×
CIO Applications Weekly Brief
Be first to read the latest tech news, Industry Leader's Insights, and CIO interviews of medium and large enterprises exclusively from CIO Applications
Cognitiv: Self-Trained Deep Neural Networks to Transform Marketing
Jeremy Fain, Co-Founder & CEOCognitiv—Ideal Pathway to Identifying and Reaching the Right Customer
As we know, consumer behaviour is very complex. Why someone buys a product, what stores they go to, which products they view or buy, what apps they have on their mobile phones, which websites they visit, what content they read, what time they decide to buy a product, demographic information—these are all important data points that can help illuminate a consumer’s behavior. With the help of such comprehensive data, brands should be able to better understand their consumers but it is just too much data to process with traditional data science teams. At Cognitiv, we have the ability to plug all of this information into our proprietary technology, which then allows us to predict consumer outcomes more accurately than ever before, which in turn leads to increased sales at lower costs. By plugging a marketer’s own Big Data into our deep learning platform Neural Mind, the resulting neural network algorithms are able to accurately predict the outcome of buying impressions for a user at a specific time and place.
When Deep Learning Meets Consumer Data: The First Neural Network Technology for Marketers
Marketers have been experimenting with machine learning for years now, with mixed results. Predicting consumer behaviour is extremely complicated because of the sheer number of data points that are required. Cognitiv’s technology marks the first time that marketers can enter an unlimited number of inputs because of the power of deep learning.
This huge leap forward in machine learning and AI means that our algorithms are able to train themselves by taking raw inputs and finding the patterns that lie within, transforming the data into an artificial super brain that can guess the right answers. Cognitiv’s neural network technology isthe first to be able to train itself to do this for marketers, a milestone in the development of deep learning.
Deep learning itself is in the midst of taking a huge leap forward, and is being applied to solve problems across many different industries. Cognitiv’s aim is to show how deep learning can be applied to marketing, which is especially important at a time when businesses can no longer afford to waste any money on poorly targeted advertising. Instead, marketers need to find solutions that will allow them to target the right people at the right time at scale.
Cognitiv trains computers to play the game of marketing
For instance, a large company like Target wants to figure out how to target their advertising to the people who have either never heard of them or who are dissatisfied with their existing one-stop-shop; otherwise, they’ll end up spending a lot of money advertising to those who already shop there, or to people who have no intention of switching retailers.
To give another example, how can you find the people who are more likely to buy a BMW over a Mercedes? Given that both companies market to roughly the same demographics, it’s not exactly clear how one company might triumph over the other. Finding out what makes someone a BMW loyalist or a proud owner of a Mercedes is complicated, and marketers have thus far been unable to make that leap consistently due to lagging technology—but no longer.
Relieving the Pain in Manual Optimization with Hyper-Optimized Automated Marketing
Marketers spend a lot of money and time online, but the tools their agencies use require them to hire large workforces and optimize campaigns manually. Cognitiv gives marketers the ability to automate that work and optimize their campaigns much more effectively, freeing up marketers to think more creatively about their overall campaign strategies. Today, advertisers want to be more specific with their targeting and get better results, and our technology enables them to do that without adding hard-to-find data science resources or large manual optimization teams.
Moment of Pride: Proof of Cognitiv’s Ability to Drive Customers
Our algorithms can predict any kind of outcome, depending on a marketer’s needs. A company like Uber might want to identify the habits of their most loyal customers, defined as the people who use their service at least five times a week, and use that information to attract new customers. By looking at the list of people who fit this consumer profile, we can create a neural network that determines other consumers who are likely to exhibit similar types of behaviour, and optimize advertising accordingly. Similarly, we can figure out who is going to ask for an auto insurance quote, or how much money someone is going to deposit with a bank or brokerage. All that is measurable or has an existing database, such as a marketer’s customer list, can be used to create algorithms unique to that marketer’s specific need and make it easy for them to find new customers.
Out of the many organizations we have served, one interesting example involves a top mobile phone wireless carrier that wanted to leverage the Cognitiv platform to drive more in-store visits. In order to run that campaign, we used a third-party location data provider to develop a list of the device IDs that had gone to the carrier stores in the last 60 days, as this was the type of person that we wanted to find more of. We then trained our neural networks using that visit data, and advertised to people who the algorithm believed were good prospects (and not existing customers). Not only did we beat our client’s goal by 773 percent, we were also able to drive more in-store visits than any other media partner—and with a cost savings of 50 percent. This is just one example of just how effective our algorithms are at targeting the right people, in the right environment, at just the right time.
Strategic Approach to Reign over the Marketplace
Our engagement with IBM Watson has been an important partnership for us, especially as it validates our technology. IBM Watson has a history of being a strong player in Natural Language Processing (NLP) and image recognition. IBM was looking for a partner that had built deep learning technology with all the links to digital marketing platforms, and we were excited by the opportunity to add our technology to IBM Watson’s portfolio of AI capabilities. That is a huge accomplishment for Cognitiv, which was only two years old at the time of the implementation. As a result of this partnership, our technology continues to advance and be used by more and more marketers each month.
We are are also working on developing other toolsets for marketers that are very specific to mobile application marketing and looking to expand our footprint globally. As it stands, our technology has achieved great results for businesses in several verticals, including financial services, retail, and the automotive industry.
