How to Pick Your Next Side Project (A Decision Framework)

Most people with more than one idea use the wrong filter to choose between them.

They pick the one they're most excited about. Or the one that seems most technically interesting. Or the one a friend said sounded cool. Or they pick based on market size estimates from a Google search and a Statista report that may or may not be accurate.

None of these filters reliably identify which idea will be worth building. Excitement fades. Technical interest isn't correlated with commercial viability. Friends are unreliable market validators. Market size estimates are accurate only at scales too large to be useful for a solo founder.

What does reliably separate the ideas worth pursuing from the ones that look good until you start working on them: five specific questions, each probing a different dimension of whether the project is genuinely sound.

The Five Filters

Filter 1: Do You Have a Distribution Advantage?

This is the first filter and the most underrated one.

Every product needs distribution. Distribution -- getting the product in front of people who have the problem -- is often harder and more expensive than building the product itself. The founders who figure this out late spend months building something excellent that nobody knows about.

The question is not "can I build distribution?" It's "do I already have it, or do I have unusually good access to the community that would buy this?"

A distribution advantage looks like:

• You have an existing audience (newsletter, Twitter, YouTube, podcast) that includes your target customer
• You're an active, trusted member of a community where your target customer gathers
• You work in the industry your product would serve and have direct access to potential buyers
• You have specific relationships with people who are influential to your target customer

If you have a clear answer to "how will the first 100 people find out about this?" that doesn't involve paid ads or cold outreach to strangers -- you have a distribution advantage. Build that idea first.

If your answer is "I'll figure out distribution once the product is done" -- the project is significantly riskier than you're accounting for. That doesn't mean kill it, but it means weight it lower against an alternative where you have distribution clarity.

Filter 2: Do You Have Genuine Problem Familiarity?

There's a difference between knowing a problem exists and knowing how it actually works.

Knowing it exists: "I've heard that invoicing is annoying for freelancers."

Knowing how it works: "I spent three years freelancing, chased payment on 40% of my invoices, lost one client relationship because of a payment dispute I handled wrong, and still don't have a clean system for knowing which clients are overdue without opening a spreadsheet."

The second person has problem familiarity. The first has problem awareness.

Problem familiarity is valuable because:

1. It gives you accurate vocabulary. Your copy and positioning will use language your customers actually use rather than the language you think they use.
2. It makes customer interviews faster and richer. You can recognize meaningful answers because you have context against which to evaluate them.
3. It sustains motivation during the hard middle of building. When you personally know the problem is worth solving, the slow weeks feel purposeful rather than arbitrary.

The test: can you describe a specific time when this problem cost you something -- time, money, or an outcome you cared about? If yes, you have problem familiarity. If your knowledge of the problem comes primarily from reading about it or talking to people who have it, you don't. That's workable but it's a weaker starting position.

Filter 3: Is the Buyer Obvious and Can They Pay?

A product without a clear buyer is a science project.

The buyer question has two parts. First: who would pay for this? Second: are they able and willing to pay at a price that makes the project viable?

Ability to pay is a function of budget and priority. A small business paying $300 per month for accounting software is able to pay for software that solves their problems at similar prices. A college student who spends $5 per month on apps has a very different ability-to-pay ceiling.

Willingness to pay is about whether the problem is painful enough and the solution specific enough that the buyer would exchange money for it. People will pay for things that save significant time, generate significant revenue, reduce significant risk, or produce significant pleasure. Problems that sit outside those four categories are very hard to monetize.

The specific questions:

• "Who is the person who would pay for this, by job title or specific description?"
• "What budget category does the payment come from?" (Operational budget, personal spending, team tools allocation)
• "What is a realistic monthly or annual price, and does that amount matter to the buyer at all?"

If you can't answer these with reasonable specificity, the idea hasn't been thought through enough to evaluate against other ideas.

Filter 4: Can You Ship a Testable Version in 4 Weeks?

A testable version is not a finished product. It's the minimum experiment that would tell you whether demand is real.

For many products, this is a landing page with an email capture form and a 3-email nurture sequence. For some products, it's a manual version of the service before the software is built. For others, it's a prototype that does one core thing well enough for five people to try it.

The constraint is four weeks. Not because four weeks is objectively right, but because it's the outer boundary of a testable hypothesis cycle for a part-time founding context. Beyond four weeks, you're either building something too complex for the stage or spending too much time on something that hasn't been validated.

The filter question: taking stock of your current skills, available time, and the technical requirements of the project -- can you get to a testable experiment in four weeks or fewer?

If yes: proceed. If no: is there a scoped-down version that gets to the test faster? If yes: scope it. If no: the project requires more resource than you currently have. Consider it lower priority until your situation changes.

Filter 5: What's the Learning Outcome If It Fails?

This filter is the one most founders skip.

Every project fails to find revenue in the majority of scenarios. Before committing, ask: if this project produces no paying customers after three months of good-faith effort, was the time worth spending anyway?

The answer should be yes for one of these reasons:

• The skill development from building it is directly applicable to the next project
• The audience or community relationships you'd build are valuable independent of this product
• The research into this problem domain gives you knowledge that informs your next attempt in the same space
• The project produces something that has personal value (a tool you'd use yourself) even without customers

If the answer is no -- if the only value is commercial success and failure leaves you with nothing useful -- then the project has no floor. The downside is pure time loss. Projects with no floor require very high confidence of success to be worth starting. Adjacent projects with high learning floors are better bets for early-stage builders.

The Scoring Matrix

Run each idea through the five filters and score each on a 1-3 scale:

Filter	1 (weak)	2 (moderate)	3 (strong)
Distribution advantage	No existing access	Some adjacent audience	Direct access to buyers
Problem familiarity	Read about it	Talked to people with it	Have the problem yourself
Buyer clarity + ability to pay	Unclear	Somewhat clear	Very clear, verified
4-week testability	Requires months	Could scope down	Testable as-is
Learning floor	Nothing if it fails	Some skill gain	High learning regardless

Maximum score: 15. Minimum passing score for the idea worth pursuing next: 10 or above, with no individual filter below 2.

An idea with a 2 in distribution, 2 in familiarity, 3 in buyer clarity, 3 in testability, and 2 in learning floor scores 12. That's worth pursuing.

An idea with 3 in testability and buyer clarity but 1 in distribution, 1 in familiarity, and 1 in learning floor scores 9. It might still be the right idea -- but it's higher risk than the 12-score alternative if you have one.

The Ideas to Be Cautious About (The Anti-Portfolio)

Some idea types consistently score poorly on these filters despite seeming attractive.

"I'll build a platform that connects X with Y": Marketplace businesses have a brutal distribution problem (both sides need to join, neither will without the other), require high familiarity with both sides of the market, and are almost never testable in four weeks. Very hard to pursue as a solo side project.

"The [existing popular product] but for [underserved niche]": Often high buyer clarity but low distribution advantage (there's a reason the big product isn't serving the niche) and moderate familiarity. Usually viable but underestimated in difficulty.

"An AI wrapper for [existing task]": Low barrier to build, moderate buyer clarity, but very low learning floor (the AI layer is commoditized; what you learn is minimal) and high testability but also high competitive risk.

"A developer tool that solves a problem I personally have": Often strong on familiarity and testability, potentially strong on distribution (developer communities are accessible), and high learning floor. Usually underestimated. Worth more consideration than founders give it.

The 48-Hour Sit

After scoring your ideas and identifying the top candidates, don't immediately start building the one with the highest score.

Sleep on it for 48 hours. During those 48 hours, notice whether you're thinking about it when you're not actively working on the evaluation. Whether the idea surfaces when you're doing something else. Whether you're inclined to talk about the problem with people you encounter.

This is not mysticism. It's testing for the genuinely sustainable form of motivation -- the kind that persists past the first week when the novelty wears off and the actual work begins.

The highest-scoring idea that you're not inclined to think about spontaneously will underperform a slightly lower-scoring idea that you are genuinely curious about. Scores are a framework, not a verdict. They make the decision better than pure intuition. They don't make it mechanical.

Run the filter. Score the ideas. Sit for 48 hours. Then pick one and give it your full focus for the month.

The worst outcome is analyzing indefinitely and starting nothing. The second worst is committing to the idea that excitement selected rather than evidence screened. The framework is the path between them.