What Obama's 2008 Burnout Paradise Ads Actually Bought

In October 2008 a set of screenshots began circulating on gaming forums. They showed billboards in Burnout Paradise, the open world racing game, carrying a blue Obama campaign banner and the line "Early voting has begun" with the address VoteForChange.com. Some players thought it was a mod or a joke. It was a paid placement, and the Obama campaign became the first presidential campaign to buy in-game advertising on a console network.

I have the screenshots in my archive alongside the TV spots from the same month, and for years I filed them under curiosity. Looking at them again, I think that undersells what happened. The buy was small, but it tested three things that political advertisers still argue about: geographic targeting inside entertainment, reaching people who avoid traditional ads, and how much of an ad's value comes from the coverage about it rather than the ad itself.

What the campaign actually bought

According to reporting at the time, the ads ran from early October until the day before the election, across 18 Xbox 360 titles. Electronic Arts handled placements in its own games, and Microsoft's in-game ad unit Massive Inc. served the rest. Besides Burnout Paradise, the list included EA sports titles like Madden NFL 09 and NBA Live 08, plus racing and music games. The spots were limited to around ten battleground states, including Ohio, Florida, Iowa and Colorado. That restriction worked because the Xbox 360 was online, so ad servers could estimate a player's location from their connection and decide which billboard texture to load.

The campaign never published the cost. Industry estimates floated in the press suggested a modest figure, the kind of money that would barely buy a week of local TV in a single mid-sized market. I would treat any precise number from that period with caution.

The creative was almost identical to the campaign's web banners: a photo of Obama, the early voting message, the website. Nothing was designed around the game itself. In Burnout Paradise the ad simply occupied the billboards that also carried ads for real products, which the game had sold since launch.

Why the reach numbers mattered less than you would think

If you measure the buy the way a TV planner would, by impressions delivered, it was a footnote. Even within the swing states, the share of voters playing those 18 games in that month was small.

The audience, though, was specific. Console players of racing and sports games skewed young and male, a group that watches less live television and was known to be hard to reach through the networks. The campaign was not trying to persuade this audience so much as remind it that early voting existed. A billboard flashing past at 180 miles an hour is a poor vehicle for argument and a reasonable one for a date and a URL.

The bigger return came from coverage. Within days, the placement was written up by gaming blogs, the tech press and then mainstream outlets. Each story repeated the message, the URL and the idea that Obama's campaign understood young voters' media habits. That last part was arguably the real product. The campaign was buying a signal about itself, and paying for it at console ad rates rather than prime-time rates.

The audience was never the point. The press release was the point, and the billboards were the proof that the press release was true.

I want to be careful with that claim. I do not have the campaign's internal reasoning, and people who worked on the digital side have described the buy in different ways. But the pattern of value, small spend and outsized earned media, is visible in the clipping file.

The mistake later campaigns made copying it

The lesson most political marketers drew from 2008 was "advertise where young people play." It sounds right, and it led to a decade of rather awkward attempts to place candidates inside games and virtual worlds. We wrote about the earlier version of that hope back in 2007 in our post on Second Life campaign offices, and most of those offices ended up empty.

I think the more accurate lesson is narrower. The 2008 buy worked because it was first. The novelty generated the coverage, and the coverage carried the message. A second campaign doing the same thing four years later would get a fraction of the attention, and the underlying impressions were never large enough to justify the buy without that attention.

The other overlooked factor is fit. Burnout Paradise already had billboards for real brands, so a political ad in that space read as part of the world. Placing a candidate into a fantasy game with no existing ad surfaces would have felt like an intrusion, and players do notice. The forum threads I archived from that October are mostly amused rather than angry, and a big reason, I suspect, is that the billboards were exactly where players expected ads to sit.

How it compares with later in-game political ads

A quick side by side helps show what changed after 2008.

Programmatic ad networks now serve ads inside mobile games at enormous scale, and political advertisers use them routinely. Those buys are cheaper, far more targeted, and almost never covered by the press, which is the reverse of the 2008 situation. They deliver impressions without the earned media. If you want the targeting side of that story, our explainer on how microtargeted political ads find you covers the mechanics.

Disclosure has also become a live issue. A billboard in Burnout Paradise carried a clear campaign identity. A small banner inside a free mobile puzzle game may not carry the same clarity, and the rules about disclaimers on small digital ads have been argued over for years. For the wider history of how ads have moved online, the political ads section keeps a running thread.

One thing to look at this week

Pick one paid placement from a current campaign, a game ad, a streamer sponsorship or a podcast spot, and search for news coverage of it. If you find more articles about the placement than you would expect from its likely reach, you are looking at a 2008-style buy, one that paid for the story rather than the screen. Note which it was before you judge whether it worked.