Ever since the popularization of the internet, it seems like media and online advertisers have been in an ongoing battle just to keep up with the ever-changing consumer consumption patterns. Looking back, the high speed of innovation and change has often led to opaque reporting and a lack of checks and balances.

A prime example of this can be seen in Facebook’s handling of video advertising on their platform over the past few years. Soon after the release of the new marketing method that would directly inject video advertisements into the social giant’s user’s newsfeeds, Facebook began reporting eye-popping engagement metrics. It all happened so quickly, and media outlets felt as if they had no choice but to quickly adapt or be left behind with no traffic. Writers were fired, entire strategies were constructed, and “pivoting to video” memes were created based upon this reporting.

The only problem? The reporting was false. Or, at least over exaggerated. In late 2016, Facebook admitted that the numbers it had been reporting were artificially inflated by 60-80%.

That’s just one example of the murky waters marketers have to navigate these days. Not to fear though, there are companies working to help bring clarity. Companies like LogoGrab.

Founded by Luca Boschin and Alessandro Prest, who serve as the startup’s CEO and CTO, respectively, LogoGrab is a visual AI company that looks to help marketers quantify the impact of their brand at scale. The two co-founders, and friends for 10+ years, observed the popularization of images on social media as new platforms took off and saw an opportunity. Twitter, Snapchat, and Facebook all were enabling the creation of images and pictures, with the advertising market also moving in that direction. But it wasn’t easy to parse these images, which is where LogoGrab comes in, per Prest.

“Let’s say you’re a company that wants to find out all the tweets where people mentioned your brand in the text. That’s easy to do via the search functionality built into the website. But if you want to do the same for images, there wasn’t an equivalent. So, we set out with this mission of bringing that functionality to the visual dimension, instead of only having it be possible for text.”

At the root of their offering is a specialized machine learning-based system that was purpose built to be able to take an image and then identify it in various videos or pictures within minutes. This quick turnaround is core to the benefit many of the startup’s customers see, as other methods of ML can take much longer and need a much larger data set.

As Prest puts it, “A popular method for training ML models these days include throwing mass amounts of data at it and have the machine figure it out. Process 10,000 pictures of a dog and it’ll get a pretty good idea that they tend to have floppy ears, etc. For our clients, they don’t have time to collect 10,000 images of a brand, and they don’t have the time to wait.”

That speed is why marketing firms, such as Brandwatch and Sprinklr, trust LogoGrab’s tech to help them provide insights to their customers. With features like an instant activation, which allows for customer to input a URL and press a button to start tracking, it’s an easy and quick setup.

For the infrastructure behind the scenes, the startup trusts much of it to AWS, per Prest. “Everything which is mission critical for us is on AWS. We’ve always been very early adopters (even before LogoGrab, we loved the richness of the service offering), and AWS continually introduces new services that we enjoy digging into. Also, the stability and security aspects are hugely important to us. With all that, we can’t help but notice how every other cloud provider seems to just be playing catch up.”

And while the technology and its applications are inspired, perhaps one of the more notable qualities of the startup is how capital efficient they’ve been able to be. LogoGrab has raised less than $3 million since its founding and is already profitable, a milestone that all startups strive for but few are able to achieve. But they’re far from satisfied.

Looking ahead, LogoGrab have moved beyond their beginnings in logo detection and are applying the technology to a variety of other industries. Think identification of counterfeit goods at scale, or protection of brands and rights holders by automatically analyzing images and videos with their products included (they’ve already locked in key customers in these areas, such as ebay and Nagra). With that expansion in mind, the team is set to rebrand the company to represent its future aspiration—2020 will bring more features, and a new name: VISUA.

from AWS Startups Blog: https://aws.amazon.com/blogs/startups/logograb-uses-ml-to-help-marketers-quantify-roi-at-scale/