Skip to main content

Part of WatEase, India’s agentic AI commerce and marketing platform

Causal Inference

Statistical methods that estimate causal effects from observational or experimental data.

Last updated

Definition

Five methods are most commonly applied in marketing: Difference-in-Differences (DiD), Synthetic Control Method (SCM), Propensity Score Matching (PSM), Regression Discontinuity Design (RDD), and Instrumental Variables (IV). Each suits a different data shape — DiD needs treated + control groups around an intervention; SCM builds a synthetic counterfactual from untreated cohorts; RDD exploits sharp thresholds (free shipping over ₹999). Together they triangulate true incremental lift from observational marketing data.

How it applies in India

Every Indian delivery address carries a PIN code, so geography is a practical unit for a causal test: some towns can be held back from a campaign and compared with the rest.

An example

A brand’s sales rise 20% in the month it starts WhatsApp broadcasts. Festive demand, a price cut and a new product all landed in the same month, so the rise on its own proves nothing. Causal inference is the discipline of separating the broadcasts’ share from everything else, ideally by comparing people who received them with similar people who did not.

How WatEase uses it

WatEase draws its causal conclusions from experiments rather than from modelling observational data. It uses randomised holdouts (shoppers chosen by a stable hash on recovery messages, offers and broadcasts launched by agents), the town-level difference in differences of the quarterly geo-lift test, and two-proportion tests for campaign holdouts, each reported with a figure for how sure the result is. It does not use propensity matching, regression discontinuity, instrumental variables or synthetic controls. Its methods are published on the research page.

Frequently asked questions

Which causal methods does WatEase use?

Experiments: randomised holdouts, a town-level difference in differences for ads, and two-proportion tests, each with a figure for how sure the result is. Data-driven attribution is observational, so WatEase caps the credit it gives ads at the lift the latest holdout measured.

Related terms

Try WatEase

Run your WhatsApp commerce on the platform built for India — Cloud API, GST invoices, UPI checkout, opt-in tracking, and 0% markup on WhatsApp messages (competitors add 20–25%). Pricing is on request — quoted by the WatEase team.

Get Started Today