Twitter Advertising Becoming Much More Aggressive

Twitter has announced it will run personalised advertisements based on what an individual user is writing about. It's an attractive proposition for advertisers but may come across as creepy to users of the social media service.

Most Twitter ads are based on the people a user follows and the types of messages that users retweet (or share with other users). (Source; pcworld.com)

The new advertising model will go beyond that and incorporate technologies from two popular ad-funded web giants. Like Facebook, Twitter ads can now be targeted to people of a particular gender, living in a particular location, or using a particular device.

One big limitation here is that, by comparison to Facebook users, Twitter users tend to provide far less personal detail when setting up an account.

Advertisers Pay Based On What You Say

Twitter will also begin keyword targeting, meaning advertisers can have messages appear when a user includes a particular word or phrase when posting a message.

Advertisers can combine this with geographic targeting based on the location from which a user makes a post. Twitter gives the example of somebody posting a message about a band releasing a new album and then seeing an ad for tickets when that band performs at a local venue.

According to Twitter, the changes won't make any difference to the number of ads users see or the way in which they are displayed. It says that the change should mean ads are less annoying because they are more relevant. (Source: twitter.com)

Twitter Ad Targeting May Be Too Literal

Will the changes actually draw more clicks? The answer will depend on how intelligently the keyword matching works. The key could be whether Twitter can use a complex technology called 'sentiment analysis', which tries to carry out the tricky task of judging the tone and intention of the language a human uses.

Without sentiment analysis, it's possible a tweet reading "I've had it up to here with Justin Bieber," would trigger an ad for the singer's latest album. That would be counterproductive, to say the least.

Sentiment analysis can also struggle to identify irony or sarcasm, creating the risk that somebody tweeting "Oh, I'd just love to hear more about Justin Bieber, I don't see enough of him on TV," might get an ad that reflects exactly what they say and not what they mean.