Siri's inaccuracy may keep it a novelty

GLOBAL - Nearly a year after its launch, Apple's Siri is still far from the accurate and helpful personal assistant that Apple has been advertising, according to a study by investment bank and asset management firm Piper Jaffray.

by Emily Tan | 07/24/2012

New study finds Siri isn't quite as accurate as her advertising implies

The report, cited in Fortune, found that Siri only understood 83 per cent of queries in noisy conditions and 89 per cent in a quiet room. Perhaps worse, Siri only answered accurately 62 per cent of the time in a street and 68 per cent in a quiet room.

By contrast Google understood 100 per cent of the questions and replied accurately 86 per cent of the time. Overall, Siri received an accuracy grade of D and Google a B+.

The study was conducted by a research analyst at Piper Jaffray, Gene Munster, who asked an iPhone 1,600 questions, 800 on the streets of Minneapolis and 800 in a quiet room. He also compared Siri's performance against Google search (Fortune assumed that the search was keyed in and not voice-activated). Examples suggested that even apart from the voice-recognition challenge, Siri performs worse than Google at interpreting the intent of the user's question when wording is ambiguous.

In conclusion, Munster wrote that Siri simply didn't stack up against Google and was therefore not ready to be a viable mobile search alternative.

By Munster's estimation, Siri lags about two years behind Google but expects Siri to "improve meaningfully" with the release of iOS 6.

For advertisers, this could mean a delay in tapping into the potential of voice-activated search, whether via search engine marketing or through voice-activated coupons.

| research