How do I measure a brand campaign?

You cannot pause it, so extend the pre-period trend and compare.

Brand search rose—was it the campaign?

Not provably. PR, launches, and seasonal demand create the same shape. A control series or geo split is needed to remove shared movement.

How much history do I need?

Enough to estimate the pre-intervention trend. With only a few pre-period points the slope estimate is unstable and no conclusion is warranted.

Why correct for autocorrelation?

Adjacent periods in a time series resemble each other, so uncorrected standard errors come out too small and results look more significant than they are.

Brand lift

Brand campaign incrementality

Compare brand search, direct traffic, and signups against the pre-campaign trend.

What data do you have?

★ I have a control region or audience

A concurrent unexposed comparison enables the strongest holdout / DiD design.

Open control-group incrementality

△ I only have before / after totals

An exploratory pre/post read is possible, but seasonality and common change remain mixed.

Open pre/post comparison

Prepare ITS data

Check whether brand search, direct traffic, or signups actually rose after a brand campaign started, against a pre-period trend baseline (ITS).Needs: Date · Outcome · Brand campaign status3 required columns

⇧

Drag & drop a CSV

or click to choose a file

Loading…

Reuse this analysis setup

Keep the mappings, filters and model options in your project.

Open saved setups

Required to saveProProject with device storage

Next step

Proof

Common next stepPrepare a new dataset

Incrementality analysis

Find other analyses