Managing ideas and hypotheses

What you'll understand by the end of this lesson

• Why a simple list is the most powerful tool a CRO practitioner owns
• How to turn a vague idea into a testable hypothesis using a single sentence
• Why writing one hypothesis often reveals three or four testable ideas hiding inside it
• How to rank competing hypotheses so the most valuable ones get tested first

---

Ideas are not the problem

Every team working on a website has more ideas than time. The designer wants to simplify the layout. The marketer wants a stronger headline. The CEO just saw a competitor's site and wants to know why yours looks different.

Without a system to manage ideas, you end up in one of two equally bad positions:

• Ideas get lost. The good observation from Monday's customer call doesn't make it to Friday's planning meeting.
• The loudest idea wins. Whoever is most senior or most persistent gets their idea prioritised — not whoever has the strongest evidence.

CRO's answer to both problems is the same: a structured idea management process. Three components. A list to capture everything. A hypothesis format to make ideas specific. A scoring framework to rank them honestly.

---

Step 1: Keep a list

The first tool is embarrassingly simple. You need a place — a spreadsheet, a Notion table, a shared doc — where every idea goes when it surfaces.

Each row captures four things:

Field	What to write
The idea	A brief description of what someone is suggesting
Who suggested it	Name or role — useful for follow-up
What it addresses	Which page, flow, or campaign
Status	Raw idea / hypothesis written / in test / done

The list does two jobs simultaneously.

It's your work queue. When you're deciding what to test next, you pull from the list — you're not inventing priorities from scratch.

It's a professional buffer. When someone walks in with an idea they saw on a competitor's site, you don't have to argue or immediately drop what you're doing. You say: "That's worth looking into — I'll add it to the list and research it." The idea gets a fair process. Your current work doesn't get disrupted.

---

Step 2: Write a hypothesis

Once an idea is on the list, the next step is turning it into something specific enough to test.

The problem with raw ideas is that they're almost always vague. Consider:

• "We should refresh the website."
• "The homepage headline doesn't feel right."
• "Our competitors have video on their product pages."

None of these can be tested — none of them say what to change, what would improve, or how you'd know if it worked. They're opinions, not experiments.

A hypothesis forces specificity. The format is:

If I [make this specific change], I expect [this specific outcome], as measured by [this specific metric].

"Refresh the website" becomes three distinct hypotheses:

• If I remove three fields from the sign-up form, I expect completion rate to increase, as measured by GA4 goal completions.
• If I add whitespace between sections on the product page, I expect users to scroll further, as measured by average scroll depth.
• If I rewrite the hero headline to focus on the outcome rather than the feature, I expect click-through on the primary CTA to increase.

Each one names a change. Each one makes a prediction. Each one says how you'll measure success.

---

Step 3: Score with ICE

After converting ideas into hypotheses, the challenge is: which one do you work on first?

The ICE framework gives you a consistent way to rank. For each hypothesis, score three things from 1 to 5:

Impact — If this test wins, how much will it affect the metric? A change to the hero headline on your highest-traffic page scores higher than a change to a footer link.

Confidence — How much evidence already exists that this problem is real? A hypothesis backed by session recordings and customer interviews scores 5. One based on a gut feeling scores 1.

Effort — How much work does building and running this test require? Effort works in reverse — low effort pushes an idea higher. A button label change is a 1. Producing video for every product page is a 5.

Add the three scores and sort highest to lowest. The top-scoring ideas are your starting point.

Three things worth knowing about ICE

Consistency matters more than accuracy. You're not predicting the future — you're creating a shared language for comparing ideas. Score consistently, and the ranking becomes useful.

It tells you where to research first. Do a quick first pass on all hypotheses, then spend time on the top candidates. Check your analytics, watch session recordings. As evidence grows, update the confidence scores.

Low-effort ideas naturally float up. A hypothesis with moderate impact and high confidence becomes very attractive when it's also quick to build. These are your early wins — they build momentum and buy credibility for bigger bets later.

---

The full workflow

Here's how the three steps connect in practice:

1. Capture every idea without filtering. Ideas come from data, colleagues, customer calls, your own observation. None of them are wrong at this stage — they're all inputs.
2. Write a hypothesis for each idea. Not all ideas survive this step — some turn out too vague. Discovering that early is valuable.
3. Score on ICE — a quick first pass. Don't overthink the numbers. Your goal is relative ranking, not absolute prediction.
4. Research the top candidates. Look at analytics, watch session recordings, review customer feedback. Update confidence scores as evidence grows.
5. Run the highest-scoring, best-evidenced hypothesis. That's your next test.

Every test teaches you something. Over time, the backlog gets sharper, scoring gets more calibrated, and the win rate rises.

---