find.
Answer these six questions:
1. What it sells, in one plain sentence.
2. Who pays: customer type(s) and current pricing tiers, with the date checked.
3. The problem it solves, and how badly customers would miss it.
4. How it reaches customers: its distribution.
5. What could make it defensible: only moats that actually apply.
6. What could limit it: dependence on third-party AI models, cost to
serve $1 of revenue / gross margin, competition, and any legal or
regulatory risk.
Then give me:
- A 5-row economics summary (revenue estimate + date, cost to serve,
gross margin positive/negative/unknown, why customers stay, what the
business keeps).
- A one-line take: great product, great business, both, or not
enough information yet, and say what evidence would change it.What to do with it
- Run the prompt, then open the company's own pricing and product pages yourself. Prices and plans change, so check them on the day and write down the date.
- Verify the few numbers that matter (revenue, last funding round, anything about margins or legal exposure) against a named source. If you can only find it on an aggregator site, treat it as an estimate, not a fact.
- Fill the memo template below, sorting each line by how solid it is: a sourced fact, a company claim, an estimate, or your own read.
- Write your one-line take last. If you can't say it in a sentence, keep researching.
Big acquisitions, court rulings, sudden pricing changes. These move fast and get reported before they're settled. A reported headline is a lead, not a fact. Before you rely on anything dramatic and recent, chase it to a primary source: a court filing, the company's own post, an official ruling. If you can't confirm it, label it reported and unverified, and move on.
A worked example: Suno
Here's the framework applied to Suno, the AI music app. You type a prompt and it produces a full song with vocals. I picked it because it's a great teaching case: a product people clearly love, reported revenue, and one big question sitting over all of it. I've filled in the parts that matter most below; the blank template has room for the rest. Verify anything you'd act on.
The evidence ledger
Here's the same information, sorted by how solid it is: what's a sourced fact, what's the company's own claim, what's a reported estimate, and what's my read.
When I say what a business keeps, I mean revenue minus the direct cost of serving customers, its gross margin, not final profit. Salaries, marketing, and everything else still come out of that. It's a useful starting point, not the bottom line.
This space moves fast, the legal picture especially. New suits, rulings, and licensing deals land often and change the math. Before you use any of this for a real decision, check the current status against primary sources.
Why it works
The six questions walk down the chain that actually sets value: a product creates demand, demand at a price creates revenue, revenue minus the cost to serve is what the business keeps, and whatever it keeps is only safe if something makes it defensible. Headlines usually describe the top of that chain. What a business actually keeps is decided at the bottom.
Suno adds a sixth-question lesson most tool reviews skip: a legal or regulatory question can outweigh great product-market fit. People loving the product tells you demand is real. It does not tell you the company gets to keep selling what it sells. Sorting facts from assumptions from your own read is what keeps you honest about the difference.
Take the template
Here's the blank memo, ready to fill in for any company you're curious about. Same structure you just saw, including the ledger.
Business research memo (template)
Markdown, so it opens in Notion, Google Docs, or any notes app, and you can paste it straight into an AI assistant. Free to use and adapt for your own work; please don't resell it. See the terms.
Sources and updates
The key figures in the example are tagged and linked below. A note on status: I've verified these sources, but I haven't run a fresh end-to-end walkthrough with a reader yet. Figures were checked on 13 Sep 2026 and will drift, and the revenue and subscriber numbers are company-stated, not audited. Always re-check on the day.
- Suno pricing & product, suno.com/pricing (checked 13 Sep 2026)
- $400M round at $5.4B valuation, Variety, June 2026, link; Fortune, June 2026, link
- Revenue / subscribers (company-stated) & company history, Contrary Research, link
- Copyright cases (UMG/Sony, expansion, German ruling), AI Lawsuit Tracker, link; background, Wikipedia
Author: Riya Daftary. Reviewed 13 Sep 2026. Educational content, not financial advice. Found something I should update? Tell me →