CES is the year’s first industry mega-event where the tech community unites for product launches, collaboration, networking and thought leadership. This year CES was virtual and we were proud to be included with other amazing companies and speakers who are at the forefront of tech. 

Our Chief Technology Officer, Peter Day, presented on the challenges and benefits of audience insight platforms. He provided background on his own discovery track, trying to define not only what an insight is, but also how it can be harnessed in the real world.

What is an insight?

The goal of insights is to help marketers build marketing intuition so they can do a better job of reaching and influencing people. Key to this thinking, Peter asserted that there are three things an insight must provide. 

An insight must be….

  1. Useful for decision making, otherwise it’s just trivia, right?
  2. Believable, yet non-obvious.
  3. Builds marketing intuition, re-enforces challenges, deepens our current understanding not just of current customers, but potential prospects.

Peter also addressed the challenges of building an insights platform, explaining how as CTO he has built and iterated on Quantcast’s audience planning tools. Three big challenges emerged. 

  1. Aggregate information often hides reality, so you need to have access to all the data

    We’ve accumulated more data in the last two years than we have in the entirety of human history that came before it, so we’re not short of data. But as soon we start aggregating it and looking at things like averages, you tend to hide the reality. And that’s true even on simple things such as the average frequency that an individual has received an ad. You might look at the average and find it’s seven, but that seven often hides the fact that a million people saw the ad once and one person saw the ad a million times. So to find that needle in the haystack, you need access to a lot of data.

  2. Predicting where insight will be found is impossible, so it needs to be interactive

    There might be 45,000 different ways of slicing and dicing data, and one of those will have some sort of insight. But in order to discover this, it needs to be somehow interactive so someone can really drill down and find that “aha” moment that will help them build marketing intuition.

  3. To be useful, the why matters

    Data without theory is about as useful as theory without data. Marketers need to be able to apply their intuition to find patterns in the data — some sort of nugget. And so being able to translate the hypothesis on the why is really what’s essential. To do that, we need tools that marketers can use, and those tools need to be interactive and user friendly.

Insights Platforms

To build the Quantcast insights platform, he used a three-phased approach:

  1. Build a live behavioral graph, which is constantly changing every millisecond. It’s updating as people move around the internet.
  2. Enrich with topic and sentiment information extracted from pages.
  3. And in order to make it interactive, they had to build a queryable through our proprietary database technology from the ground up, so they could get answers back fast enough and allow real-time scale to allow marketers to really drill into data.

Inside the Data Black Box

Peter ended the session with a live demo of Quantcast’s behavior graph to visualize what Quantcast’s footprint on 100MM web destinations really looks like.

Embedding Projector Visualization Of High Dimensional Data

He explained that every little blue dot represents an individual URL and its position is dictated largely by the content on the page.

  1. We crawl the entire internet
  2. We extract the text from the pages
  3. We use the language processing to find the common words and remove the uncommon words
  4. Then it moves every URL into a position in space so that URLs that have similar content and concepts are close to each other, and URLs that have completely different concepts are further apart

This technology allows us to associate content in terms of position in space. In reality, our models are across hundreds of dimensions but our visual cortex can only handle three, so this is projected down to three dimensions. Computers are mathematical beasts, so as soon as they have numbers associated with things, they can make sense to the internet in terms of user behaviors. It is this technology that allows us to build predictive algorithms of who is more likely to buy your product.

Watch the full VOD demo.

We encourage you to watch the full spotlight presentation here, which includes the full live demo of both our Starmap and Audience Planner tools.

About Quantcast:

Quantcast is an audience intelligence and measurement company headquartered in San Francisco. Combining machine learning, a privacy-by-design approach, and live data drawn from more than 100 million online destinations, Quantcast provides software, information, and advertising services for marketers, publishers, and advertising agencies worldwide. Founded in 2006, Quantcast has employees in 20 offices across 10 countries.