Faisal Mahmud TusherDigital marketing, SEO & analytics

WebmailLet's Talk

with Faisal Tusher

Resources · Experiments

Marketing Experiments

Opinions are cheap; experiments settle them. The discipline is in the design — a hypothesis, one variable, enough data.

Try the interactive tool

Faisal Tusher at a heritage building

Marketing Experiments, my way

5

building blocks

2

KPIs

3

tools

The building blocks

What goes into Marketing Experiments

Hypothesis
Because we observed X, we believe Y will improve Z.
One variable
Isolate what changes.
Sample size
Enough conversions to detect the effect.
Significance
Avoid calling winners on noise.
Learning log
Record results — including failures.

How I do it

The process

1
Hypothesis
Evidence-based.
2
Design
Variants & metric.
3
Run
Until sample size.
4
Analyse
Significance.
5
Log
Learning.

Principles I follow

Prioritise tests with ICE.
Pre-register the success metric.
Don't stop tests early on a lucky streak.

Measurement

The numbers I watch

Test velocity
Tests per month.
Win rate
Share of tests with significant improvement.

Tools I use

G4Google Analytics (GA4)
AMMeta Ads Manager
GAGoogle Ads

What you get

Test roadmap
Learning log

Creative test

Two ad concepts, similar CTR after 2 days.

Keep running until each has enough conversions; check significance before deciding.

Illustrative example

What confidence level should I use?

95% is conventional; lower it only knowingly for low-risk decisions.

Need help with Marketing Experiments?

Let's talk about what you're trying to achieve.

Related