- Adstock explains why sales lag spend.
- Half-life tells you how many weeks until an effect halves.
- High-carryover channels suit pulsed spend.
The idea from pharmacology
A drug dose stays active in the body and fades at a steady rate. Advertising behaves similarly. A video seen on Monday can still influence a purchase the following week.
The geometric adstock formula adds a share of last week's effect to this week's spend. That share is the carryover rate, written θ.
Reading the half-life
Half-life equals ln(0.5) divided by ln(θ). With θ of 0.6, the half-life is about 1.4 weeks. With θ of 0.8, it is about 3.1 weeks.
The longer the half-life, the more you should judge a campaign over several weeks and the less you should react to a single weak week.
Typical ranges by channel
In our planning models for US brands, search ads often carry over at 0.1 to 0.3, social at 0.3 to 0.6 and video or brand campaigns at 0.6 to 0.8. These are starting assumptions to test, not fixed truths.
How it changes your plan
High-carryover channels can run in flights with short gaps, which saves budget. Low-carryover channels need steady presence because the effect ends quickly. Test a gap before you commit to it.
How long does this week's spend keep working?
Questions about Paid Media Math
How do I estimate my own adstock rate?
Fit it in a media-mix model on weekly spend and sales, or compare results across flights and gaps.
Is adstock the same as frequency?
No. Frequency counts exposures. Adstock models how the effect decays over time.
Does adstock apply to email?
Usually the carryover is short, but the same math applies.
Key terms
Keep reading