Brief us

WP-36: Delivery

Eleven billion dollars, thirteen states

The 2024 cycle set a record for political advertising, and two thirds of it landed in thirteen states. Broadcast fell below half for the first time, streaming took a fifth, and the hardest question was still the old one: what did it do?

By Brushworks research deskNovember 2026Working paper

Abstract. Political advertising reached $11.1 billion in the 2023–2024 cycle, by AdImpact’s count, with two thirds spent in thirteen states and more than half in the final eight weeks. Broadcast television fell below half of the total for the first time and connected television rose to 21 percent. We set out where the money went, what the research says about what television and streaming advertising do, and why the growth of addressable media makes the measurement question more urgent rather than less.

In brief

The finding
Political money is more concentrated, later and more addressable than ever. The share that can be measured against a control has grown faster than the share that is.
For a campaign
Buy addressable media from owned accounts and seats, hold out a control, and judge each medium by measured lift per dollar.
For a donor or a committee
Ask what share of your spend landed in markets where every voter was already saturated, and what it moved.

The 2023 to 2024 election cycle was the most expensive on record for political advertising: $11.1 billion, by AdImpact’s count, against roughly $9 billion in 2020 and $8.9 billion in 2022.1, 2 It was also the most concentrated. Thirteen states received $7.4 billion, 67 percent of the total, and 51 percent of all spending ran in the final eight weeks.1 This paper sets out where the money went, what research says it does, and what a campaign should ask of every dollar.

Where the money went

Broadcast television remained the largest medium, at $5.35 billion. For the first time it took less than half of the total: 48 percent. Connected television, streaming delivered to a television set, took 21 percent ($2.33 billion) in its first fully tracked cycle, ahead of digital (15 percent) and cable (12 percent). Radio took 3 percent and satellite 1 percent.1

The 2024 political advertising dollar by medium.

The shift to streaming was fast. CTV’s share of political spending grew from about 12 percent in 2022 to 21 percent in 2024, an increase of more than $1 billion, while broadcast’s share fell by about three points.3

Streaming’s rise in one cycle. Connected television’s share of political advertising, 2022 and 2024 cycles.
Total political advertising by cycle, billions of dollars.

Thirteen states

Two thirds of the money went to thirteen states. Pennsylvania alone received more than $1 billion, over 12 percent of all political spending in the country, and its broadcast viewers saw more than 500,000 political ad airings in the year.1, 4 The presidential race accounted for more than $3 billion.5

Where the 2024 political advertising dollar landed. Each square is one percent of the total.

Concentration has a cost that does not appear in a media plan. A voter in a saturated market has already seen each argument many times. The marginal ad competes with the opponent’s ad in the same break and with the voter’s own fatigue. The research on television advertising finds real effects that are small on average, larger down the ballot than at the top, and short-lived.6, 7 At the margin of a saturated market, the effect of the next spot is likely smaller still.

Addressable media and the measurement question

Streaming and digital media can be bought against lists of households and measured against households that were held out. That makes them the most measurable media a campaign has ever bought. It does not make them measured. Much programmatic political spending passes through layers of resellers, each with its own reporting, so that the buyer cannot say exactly which household saw which ad (WP-12).

What a buy needs before its effect can be measured. The audit trail a campaign should require from every addressable placement.

Three practices close the gap. Buy from the campaign’s own accounts and seats, so that every impression has one origin. Seal a holdout of households before the first flight, so that the comparison is randomized rather than assumed. And judge each medium by measured lift per dollar, not by reach or by cost per thousand.

How the money moved in one decade

Political advertising has grown in each presidential cycle, but its composition has changed faster than its size. AdImpact’s projections for 2024 expected broadcast to hold just over half of the total and cable to slip slightly. The actual cycle moved further: broadcast fell to 48 percent, and streaming took a fifth of the money after taking about an eighth two years earlier.1, 3

Two forces drove the shift. Viewers moved: a growing share of American households watch television through streaming services, and many of the voters a campaign most wants to reach, younger and more mobile, watch little live broadcast at all. And buyers moved: streaming inventory can be bought against a list of households, which lets a campaign reach the people in its plan and skip the people outside it, at a higher price per impression and a lower price per relevant one.

Where the growth went. Change in spending by medium between the 2022 and 2024 cycles, billions of dollars, from AdImpact’s published figures.

The concentration of spending moved the other way. Two thirds of 2024’s money went to thirteen states, and a single state took more than a billion dollars.1, 4 For a voter in one of the thirty-seven states outside that group, the presidential campaign was a rumor. For a voter in Pennsylvania, it was a wall.

These changes set the questions a campaign should ask of its own plan. Is the money going where the undecided are, or where the race is close on the map? Is the addressable media actually aimed at the households in the plan, and is any of it held out? And how much of the spend in a saturated market is buying the second hundred airings for people who made up their minds after the first ten?

What saturation looks like

In a contested market in the final weeks, a voter who watches local television sees political ads in nearly every break. Pennsylvania’s broadcast viewers saw more than half a million political airings in 2024.4 Each additional airing reaches someone who has already seen the argument, often many times, and who is now hearing it next to the opponent’s answer.

Measured lift per additional million dollars, by cumulative spend in one market. Illustrative of diminishing returns; no single program’s data.

The shape of diminishing returns is not controversial. What is rare is a campaign that measures where it is on the curve. Without a control, the last million in a saturated market looks exactly as effective as the first million did, because the poll is moving for every reason at once.

A worked example: the last two million dollars

The figures in this example are illustrative. A campaign has $2 million left for the final three weeks in a market where it has already bought heavily. Option one buys more broadcast. Option two buys addressable media aimed at households the campaign has not reached, with a sealed holdout, and a turnout program for supporters who have not yet voted.

Modeled net votes from the last $2 million under two plans. Illustrative.

The point of the example is not that one medium beats another. It is that the second plan produces a measurement, so that the campaign knows what its last dollars did and the next campaign does not have to guess.

Objections

Broadcast reaches the older voters who decide elections. It does, especially in local and off-cycle elections (WP-35). The question is how many times each of them needs to see the argument, and what the next airing adds.

Streaming targets households, not people. For most persuasion and turnout work, household attribution is enough; where it is not, the plan should say so (WP-24).

Measurement takes too long in the final weeks. A holdout drawn before the first flight reads within days. A campaign that waits until October to start measuring has chosen not to know.

Questions to ask of a media buy

  • What share of the budget lands in markets where every likely voter has already seen the argument?
  • Which placements are bought from the campaign’s own accounts and seats, and which through resellers?
  • Where is the holdout, and when was it drawn?
  • What is the measured lift per dollar by medium, and when will it be reported?

Further reading

  • Sides, J., Vavreck, L., and Warshaw, C. (2022), cited above: the broad estimate of what television advertising does in American elections.
  • Gerber, A. S., Gimpel, J. G., Green, D. P., and Shaw, D. R. (2011), cited above: how quickly the effect fades.
  • Kohavi, R., Tang, D., and Xu, Y. (2020). Trustworthy Online Controlled Experiments. Cambridge University Press: how organizations that run many experiments learn to trust their measurements.

Implications

For a campaign. Treat each medium as a hypothesis. Measure its lift against a control in the first flights, and move the money to what works before the expensive weeks.

For a donor or a committee. Ask how much of the budget lands in markets that are already saturated, and what measured evidence says the last million there will do.

For both. The growth of addressable media is an opportunity to know what political advertising does. It is only realized by buyers who insist on it.

Notes and references

  1. AdImpact, Cycle in Review 2023–2024, December 20, 2024; summarized by TV Tech.
  2. AdImpact cycle totals for 2020 and 2022 as reported by TV Tech and in AdImpact’s 2024 projections report (2020 cycle, $9.02 billion).
  3. Locality, citing AdImpact on CTV’s share in the 2022 and 2024 cycles.
  4. AdImpact, “Pennsylvania saw a record-breaking $1.2B in election ads.”
  5. AdImpact figures on presidential spending, as reported by NBC News, November 2024.
  6. Sides, J., Vavreck, L., and Warshaw, C. (2022). The effect of television advertising in United States elections. American Political Science Review, 116(2), 702–718.
  7. Gerber, A. S., Gimpel, J. G., Green, D. P., and Shaw, D. R. (2011). How large and long-lasting are the persuasive effects of televised campaign ads? Results from a randomized field experiment. American Political Science Review, 105(1), 135–150.

Basis of figures. Spending by medium, state and period from AdImpact’s published cycle review; CTV’s 2022 share as reported from AdImpact. Totals for earlier cycles differ slightly by source and are rounded. The returns curve and the last-two-million example are illustrative. No Brushworks program figures appear in this paper.

Next paperThe Review

Start a conversation

Brief us.

Tell us the race, the calendar and the margin. Within a week we will tell you whether we can move it and what it would take. Candidate work and independent-expenditure work are taken by separate teams; say which you are. Governments and institutions: tell us the mandate and the jurisdiction, and we will tell you what registration it requires before we say anything else.

Send a briefbrief@brwrks.com

  • Candidate practice
  • Independent-expenditure practice
  • International practice

Brushworks Strategies Inc., a Wyoming corporation · Washington, New York, Ottawa, London, Munich, Dubai, Tokyo