What Is Adstock? How Carryover Effects Shape Your Marketing Results

Adstock is the lingering effect of advertising on consumer behavior after the initial exposure. When someone sees your TV commercial on Monday, they don't forget about it by Tuesday. That awareness sticks around, decaying gradually over days or weeks. Adstock is both the concept that describes this carryover and the mathematical framework used to model it.

The term was coined by British researcher Simon Broadbent in his 1979 paper "One Way TV Advertisements Work," published in the Journal of the Market Research Society. Broadbent observed that television advertising didn't produce instant, one-time effects. Instead, each exposure built on previous ones, and the accumulated impact decayed slowly over time.

This matters for marketers because if you ignore adstock, you undercount the value of channels with long carryover (like TV) and overcount channels with short carryover (like paid search). Your budget ends up in the wrong places.

How Adstock Works

Think of adstock like a bathtub. Each ad exposure adds water. Between exposures, the water slowly drains. The drain speed is the decay rate. If you keep running ads, the water level stays high. If you stop, it gradually drops to zero.

More precisely, adstock at any point in time equals the current week's advertising weight plus a fraction of last week's adstock. That fraction is the decay parameter, typically represented as a number between 0 and 1.

Here is a simplified example. Say you run a TV campaign that delivers 100 gross rating points (GRPs) in Week 1, then stop. With a decay rate of 0.7 (meaning 70% of the effect carries into the next week):

Week	New Ad Weight	Carried Over	Total Adstock
1	100	0	100
2	0	70	70
3	0	49	49
4	0	34	34
5	0	24	24

Even though you stopped advertising after Week 1, the effect persists for weeks. By Week 5, about a quarter of the original impact remains. This is why brands that go dark on advertising don't see sales drop off a cliff immediately. The accumulated adstock provides a buffer.

Now consider what happens when you advertise continuously. Each week's new exposure stacks on top of the residual effect from previous weeks. The adstock builds to a higher steady state than any single week's effort would produce on its own. This compounding is why consistent advertising tends to outperform sporadic bursts of the same total spend.

Why Adstock Varies by Channel

Not all advertising decays at the same rate. A TV commercial with strong creative and emotional resonance sticks in memory longer than a banner ad you scrolled past. The channel, the format, and the creative all influence how quickly the effect fades.

Research from media mix models across industries shows consistent patterns:

Slow decay (high adstock) channels:

• Television: Average adstock rates around 60-65%, meaning roughly two-thirds of TV's impact carries into the following week. TV benefits from multi-sensory engagement (sight, sound, motion) and typically higher attention levels during viewing.
• Out-of-home (OOH): Billboards and transit ads show adstock rates in the 50-55% range. Repeated physical exposure along daily commute routes reinforces the message.
• Print: Magazine and newspaper ads tend to have moderate-to-high carryover, partly because readers may revisit the publication multiple times.

Fast decay (low adstock) channels:

• Paid search: Near-zero carryover. Someone clicks your Google ad, converts or doesn't, and the effect is essentially over. Paid search is an intent-capture channel, not a memory-building one.
• Social media ads: Adstock rates typically fall in the 10-20% range. The feed moves fast, attention is fragmented, and individual impressions are quickly forgotten.
• Display/programmatic: Similar to social, with effects that peak within 1-2 weeks and become negligible after 3-4 weeks.

These differences have direct implications for how you allocate your marketing budget. If you evaluate TV and paid search using the same one-week measurement window, you will systematically undervalue TV (which is still producing results in weeks 2, 3, and 4) and overvalue search (which front-loads all its impact into the conversion week).

Adstock in Media Mix Modeling

Adstock is a core component of media mix modeling. When an MMM estimates the relationship between your ad spend and your business outcomes, it needs to account for the fact that this week's sales are influenced by both this week's advertising and the residual effects of past advertising.

Without adstock transformations, an MMM would treat each week's advertising as independent, ignoring carryover entirely. That produces biased estimates: channels with long carryover look weaker than they actually are, because the model attributes their delayed effects to other factors (or to the baseline).

How MMM Models Adstock

There are two main mathematical approaches to modeling adstock, as outlined in Google's influential research on Bayesian media mix modeling:

Geometric decay (Broadbent's original model):

This is the simplest and most common approach. The adstock at time t equals the current advertising plus a fixed percentage of the previous period's adstock:

Adstock(t) = Advertising(t) + decay_rate x Adstock(t-1)

The decay rate is a single number between 0 and 1. Higher values mean slower decay. A decay rate of 0.8 means 80% of the effect carries forward each period. This model assumes the peak effect happens immediately upon exposure and then decays from there.

Delayed adstock (Weibull decay):

Some channels don't peak immediately. A podcast ad might take a few days to sink in. A TV branding campaign might build awareness gradually before influencing purchase behavior. The Weibull function allows the response to build to a peak at some point after the initial exposure, then decay from there. It uses two parameters (shape and scale) instead of one, giving it more flexibility.

The choice between geometric and Weibull decay depends on the channel. Performance-oriented channels like paid search and social typically follow geometric decay: the impact is immediate and fades quickly. Brand-building channels like TV and podcasts often show a delayed peak that Weibull handles better.

The Relationship Between Adstock and Diminishing Returns

Adstock and diminishing returns are separate concepts, but they work together in an MMM. Adstock handles the time dimension (how effects carry over across weeks). Diminishing returns, modeled through saturation curves, handle the volume dimension (how each additional dollar of spend produces less incremental impact).

Consider a Facebook campaign:

• Adstock tells you: The effect of this week's spend persists into next week at roughly 15% strength.
• Saturation tells you: Going from $5,000 to $10,000 in weekly spend does not double the impact, because you start reaching less responsive audiences.

Both effects matter for calculating marginal ROAS. If you only model saturation without adstock, you miss the carryover value. If you only model adstock without saturation, you overestimate what happens when you scale spend. A properly specified MMM captures both.

This is also why incrementality testing results sometimes differ from MMM estimates. A short geo-lift test might not run long enough to capture the full carryover effect of a channel, especially for high-adstock channels like TV. Using both methods together, with incrementality tests calibrating the MMM, gives you the most accurate picture of marketing effectiveness.

Why Getting Adstock Right Matters for Budget Decisions

Mismeasuring adstock leads directly to misallocated budgets. Here are the most common ways it goes wrong:

Undervaluing brand channels. If your measurement window is too short, channels with long carryover (TV, OOH, podcasts) look like they underperform. The true return on these channels extends well beyond the campaign flight dates, but a last-click or same-week analysis misses most of it. Teams that rely solely on platform-reported attribution for budget decisions tend to over-index on short-carryover digital channels as a result.

Cutting spend too aggressively. Because adstock creates a buffer, the negative effects of pausing a channel don't show up right away. A marketing team cuts TV spend, sees no immediate drop in sales (thanks to residual adstock), and concludes TV wasn't working. Three months later, when the adstock has fully decayed, sales start declining and nobody connects it back to the TV cut.

Misreading test results. When you run an incrementality test on a high-adstock channel, the measurement period needs to be long enough for carryover effects to play out. A two-week holdout test on TV will understate the true incremental impact because the control group still has residual adstock from pre-test advertising. The Marketing Science Institute has highlighted this as a common source of error in marketing experiments.

Ignoring the compounding benefit of consistency. Sporadic advertising never builds steady-state adstock. Two $50,000 flights separated by two months of silence produce less total impact than $25,000 per month for four months, even though the total spend is identical. The continuous schedule maintains higher average adstock levels throughout the period.

How to Think About Adstock for Your Business

You don't need to calculate adstock by hand. A well-built media mix model estimates adstock parameters from your own data, giving you channel-specific decay rates based on your actual advertising and sales history.

But understanding the concept changes how you think about a few key decisions:

When to measure. If you are evaluating a channel that you suspect has long carryover, make sure your analysis window is long enough to capture it. Looking at same-week ROAS for a TV campaign is almost guaranteed to undercount its value.

How to schedule. For high-adstock channels, maintaining consistent spend (even at a lower weekly level) typically outperforms the same budget concentrated into short bursts. The compounding effect of sustained adstock more than offsets the reduced weekly weight.

What to trust. Platform-reported marketing KPIs like ROAS and CAC don't account for carryover at all. They credit all the value to the moment of conversion and ignore the weeks of accumulated awareness that preceded it. This is one of the core reasons data-driven marketing requires modeling beyond what the ad platforms provide.

How Formula Handles Adstock

Formula's media mix modeling platform estimates channel-specific adstock parameters directly from your spend and revenue data. Instead of assuming a fixed decay rate for each channel, the model learns the actual carryover patterns from your business. That means your TV adstock estimate reflects your creative, your audience, and your competitive environment, not an industry average.

These adstock estimates feed directly into the response curves and marginal ROAS calculations that drive budget optimization, so when you shift dollars between channels, the recommendation accounts for both the immediate and the lingering effects of your advertising.

See how Formula measures the true carryover value of your marketing channels.