Correlation vs Causation
AnalyticsAlso: Correlation Is Not Causation · Causal Inference
Quick definition
Correlation is when two things move together in a pattern. Causation is when one thing actually makes the other happen. Marketing data is full of strong correlations that get reported as causation without any test to prove it, which leads to decisions based on coincidence rather than evidence.
How it varies across Australia
Most Australian marketing teams still make budget decisions off correlation alone, especially with attribution reporting and dashboard trends. The gap between businesses that test causation with proper experiments and those that eyeball a chart is one of the clearest divides in data maturity we see across industries.
See data and tracking maturity across Australian industries →What it actually means
Ice cream sales and shark attacks both rise in summer. Nobody thinks ice cream causes shark attacks. The real driver is warm weather, pushing people toward both ice cream and the ocean at once. That's a confounder, a third factor causing two things to move together without either one causing the other.
Marketing dashboards produce this pattern constantly. Email open rates rise the same week revenue rises. Instagram followers grow the same month a product launches. It's tempting to draw a straight line between the two and call it proof. Usually it's two effects of the same underlying cause, like a seasonal spike or a broader demand shift that lifted every metric on the dashboard at once.
Attribution and correlation are close cousins here. Attribution decides which touchpoint gets credit for a conversion. It's still just describing a pattern in the data, not proving the touchpoint caused the sale. The only way to move from correlation to causation is a controlled test: hold a group back, change one variable, measure the difference. A-B testing does this. So does a proper incrementality test. Everything else is an educated guess dressed up as an insight.
A chart with two lines going up together is not evidence. It's a coincidence wearing a lab coat.
How it shows up
It shows up in the boardroom sentence 'our conversion rate went up when we launched the new brand campaign, so brand drove sales'. It shows up in SEO reports claiming a ranking jump caused a traffic spike when Google also rolled out an algorithm update that week. It shows up whenever someone says 'the data shows' and means 'two lines on a chart moved in the same direction'.
The Australian context
Australian retail and finance data is particularly prone to seasonal confounders. Tax time, EOFY sales periods, and the runup to Christmas all push multiple metrics up together regardless of what marketing did. Teams reviewing year-on-year data need to control for these known seasonal patterns before crediting any single channel or campaign with the lift.
Where people get this wrong
Correlation vs Causation vs Attribution
| Correlation vs Causation | Attribution | |
|---|---|---|
| What it answers | Do these two things move together | Which touchpoint gets credit for a sale |
| Proves causation? | No | No, on its own |
| Requires a test? | Yes, to move beyond pattern | No, it's a credit-assignment model |
| Common failure | Mistaking a shared driver for cause | Mistaking assigned credit for proven impact |
Related terms
Common questions
How can I tell if a correlation is actually causal?
Run a controlled test. Hold a comparable group back from the change, measure the difference between groups, and check the result is statistically significant. If you can't isolate the variable with a test, treat the correlation as a hypothesis, not a conclusion.
Isn't attribution modelling a way to prove causation?
No. Attribution assigns credit across touchpoints based on rules or statistical weighting, but it still describes a pattern in existing data. Proving a touchpoint actually caused the conversion requires an incrementality test with a holdout group, not just a model.
What's a confounder in marketing data?
A confounder is a third factor that causes two metrics to move together without either one causing the other. Seasonality is the most common marketing confounder, lifting sales, traffic and engagement all at once regardless of any specific campaign.
Why do marketing teams keep mixing this up?
Dashboards make correlation easy to see and causation hard to prove. A rising chart next to a campaign launch is instantly visible. Setting up a proper test takes planning, a holdout group and patience, so teams default to the easier, weaker evidence.
Debrief
Get the next one
No spam. No fluff. Just the next article, straight to your inbox.
Keep exploring
About New Rebellion
New Rebellion is a marketing intelligence consultancy. We build tools, score Australian businesses on how their marketing actually performs, and publish Debrief every day. This dictionary is part of how we work in the open.
How we think →