What is adstock?

Advertising does not stop working the moment a campaign ends. Someone sees a TV spot on Tuesday and orders two weekends later; a radio campaign keeps a brand in mind for a few weeks. Adstock is the name for that carryover: the share of a week's advertising effect that lands in later weeks, and the rate at which it decays.

Every channel has its own carryover. Paid search converts almost immediately and carries little. TV, radio and out-of-home spread their effect over several weeks. The model estimates both the amount and the decay for each channel from the data, starting from a prior.

In the model, adstock transforms the spend series before it meets the response curve. A geometric adstock keeps a fixed share of last week's effect each week, so the effect fades steadily; other shapes let the effect peak a week or two after the spend. The practical result is that a week's sales are explained by this week's spend plus the fading tail of earlier weeks, channel by channel.

Carryover is also how TV's pull-through shows up. When heavy TV lifts search and social sales in the following weeks, that lift is credited to TV, because the model sees TV spend followed by higher sales. The model does not estimate interactions between channels; natural variation in the history is what lets it separate channels that often run at the same time.

Without carryover, a model credits the wrong week. A TV flight in week 1 followed by a search push in week 2 would look like search drove the week-2 sales. With carryover estimated per channel, both get their share, and the medium-term effect of TV and radio is measured rather than lost.

Adstock covers the weeks after a campaign, not the years. The long-term brand effect, the slow build of baseline demand from sustained brand investment, is a separate question and is covered under long-term brand effect.

• Long-term brand effect: The slow lift in baseline demand from sustained brand investment, beyond the weeks adstock covers.
• Saturation curve: The response curve per channel: how sales respond to spend, and where the next unit stops paying.
• Channel contribution: What each channel added to sales above the baseline over a period. The decomposition behind every ROAS figure.
• Marketing mix modeling (MMM): The statistical method behind everything else on this list: spend in, sales out, every channel on equal terms.

Go one level deeper: Measuring TV, connected TV and YouTube with MMM. How the model works. Or go back to all terms in the glossary.