Proof of concept on synthetic data. This page is a demonstration, not a production system.
Greenlight Content Portfolio Intelligence
A decision tool for content finance

Which shows and films actually make money?

Greenlight is a simulated streaming service, modeled end to end. It shows which titles earn their keep — once you account for what they cost to make, how that cost is spread over the years people keep watching, and how much they’re actually watched. Ask it anything in plain English, see the exact query behind every answer, and test “what if we spent differently.”

As of —· — titles· — monthly facts· engine loading…
The idea in 60 seconds

What you’re looking at

1 · The setup

A made-up streaming service with a $16B slate — about 300 shows and films, across four regions and several languages. Every figure on this page is calculated live from that data.

2 · How the money works

You spend to make a title (its content cost). Because people keep watching it for years, that cost is spread out over time — that’s amortization — instead of hitting all at once. Each title is then credited a share of subscription revenue based on how much it’s watched. So profit = credited revenue − the spread-out cost − marketing. When you see “EAM,” that’s just this profit number.

3 · Why you can trust it

Every number has one official definition. When you ask a question, the tool turns it into a real database query, shows you that query, and explains the result in plain words. If a question is unclear, it asks instead of guessing.

The slate

Where the money is — and isn't

Series makes most of the profit, Film currently loses money, and Unscripted is small but efficient. Everything below is the same handful of numbers — each defined once — sliced different ways. “Profit” means credited revenue minus the spread-out content cost and marketing.

Profit by type of content

Profit (credited revenue − spread-out cost − marketing) for Series, Film and Unscripted.

How revenue turns into profit

Start with the revenue the slate earned, take out cost and marketing, and this is what’s left.

Profit by quarter

Profit booked each quarter as titles age and their costs get written down.

Profit per $1 of spend, by genre

For every dollar of content spend, how much profit each genre returns (its ROI).

The slate, title by title

Each block is a large title — bigger = more expensive; teal = making money, clay = losing money.

Cost vs. profit, per title

Each dot is a title: what it cost (left→right) against the profit it made (up→down). Dots below the line lost money.

Profit by region

The four reporting regions — U.S. & Canada, Europe/MEA, Latin America, Asia-Pacific.

Make vs. buy

For every dollar spent, do titles we produce ourselves (Originals) or license from others return more profit?
The analyst

Ask in plain English. Get an answer you can check.

This analyst is a rule-based resolver, not an LLM: it matches your words against a controlled vocabulary of defined metrics. Type a question the way you’d say it out loud. The tool works out which official number you mean, runs the real query (and shows it to you), and explains the answer in a sentence. If the question is vague, it asks you to be more specific rather than inventing a number.

▸
The scenario studio

Move the money around. Watch profit change.

From the history, the tool learned how much extra profit each category returns for each extra dollar — with diminishing returns as you pour in more. Drag the sliders to shift budget between Series, Film and Unscripted, or let it find the mix that makes the most profit for your budget.

Budget—
Baseline EAM—
Projected EAM—
Δ EAM—
Profit today → projected profit, by category
Move a slider or hit optimize to see the projected impact.
How it works

You can check its work

Nothing here is a black box. Your question is matched to one official number, run as a real query, and explained from the actual result — with the query shown, so you can verify every figure yourself.

01

Understand

Reads your plain-English question and maps it to one official number, plus any filters — region, genre, time period.

→
02

Use the definition

Pulls that number’s single official formula, so every answer uses exactly the same math.

→
03

Run it for real

Turns it into a database query and runs it in a real engine, right here in your browser.

→
04

Explain

Writes a plain sentence from the actual numbers — including why something went up or down.

→
05

Show the work

Puts the query and the result on screen, so you can check every figure yourself.

🔒

Defined once

Profit, margin and ROI each have one official definition. Two questions about the same thing can’t come back with different numbers.

🛑

Won’t guess

If your question doesn’t match a defined number, it asks a follow-up instead of making something up.

🧾

Shows its work

The exact query appears with every answer. You never have to take a number on faith.

Scope · what production adds
Greenlight is a proof of concept on synthetic data, built to demonstrate the approach end to end. In a production build the same design runs against the live content warehouse, the plain-English analyst becomes a governed LLM interface (constrained to the defined metrics so it still can't invent a number), and the scenario model is calibrated against actuals. The core idea — governed metrics → live compute → explained result — is what carries over; production just scales it.