Advertisers target consumer computer searches and purchasing activity so that ads can be matched to them. That is why, if you’ve shopped for anything from shoes to kayaking gear recently, images of shoes or kayaks miraculously show up on your computer.

Mining Social Media Accounts

Technology news indicates that searches and buying activity are not the only things being mined on the web. Fast Company recently cited a Canadian company, Cluep, as a firm that actively mines social media accounts for image recognition in an attempt to find products, brands, and activities that will help their clients. 

Example? Under Armour, the sports apparel company. Cluep scores social media accounts across Twitter, Facebook, Instagram, and more to find images of people wearing products from Under Armour's competitor, Nike to match ads to searches. If they are shown wearing any of its gear, they are targeted to see a video of Run Camp, an Under Armour event for serious runners.

Cluep’s product can target not just products, but scenarios. So you don’t have to be wearing Nike; running or on a track will do.

According to company executives, data mining of images like this is part of its business strategy now because sufficient scale is available. 

It also, according to Cluep, is an excellent way to get people involved in an advertisement. The company reports that metrics like conversions and clicks are from five to ten times higher than with more standard methods.

Despite the company’s contention that only recently has scale been available, the mining of images has been going on for some time. Advertising Age profiled a similar company, Ditto Labs, three years ago, noting that it could process 750 million images per day, focusing on selfies throughout social media. 

Ditto Labs also looks for logos, products, and activities, but adds to the mixed sentiment indicators such as smiles.

Privacy and Accuracy Concerns?

Both Fast Company and Advertising Age raise concerns consumers may have about image mining and consequent ad targeting. It’s one thing to be targeted for ads related to a product one has actively looked for. It’s quite another to be targeted based on logos and perceptions of sentiment. Did consumers sign up to be mined when they posted their pictures?

Well, basically, yes. For most social media companies, users sign a standard form that indicates they know their posts can be shared. Unless specifically marked as private, text posts and images can be viewed and downloaded by anyone. Corporations are no exception. In addition, sites marked as private are private: they aren’t accessible to any entity the user hasn’t approved.

How about the accuracy? It might be argued that logos, for example, are not necessarily indicating product purchase. They might just as well be indicating giveaways or the preferences of a consumer’s Aunt Edith. Similarly, a couple holding a baby and dressed in logo-bearing t-shirts may not be smiling broadly because of the logo.

But companies engaged in image scanning and mining are apparently quite content with the metrics they garner. Look for image mining not only to continue but to move into its next frontier: videos.