The analysis of some of these scribbled messages revealed the walls of the wealthy were highly sought after, especially for political candidates hoping to gather votes.
The findings suggest that Pompeii homeowners may have had some control over who got artistic on their walls, said study researcher Eeva-Maria Viitanen, an archaeologist at the University of Helsinki.
"The current view is that any candidate could have chosen any location and have their ad painted on the wall. After looking at the contexts, this would not seem very likely," Viitanen told LiveScience.
"The facades of the private houses and even the streetwalks in front of them were controlled and maintained by the owner of the house, and in that respect, the idea that the wall space could be appropriated by anyone who wanted to do it seems unlikely," she said.
Pompeii, which was famously destroyed and frozen in time by an eruption of Mount Vesuvius in AD 79, was a city of "avid scribblers," she said.
People scratched messages into the city's stucco walls or wrote them in charcoal. They copied literary quotes, wrote greetings to friends and made notes of sums.
More From This Section
Amid all these amateur "wall posts" were political campaign ads, most of which were done by professional painters, Viitanen said.
The study found that campaign ads were almost invariably on heavily trafficked streets, Viitanen told the annual meeting of the Archaeological Institute of America in Seattle.
The second, more surprising, discovery, was that the most popular spots for ads were private houses rather than bars or shops that would see a lot of visitors.
Some 40 per cent of the ads were on prestigious houses, she said, which is notable because there were only a third as many lavish homes as there were bars, shops and more modest residences.
The discovery led researchers to believe that the ads reveal early social networking. It seems likely that candidates would need permission from the homeowner to paint their ads, suggesting the graffiti is something of an endorsement.