A software bug has resulted in TV ratings being allocated incorrectly among broadcast and syndication programming since March, according to executives at Nielsen, the measurement service whose rankings form the basis of how advertisers pay for TV commercials.

At issue is a process used by the company when it moves from its initial survey of national ratings to its final one. When Nielsen calculates its early “fast national” ratings, some sources of viewership are not properly labeled and the data is put aside as “all other tuning” until it can be identified and tabulated properly. But Nielsen has discovered that some of that early unidentified data was improperly attributed to ABC programming, explained David Poltrack, chief research officer for CBS Corp., who said CBS contacted Nielsen about the issue three weeks ago and partly spurred the company’s investigation.

Because of the error in its systems, “we started to credit that viewing to different programs,” said Patricia McDonough, Nielsen’s senior VP of planning policy and analysis, in a Friday conference call with reporters. “Some of that was done to the wrong source.”

Executives said 98% to 99% of ratings would not be affected by more than .05 of a ratings point. Cable networks and local programming are unaffected by the snafu.

The issue came to light, Poltrack said, when CBS noticed ratings for a broadcast of ABC’s “Dancing With the Stars” were increasing between the “fast national” tabulation and the final one, even though the show was preempted in two markets so that local stations could show home-team NFL football games.  What CBS found, Poltrack said, was that all of the unidentified views in the “fast national” data were “going to ABC. None of it was going to the proper network. It was all going to ABC. That resulted in ABC getting a big bump.”

In a statement, ABC said, “Our entire industry relies upon Nielsen for accuracy and veracity, and we hope that they can quickly resolve this issue.   We’re confident that the momentum we’ve seen across the network so far this season will continue, including delivery of the #1 new drama and the #1 new comedy on television.” Nielsen executives declined to specify which networks were affected most.

Nielsen said it uncovered a technical error on Oct. 6 that revealed incorrect measurement of national network TV ratings over several months, resulting in the incorrect attribution of “small amounts of viewing.” The error was introduced on March 2, the company said. The glitch affected both primetime and total day ratings, executives said.

Executives from at least two broadcast networks expect to be awarded some degree of viewership they might otherwise have lost. “The numbers are not very large for us,” Poltrack said. But the data could “make a difference between [being] first or second for the week, or being up versus last year, or being flat versus last year,” he said.

The measurement hiccup comes as Nielsen is under pressure to measure all the new ways consumers watch video content that was once solely distributed by television. Nielsen has made strides to include viewing of streaming video available on desktops and laptops, smartphones and mobile tablets. But TV networks, who have seen their traditional ratings decline as would-be TV viewers splinter off to see video through this new panoply of methods, want the company to move more quickly to draw all views into a single tabulation that would show advertisers TV draws the mass of viewers they say they covet from the medium.

Glitches in data collection are not uncommon, said Poltrack, but TV networks want to be more certain that Nielsen is monitoring its efforts. “We’d like to know what operational thing you’re going to put in place so that you discover it, as opposed to us.”