Predicting CTR and spotting bots in online advertising

In internet advertising, total measurability and trackability are both a blessing and a curse, especially when a company is paying for ads in a cost-per-click model.

The challenge

The cost-per-click model provides countless small businesses around the world an inexpensive way to advertise online. They pay only for ad clicks. For the publisher displaying ads, unfortunately, there is a strong temptation to click the ads and get paid for it. That’s fraud, and it poses a challenge.

Of course, a low ad click-through rate remains a problem for both the site displaying the ads and the businesses that have taken them out. The publisher is displaying ads that don’t make a profit.

The solution

deepsense.ai designed a model that predicts the click-through rate of internet advertisements. The model analyzes historical data on site user behavior to spot patterns and uncover anomalies. It enabled the deepsense.ai team to identify an abnormal pattern among users, which turned out to be bots engaging in fraudulent clicking.

The effect

The solution effectively spotted internet bots that click ads, significantly boosting CTR predictions. The solution spotted up to 90% of bots and improved CTR predictions by up to 35% over existing heuristics.