Every growth team has the same problem. Twenty ideas on the board, five people, one week, The question is never what to test. It's what not to test. Here's the framework I use to prioritize ruthlessly.

In any growth or marketing team working at a decent pace, you generate ideas faster than you can execute them. That's healthy. It means the team is thinking, But it creates a different problem: the idea-to-execution ratio goes upside down. You end up with a Notion board with 40 ideas and a team that ships 2 of them per month.

Most of those ideas are probably fine. Some are great, A few are terrible, The job isn't to test all of them. It's to figure out which 2-3 to run this month, confidently, so the rest of the team can focus.

Here's the prioritization framework I use to cut a list of 20 ideas down to the 3 that matter.

Why most frameworks fail

The standard advice is ICE (Impact, Confidence, Ease) or RICE (Reach, Impact, Confidence, Effort). I use these with clients all the time. They're fine, But they have one big problem: they treat every idea as independent.

In reality, your 20 ideas aren't independent. They're mostly small variations of the same 4-5 underlying bets. You don't really have 20 ideas. You have 4 ideas with 5 variations each, The first step of prioritization is to see that clearly.

Step 1: Cluster, don't list

Before ranking anything, group your 20 ideas into clusters by underlying hypothesis. What's the actual bet behind each idea?

Example: imagine a B2B SaaS growth team with these 8 ideas:

That's 8 ideas but really 3 bets:

  1. Bet A: Landing page conversion is too low. (Headline change, social proof, form simplification, video redo)
  2. Bet B: The top of funnel is too small. (LinkedIn ads, SEO pillar, podcasts)
  3. Bet C: The pricing page is confusing. (Simplified pricing)

Now you're making 3 decisions, not 8. Much more tractable.

Step 2: Pressure-test the hypothesis

For each bet, ask:

  1. Is the underlying problem real? What's the evidence that this is actually broken? Is landing page conversion too low, or is that an assumption?
  2. If we fix this, does the primary metric actually move? Even if the landing page converts 2x better, does that meaningfully move pipeline? Or is the volume so low that a 2x lift is a rounding error?
  3. Is there a faster/cheaper way to test the hypothesis? Before you redesign the whole landing page, could you test the headline change alone?

This step usually kills 30-40% of your ideas, Not because they're bad, Because they're solving problems that don't really exist or won't move the metric, An experiment that can't meaningfully move your primary metric is a distraction wearing a hypothesis t-shirt.

The test I use: if this experiment succeeds beyond our best-case expectation, would anyone outside the marketing team even notice? If the answer is no, park it.

Step 3: Score on effort vs. upside (not ICE)

For the bets that survive, score them on two axes only:

Skip "confidence" as a separate axis. Confidence is already baked into your upside estimate. If you're not confident, lower the upside.

Plot them, The bets in the top-left quadrant (high upside, low effort) run first, The bottom-right (low upside, high effort) never runs, The diagonal (high upside + high effort, or low upside + low effort) is where judgment matters.

Step 4: The "what happens if we don't do this" test

Here's my favorite prioritization trick. For each top candidate, ask: what's the cost of not doing this for 6 months?

All else equal, run the irreversible and opportunistic bets first. Save the timeless ones for slower weeks.

Step 5: The 30-day kill rule

Once you pick 2-3 experiments, set a 30-day check-in. If by day 30:

The worst outcome is a half-built experiment that lingers for months. Kill it. Move on.

What changes when you start prioritizing

Teams that adopt tight prioritization get three benefits that compound:

1. Focus. Everyone knows what matters this month. Nobody is spread across 6 half-done experiments.

2. Speed. The cycle from idea to learning gets shorter. Fewer experiments = more iterations.

3. Honest conversations. When you only run 3 experiments, the team can actually talk about why they worked or didn't. Thirty experiments running in parallel is too noisy to learn from.

The job of prioritization isn't to pick the "best" idea. It's to make sure the 3 you run are the ones that teach you the most about your business this month.

The short version

When you have 20 growth ideas: cluster them into underlying bets, pressure-test the hypothesis, score on upside vs effort, apply the "what if we don't" test, and commit to 2-3 experiments with a 30-day kill rule. Better to ship 3 experiments with real clarity than test 12 you'll half-understand.

Growth isn't the team with the most ideas. It's the team that picks the right ones and actually learns from them.