Retailer: Meta makes people search for you. For about two days.

Case studies · done with EPYC · Ask EPYC

0.92correlation, effect gone after 48 hoursGranger causality · daily Meta spend vs branded search

What we found

Spend on Meta predicts branded search volume with a one-to-two-day lag and then the effect disappears. Read: the Meta spend was promotional — it makes people look, it does not build anything that lasts. The next question writes itself: which media, if any, sustains the brand. Answering it needs TV, out-of-home, and programmatic spend in the same table as Meta.

Illustrative analysis from a platform demo; the store is not a client.

Method

01
We pulled daily Meta spend and impressions next to branded search impressions on Google — a year of data.
02
We ran a simple correlation first (0.92), then Granger causality to test direction and lag.
03
The analysis ran as code on a virtual machine inside the chat; the code was kept with the answer.

Run this test on your own account.

Start on EPYC →The growth operating system for your media: plan, run, measure, and improve every campaign from one place. Take it for a spin, on us.The platform Book a call →We run the models, design the experiments, and, if you want, your media too. You get the results. Thirty minutes with someone who does the work, not someone who sells it.The platform, humans included