Skip to content
EN
BOOK A CALL
Growth consulting

Growth Experiments: ICE Prioritization, a Two-Week Test Cycle and a Shared Test Log

Teams rarely run out of ideas. What they lack is a way to decide which idea to test first, and a habit of learning from what has already been tested. Our growth experimentation process ranks ideas with the ICE framework, launches them in two-week cycles and records every test in one shared log. Even a losing test makes the next decision easier, as long as it is written down.

Book a growth call Read the details
IdeaImpactConfidenceEaseICEOne-page checkout9757,0Delivery time6897,7Cart reminder7877,3New channel7445,0
Illustrative view: illustrative view of an ICE scoring table.
Purpose and appearance

What do we use this service for?

We build the experiment system so that budget goes only to ideas with evidence behind them. We test ideas that affect conversion rate, average order value or repeat purchases in two-week cycles, scale the winners and carry what the losers teach us into the next round of prioritization. Team meetings discuss recorded test results instead of debating ideas.

Funnel stages it serves

  1. Awareness
  2. Consideration
  3. Conversion
  4. Loyalty

How does ICE scoring work, and how do we keep it objective?

The ICE framework scores each idea from 1 to 10 on Impact, Confidence and Ease and takes the average. Its weak spot is well known: scores inflate with the enthusiasm of whoever proposed the idea. We prevent this by tying Confidence and Ease to scales written in advance, and by scoring Impact according to which branch of the KPI tree the idea touches and whether that branch is the current bottleneck.

  • Confidence 8-10: we have seen a similar result in our own data or in a previous test
  • Confidence 5-7: supported by session recordings, heatmaps or customer reviews
  • Confidence 1-4: we have only seen it at a competitor, or it rests on intuition
  • Ease 8-10: one person can do it within a few days without a developer
  • Ease 1-4: requires a developer sprint, new creative production or legal approval

Our two-week growth experiment cycle

On the first day of the cycle, we pick one to three tests from the top of the ranking and write a hypothesis and success criterion for each. Tests go live within the first week. We don't change a live test before its run time is over; we stop it only if tracking breaks or cost exceeds a limit set in advance.

At the end of the cycle, a 30-minute meeting answers three questions: what did we test, what did we learn, what will we do next round. A winning test goes into the marketing plan and the budget; a losing test is closed, but what it taught us changes the Confidence score of the next idea. We show how the two-week rhythm fits together with the monthly plan and the quarterly strategy review in our growth approach.

Writing hypotheses and keeping a growth experiment log

We write every hypothesis in one format: 'If we change X, metric Y will change, because Z.' If the 'because' part is empty, the idea isn't ready to test yet. The log also records the test's date, channel, owner, measured metric, result, decision and a screenshot. Six months later, the answer to 'didn't we try this already?' comes from this table.

We prefer the platforms' own testing tools. In Google Ads, custom experiments split traffic between the original campaign and the experiment arm; in Meta, A/B tests separate audiences so the same person doesn't see both versions. Landing page and checkout changes we set up together with our A/B testing service.

Growth Experiments (ICE): when do we recommend it?

We recommend it

  • Teams with plenty of ideas that decide which comes first by debating it in meetings
  • Brands that can't measure, or can't remember, the outcome of the changes they make
  • Growth teams that want to run tests across several channels at once

We don't recommend it

  • Sites whose weekly conversion volume is too low for a statistical comparison; here, instead of testing, we recommend applying well-evidenced changes directly and tracking before and after
  • Accounts where conversions in the ad platform don't match order data; the winner of any test would remain unclear

How we manage Growth Experiments (ICE)

  1. Idea poolWe collect ideas from teams, customer reviews and data in a single list.
  2. ICE scoringWe score with written scales and move ideas that touch the bottleneck branch to the front.
  3. Hypothesis and criteriaFor each test we write the hypothesis, primary metric, duration and stopping condition.
  4. Launch and monitoringWe launch the test with the platform's own experiment tool and verify on day one that tracking works.
  5. Cycle meeting and logWe review the result with three questions and record the decision and learning in the test log.

Growth Experiments (ICE): metrics and requirements

Metrics we track

  • Tests completed per cycle
  • Winning test rate
  • Impact of winning tests on the KPI tree
  • Time from idea to launch
  • Number of tests recorded in the log

What you need to get started

  • Reliable conversion tracking
  • Permission to create experiments in the ad accounts
  • A decision-maker who attends the cycle meeting
  • A shared spreadsheet for the joint test log

Official sources

The Brand Move Growth Consulting Team

To learn more about Growth Experiments (ICE) or other growth consulting services, contact The Brand Move Growth Consulting team today.

Contact us now
Explore further

Other Growth Consulting services

Growth Consulting overview

Growth Strategy

Growth strategy consulting that ties every team to one goal through a North Star metric, a KPI tree and a channel role map.

Explore

Marketing Planning

A 90-day marketing plan with weekly owners and a TL budget calendar adjusted for seasonality, turned straight out of your strategy.

Explore

Related pages

Frequently asked questions

Growth Experiments (ICE): frequently asked questions

What is the ICE framework?

ICE is a prioritization method for growth experiments. Each idea gets a score from 1 to 10 for Impact, Confidence and Ease, and the average decides the order of testing. We tie the Confidence and Ease scores to written scales so the ranking doesn't depend on who argues loudest.

Can we use RICE or another scoring model instead of ICE?

Yes. RICE adds reach as a separate criterion and works well for products with a large user base. For most marketing teams, ICE is fast and simple enough. Whichever method you choose, tying the scores to written scales is what makes the real difference.

Are two weeks enough for every test?

Not always. Two weeks is the cycle's review rhythm; a low-volume test can stay open across several cycles. We don't make a decision before the result is meaningful; in the meeting we simply mark it as 'ongoing'.

What tools do you use to run growth experiments?

We prefer each platform's own testing tools: custom experiments in Google Ads, A/B tests in Meta, and on-site tests for landing pages and checkout. The test log itself lives in a shared spreadsheet that the whole team can read.

Are losing tests of any use?

Yes, as long as they are recorded. A losing test lowers the Confidence score of similar ideas and stops the team from proposing the same idea again six months later. The losing tests in the log are often the entries that save the most time.

Let's run a first Growth Experiments (ICE) review together.

In a free call we listen to your goals, channel mix and current plan, then send you the three highest-priority steps in writing.

Book a growth call No sales pressure. Just findings.