Co.Create

Ditto Labs Is Looking To Mine Brand Insights From Your Shared Photos

Ditto Labs launches a new analytics tool with the potential to scour some 800 million selfies, shoe shots, and food porn to help marketers home in on how people really feel about their products.

800 million. That’s the estimated number of photos that get shared publicly across Facebook, Snapchat, Instagram, and Flickr. Whatsapp apparently adds 600 million more to that staggering total. Every day.

Despite the proliferation of selfies, shoe shots, and food porn, there isn't a lot of opportunity for marketers to parse all that visual information.

David Rose

David Rose believes he’s found a way to change that. The CEO of Boston-based Ditto Labs tells Co.Create that the right analytics tool has the potential to tap the promise of those photos and turn them into a branding opportunity.

A long time photographer and self-professed "rabid photo sharer," Rose says that he spied an opportunity in uploads back in the nascent days of the web. Back then, he and Neil Mayle (Ditto Labs’ current CTO) wrote several patents for online photo sharing, eventually launching Opholio, a pioneer in the space.

"We were way too early with that," Rose admits, partly because the price for digital cameras was still prohibitively high. The two eventually sold the company to former Apple subsidiary Flashpoint in 2000.

Now, says Rose, the timing is just right. Instead of focusing on the technology of uploading, Ditto Labs is homing in on the photos themselves with a logo detection engine that reveals trends, measures sentiment, and monitors new products. Though software to analyze text abounds from the likes of Salesforce and Oracle, Rose estimates that current analytics tools are missing 85$ of the millions of public photos shared daily.

The system, which runs in Amazon’s cloud and has between 200 to 300 servers processing "a firehose of images" from Facebook, Instagram, tweets, and Tumblr gets funneled into huge queue that continuously tries to look into the pixels to identify about a thousand brands from sports to consumer package goods.

KFC Brand associated Photos that were used amongst social media pages

The system is great at finding brands, he says, because they are designed to be identified. "We do machine learning to improve the precision," he says, "to get a true positive," that is, to make sure that swoosh is really a Nike logo and not some random shadow on fabric.

In addition to pinpointing logos, Ditto Labs’ system has also been taught to recognize patterns like a Burberry plaid or a Vera Bradley paisley. It then runs those images up against the people in the photos.

Face of the customer

"Humans automatically identify faces and smiles, key transmitters of sentiment," he explains, "Ditto's image recognition technology analyzes faces and smiles in pictures and captures sentiment as a result." Faces can be measured to see if they are taken as a selfie and if the brand is intentionally featured by the subject and not just photobombing in the background. Machine learning doesn’t end there.

Ditto Labs can capture people in different contexts: in groups, indoor and outdoor, in the presence of pets, all of which leads to a better understanding of "how people are using these products in the wild," says Rose. There is a human component, too. Rose says Ditto’s brand strategists who are experienced in photo analysis and insights can deliver in-depth sentiment findings to their clients.

Photo Impressions

Rose believes this can be a boon for marketers looking to leverage the somewhat elusive power of social selling. Consider these figures according to Econsultancy Digital Intelligence Briefing:

  • 25% of search results for the world’s top 20 brands are links to user-generated content.
  • 35% of bloggers post opinions about products and brands.
  • 78% of consumer trust peer recommendations. Only 14% trust advertisements.

The inspiration offered by an average person using a brand is a powerful motivator for someone else to do or buy, the same thing. "That’s how we came up with the name for the company," says Rose.

So far during its beta test, Ditto Labs analytics have offered clients such as Digitas a way to mine inspiration for a Cadillac campaign. Think: targeting ethnically diverse women instead of old, white guys. "We looked for photos that included brand and then we looked at how [users] were talking about it. That photo-ethnography gave Digitas an idea for the campaign."

Ditto Labs recently partnered with Fox Sports for a fan engagement program for the Super Bowl and are gearing up for another campaign with Fox Sports around March Madness. With thousands of images already analyzed, Rose says its simple to turn on analytics for any brand that wants to get the intelligence for a monthly fee.

As for privacy concerns, programs such as the ones using Ditto Labs’ Facebook app clearly need to be opted into. He says the analytics tools don’t collect or save any personal data, just the brand/logo and sentiments generated.

Likewise, as brands start to use this as a tool to for e-commerce, users will have to opt in. These are still early days for photo-enabled shopping, says Rose, as there isn’t a seamless way to go from Instagram to a shopping site, for example.

For now, Rose says, "Any brand we talk to wants to grow their audience and they see social photo sharing as an organic way to do that." With an estimated 20,000 to 30,000 brands in the U.S. alone, Rose says one of Ditto Labs challenges is that they can grow in so many dimensions. "We are on the lookout for a 'something else completely' application like a public health app, removing logos if people wish, brand terrorism, and a global brand trending index." Looks like its an idea whose time has finally come.

[Image: Flickr user Shuji Moriwaki]

Add New Comment

0 Comments