How Goodspeed Scores App Ideas: Behind the 100-Point Rubric

Why We Score Ideas

Most app ideas die because the builder never asked: "Is this actually worth building?"

They skip validation, spend months coding, launch to crickets, and conclude that the market is broken. But the market isn't broken. The idea was never tested against reality.

---

The 100-Point Framework

Every app idea that enters our pipeline gets a score out of 100. Most of it comes from three weighted parts that add up to exactly 100. Four adjustments then move it up or down, and an idea that gets deeper research can move a few points more. The result is kept between 0 and 100.

Each weight reflects how much its input actually separates one idea from another across the ideas already scored, adjusted where a measure turned out to mislead. A question that gives every idea the same answer carries almost no weight.

Market Opportunity (0-39 points)

Is anyone actually asking for this, and will they pay?

We measure demand by scanning real conversations wherever people air their frustrations:

• Hacker News discussions.
• App Store listings and their reviews.
• Product Hunt launches.
• GitHub issues and trending repositories.
• Stack Overflow questions.
• Y Combinator launches.
• Google Trends data.
• What people type into search autocomplete.
• WordPress plugin listings.
• Tech news feeds.

We're not counting keyword volume. We're reading actual conversations where people describe problems, frustrations, and unmet needs.

What earns the points, heaviest first:

• Willingness to pay: evidence that people already spend money on this problem.
• Demand signals: how strongly and how often people ask for a solution.
• Retention: whether people would keep coming back.
• Market size and growth: small on purpose, because almost every idea gets a similar answer here.

Competitive Landscape (0-36 points)

Competition is counterintuitive in our scoring. What matters most is how good the existing options are, not whether any exist.

What earns the points, heaviest first:

• Competitor quality: weak, clumsy or overpriced competitors score higher than strong ones.
• Competitor count: how crowded the space is. A count we could not verify can earn at most half of this, because an unverified count is a guess.
• Sentiment gap: how clearly users complain about what the current tools get wrong.

We look specifically at competitor weaknesses: low ratings, frequent complaints in reviews, and features users keep asking for. That's where the opportunity lives.

Monetization (0-25 points)

Can this app make money? Not theoretically. Actually.

What earns the points, heaviest first:

• Subscription fit: whether people would keep paying month after month.
• Revenue model: whether the way it would charge already works for similar apps.
• Revenue per user: kept small, because on its own it says little about how good an idea is.

The Four Adjustments

On top of the three parts:

• Build difficulty (down to -12): points come off when an idea is hard to build or keep running, relies on third parties whose terms can change, or risks a platform rejecting it. Being easy to build earns nothing, because an app that does nothing is the easiest of all.
• Solo viability (up to +5): a bonus when one person can realistically build, launch and market it.
• Moat (up to +3): a bonus for an advantage a fast follower cannot copy in a weekend, such as its own data, a network that grows with use, or real switching costs.
• Cold start (down to -4): points come off when the product is worth little until it already has content, supply or other users.

After Deeper Research

When an idea goes through deeper research, the score can move a few more points: up for strong early retention, healthy unit economics or word-of-mouth growth, and down for serious risks. These points are added on top of the breakdown above.

---

How the Score Is Produced

Scoring isn't a single step:

Step 1: Signal extraction. A model reads each signal (an app review, a Hacker News comment, a Stack Overflow question), identifies the core problem being discussed, and proposes app ideas that would solve it, each with its supporting evidence.

Step 2: Research. A second step researches each candidate and answers the rubric's questions: how many competitors there are and how good they are, what people already pay, how strong the demand is.

Step 3: The arithmetic. The score itself is fixed code, not a model's opinion. The same answers always produce the same score, and the breakdown you see shows which part earned or lost points. A deeper-research adjustment, when there is one, is added on top.

---

What Tends to Score Well

High-scoring ideas tend to:

• Address painful, recurring problems, not nice-to-haves.
• Target audiences that already pay for tools.
• Have a few competitors with clear weaknesses.
• Be something one person can ship and run.

Low scores usually come from one of:

• No sign that anyone pays for a solution today.
• A strong, well-loved incumbent.
• A product that is empty until lots of other people use it.

---

Transparency and Limitations

Our scoring system is not perfect. No scoring system is. Here are the known limitations:

• English-language bias: Our signal sources are primarily English. Ideas that address non-English markets may be under-scored.
• Consumer app bias: The rubric is tuned for consumer mobile apps. B2B enterprise ideas don't fit as neatly.
• No execution scoring: We score the idea, not the builder. A great idea poorly executed can still fail.
• A missing answer is not flagged: when a question cannot be answered, it scores like a middling answer rather than showing as unknown.

The rubric is versioned, and every scored idea records the version that scored it. The current version was rebuilt after we found the previous one scored very different ideas too closely together.

Want to see scored ideas? Check out our ideas page or learn more about how our discovery pipeline works. Sign up to score your own idea against the full rubric.