Category: Analysis
(8 von 100)
Why: We need to learn from both our successes and failures.
Table of Contents
In Short
Want to find out if your offer really solves customer’s problems? Experiment!
Measure the effect of the offers(treatments) on certain assumptions (outcomes).
If you already have evidence of effectiveness, then experiment with it to check its validity.
Step 1. Choose Variables
The experiment is usually the control of one set of variables and observes the consequences.
We need to control only one set at a time because we want to distinguish its effect.
- Outcome variable: measure the effect of manipulation.
- Treatment variable: manipulated variable, ones that the experimenter wants to explore.
- Control variable: measured and kept constant by the experimenter to provide reliable environment.
- Extraneous variable: uncontrollable parameters. To be noted, the effect of the treatment variable is always biased.
Step 2. Design Experiment
Experimental designs are how do you measure the effect of treatment?
There are two main ways you can do it.
- Wooden subject variation (Before vs After)
This is where the same subject is measured in terms of the outcome variable to capture the difference of the after outcome with the before outcome. - Between-subject variation (A vs B)
This is where we actually measure two different individuals or two different groups of people. But these two groups being affected by the same manipulation differently.
Conclusion
To learn about our customer’s real needs, we need to put them to the test. Here, we suggest 2 causal experiment designs, a wooden subject to learn about an outcome, and a between-subject to learn about the different reactions from different groups of customers.