DLG Project: hypothesis, goal & success metrics
Great! Now we know the problem we're solving and it's impact. Next, let's define the finer details of the experiment:
1. Problem statement & overview
2. Hypothesis, goal setting & success metrics
3. Experimentation design: Part 1
4. Experimentation design: Part 2
5. Post experiment learnings & next steps
6. Stakeholder management
Hypothesis
A strong hypothesis provides a clear and testable assumption about how a change will impact the key metric.
Framework for writing a hypothesis:
If [we implement this change], then [this result will occur], because [reasoning based on data or user behaviour].
We learnt the concept of null hypothesis to set a solid base for building experiments. However, it might require you to educate your peers/stakeholders to convey the concept. So, for this project, we’ll use a standard hypothesis framework.
For example:
If we show the product demo & CTA in first fold
then our website visit to signup rate will increase by 5%
due to clear value prop delivered & CTA being the only focus
Goal setting
Write the goal of the experiment and how it aligns with the business objective.
Good experiments are run on bets that align with the business objective rather than optimising for random metrics.
Key Aspects of a Strong Goal:
- Aligns with overall business objectives.
- Is specific and actionable.
- Focuses on a core input metric rather than vanity metrics.
For an E-commerce website:
Running an experiment with the objective of improving session conversions in a particular category would probably align with the overall business goal of revenue rather than; running endless smaller optimisations that lack focus and don't contribute meaningfully to the overall business objective.
Always break down the business objective to input levers and run experiments only on those input levers.
Success metrics
Pre-define how you would measure success. Put down the “worst case” success number & the “best case” success scenario.
Define success metrics before launching the experiments.
And, say a big no to retrospective learnings. Focus on the “why” behind your tests, and clearly define what you hope to learn from each experiment.
For example;
Worst case scenario: absolute increase in conversion by 1.5%.
Best case scenario: absolute increase in conversion by 5%
Even in the worst case, if the experiment is successful, we will scale the experiment.
What that means is; you’ve already done the cost, effort & impact math to know that the 1.5% increase if worth scaling the experiment.
Brand focused courses
Great brands aren't built on clicks. They're built on trust. Craft narratives that resonate, campaigns that stand out, and brands that last.
All courses
Master every lever of growth — from acquisition to retention, data to events. Pick a course, go deep, and apply it to your business right away.
Explore foundations by GrowthX
Built by Leaders From Amazon, CRED, Zepto, Hindustan Unilever, Flipkart, paytm & more
Crack a new job or a promotion with the Career Centre
Designed for mid-senior & leadership roles across growth, product, marketing, strategy & business
Learning Resources
Browse 500+ case studies, articles & resources the learning resources that you won't find on the internet.
Patience—you’re about to be impressed.