But media executives say the research-estimating ad spending on a per-household basis-is unlikely to help marketers make spending decisions.
The so-called "advertising atlas" consists of two separate but linked databases.
One measures direct mail, promotion and ad spending by media type in each of the top 150 markets. It also uses national and local spending data from 15 sources, including the U.S. Postal Service, media trade groups and Advertising Age.
Another tracks consumer and business-to-business spending by company using standard industry classifications, annual reports, U.S. Census Bureau and other Commerce Department and local business data. For any given market, the two sets of figures had to agree within a range of 5% before they could be projected against a national average.
The Ad Audit database, as it's called, was developed in 1988 by Kip Cassino, who had been hired to help a Philadelphia radio station calculate its share of the local advertising market. Mr. Cassino, now research manager at the Gannett Co.'s Wilmington, Del., News-Journal, expanded the study and said the research can help companies evaluate the most- and least-expensive ad markets, as measured by the value of ad spending per household.
"What you're looking at is the amount spent to get an ad message to a family in a particular city, how many messages a family in the market receives and how they receive it," Mr. Cassino said.
For instance, while New York, Los Angeles and Chicago are the markets with the most expensive ad rates, and receive by far the most spending, Portland, Maine, gets the highest spending per household, followed by other small markets.
Ranked that way, New York isn't so expensive after all, falling within the national average range of $4,100 to $4,399 spent per household.
Even among larger markets, households in Boston, Dallas, Minneapolis and Orlando are exposed to higher ad spending levels, the survey concludes, while Cleveland, Miami and Pittsburgh get per-household spending that's well below average. So do smaller markets with lower-income and immigrant populations.
Mr. Cassino explained discrepancies-even among similar-size markets-by noting newspaper and other media monopolies often force ad costs higher in certain markets. Income figures and government estimates of local "buying power" also weight national spending estimates and can vary across markets.
Some agency researchers questioned the local data's accuracy, and even Mr. Cassino acknowledges some of his findings are "very fuzzy." Even so, American Demographics Editor Brad Edmondson, who originally published the findings, said the research passed a rigorous methodology test.
Some took issue with the way the database lumps together various forms of media as "looking" (TV and outdoor); "reading" (print and direct mail); and "listening" (radio).
"He's not doing media mixes in a way that anybody at an ad agency does mixes," said Joanne Burke, senior VP-worldwide media research director, at Foote, Cone & Belding, New York.
Erwin Ephron, a partner at Ephron, Papazian & Ephron, said the research is an "interesting exercise, but I don't think it's very important to national advertisers. It's based on the idea that all advertising competes equally for attention, and that misunderstands how ads work. When an ad works, it's because of relevance, not because of the absence of other messages."
Others raised a more crucial point.
"Advertisers spend behind sales, not populations," said Jon Swallen, senior VP-director of media research at Ogilvy & Mather.
He said marketers are more apt to pick a test market based on product usage than media cost. For some new products or emerging categories, however, the data might prove useful, he said.