How 26 Top Marketing Executives Use Artificial Intelligence

With Salesforce Einstein, which brings the power of AI to every Salesforce user, marketers can predict the optimal timing, channel, content and audience for any marketing message, as opposed to only being able to look at past consumer behavior. Specifically, marketers can gauge how likely it is a customer will engage with an email, unsubscribe from an email list, or make a web purchase, and determine what is driving true engagement to better anticipate the needs of every customer. They can also build audience segments of people showing multiple predicted behaviors in common.

We’ve brought A.I. into our own practices, especially in predictive account scoring, or figuring out what companies are most likely to purchase products, as well as analyzing which of our current customers are most likely to buy additional services. These days, 60% of the buying process is done before even meeting with the prospect. These decision makers are engaging across a variety of different channels, hitting multiple advertisements or events that influence their decision. Should they decide to buy the product or services, we have to determine what method influenced their final decision. Multi-touch attribution is key here, along with associating cost.

In the next four years, customers will expect 85% of their communications to be self-service, so A.I. will be absolutely necessary to bridge these customer expectations with the ability to provide good engagement and service. To do this A.I. must better predict what customers will want based on past behavior while also appearing to be a facsimile of a human being. We are evolving our own data business at Bitly, and hope that cognitive systems will enable Bitly customers to integrate our massive first party dataset with other sources to create the ultimate customer experience.

I’ve found that having a single view running through all of the commercial and marketing operations, including all of the systems being used, with a common language around the data is what makes marketing most efficient. Something simple like a different interpretation of a lead source or success metric for customer engagement can drive confusion and inefficiency. Finally, data governance is certainly the biggest issue with any technology – with multiple stakeholders using these solutions, having that common language around the data sets for input and output will dramatically increase efficiency and ability to calculate an ROI with solid data.

Brands aren’t capitalizing on the vast amount of content related to their brand that already exists and is being created by their own customers. There are 1.8 billion photos posted each day to Instagram alone. Discovering, analyzing and recommending the right content for the right customer at scale is a challenge that only machine learning technology can solve. Stackla’s tech helps brands do just that — discover the best user-generated content (UGC) around their brand, categorize it around customer personas, and recommend the right content for the right marketing channel. Virgin Holidays increased bookings by 260% over the previous year by using UGC from Stackla’s machine learning content marketing platform. During London Fashion Week, TOPSHOP increased sales of featured online products by 75% with Stackla.

We utilize CRM, marketing automation, and customer advocacy systems, and also our own Reltio Cloud to bring all customer data together in a data-driven application to maintain a complete understanding and a single-source-of-truth of our customers and accounts so we can deliver personalized and relevant information to them in a timely fashion. Our technology is a key tool used in our marketing efforts, and central to the department’s success. Leveraging our own technology, the marketing team was able to lead a personalized video campaign that saw engagement rates exceeding 1000% our company’s average.

AI makes media buying much more efficient. First, it automates the tedious process of campaign setup, which eliminates human error and frees up media buyers to focus more on strategy. Second, it spots nuanced patterns undetectable to the human eye. Third, it breaks ad campaigns into several micro-campaigns for multivariate testing and shifts ad dollars to the best-performing targets in real time. Strike Social’s AI software is able to improve YouTube ad performance (cost per view and view rate) by 25% while reducing execution time by 75%. And yes, by execution time, we mean reducing your staff by up to 4x.

We’re taking 1st and 3rd party data combined to build learning offers that are able to capitalize on the insights that our data provides and then accelerate the learning process. We’re personalizing users’ public content feed in News Republic and are seeing a significant increase in engagement.  When we apply a learning system to provide better personalization for news and video for consumers, that same personalization can be applied to the advertising stream. Ads become smarter than the content itself.

From the omnichannel perspective, there are problems with O2O (Online2Offline) measurement, but there’s growing demand for proximity marketing around the world to measure the research online, purchase offline (ROPO) effect. Synerise uses Apache Hadoop ​& Spark technology, in-memory processing, microservices architecture​ which provides an efficient way to store and process data in various formats and also allows us to handle ​billions of events ​and people interactions in real time without batch processing​. Since the implementation of Synerise Omnichannel Ecosystem, Synerise customer Gino Rossi experienced over 300% increases in loyalty program users, number of transactions, and e-commerce revenues.

When marketing data is cleansed and prepared, it is optimized for a specific use case (e.g., programmatic, campaign reporting, media attribution) and therefore creates competing versions of the data. Further, the technology often requires unique expertise creating a bottle necks and inefficiencies. Finally, given the lack of an enterprise-grade solution for analytics, there are many missed opportunities for performance improvement. AI has a new and different opportunity to impact the entire marketing function. Inclusive and enterprise-grade AI solutions will impact organizations by 50 to 500 basis points.

Marketers are likely spending more time managing marketing tools than they are actually marketing. In enterprise search, internal teams have a need for fast and easy access to content they need to help them do their jobs, that doesn’t disrupt workflow. Swiftype built the Enterprise Search Platform because our customers communicated the need to find information related to marketing campaigns from a single app: Marketo & Salesforce campaigns, reports,  tasks and projects and also related campaign assets from Dropbox and Google Drive. AI plays an important role here by also surfacing the most relevant content right within our daily workflow.

Lucy is a cognitive companion for marketers. She is constantly learning from new data she’s fed and every query makes her smarter on a given data source. She can learn an enterprise’s data as well as common industry sources of data and becomes smarter every day. Lucy has been successfully adopted by a number of companies, including Havas Media (one of the world’s largest media agencies) who is achieving a 75% reduction in vendor cost and experiencing a 7x faster campaign deployment thanks to Lucy.

Brands spend more than $60BN globally on sports sponsorships and are unable to capture the full value of their logo impressions on broadcast TV and also social media. With recent computer vision advancements, brands can now automatically analyse each frame of a NBA game on TV and also on YouTube or Facebook. Real-time image recognition capabilities identify each logo impressions, size and recency and determine what would be the media equivalency. Without technology, analyzing a 3 hour game for these different placements can take days. A machine can simultaneously analyze every sponsor and location within every frame of the video in a matter of seconds, allowing a full game to be analyzed in a few hours or less.

We’re seeing a 76% improvement for customer service because people are reaching out to a business to try and fix a problem and the examples of the problem are a rich source of training data that allows us to better understand what the customer needs regardless of how they explain themselves. We’re also seeing a 82% increase in understanding of social listening data by using AI to answer a specific problem such as ‘what are people complaining about on social media and how do we automate responses’ or ‘which images of our product are people sharing that highlight the product in a negative way and how do we create a model to manage the customers needs’.

The holy grail for marketers has always been personalization at scale and affordable customer acquisition costs. An automotive company used machine learning-based scoring to help their dealerships generate reviews on the right mix of review sites and saw the best performers grow their car sales 13% faster than the average. Machine learning have had great impact on lower funnel activities (like recommendation-based retargeting), driving 3-4x improvement in conversions after a site visit, compared to one-size fits all retargeting approach based on a limited set of pre-defined retargeting messages.

Separation of the MarTech and AdTech stacks creates an identity gap, where we are trying to uniquely identify the person we are advertising to before we have an identifiable piece of information from them. AI and associated tools allow you to overcome some of these issues in a number of ways. The first is the use of AI on audience segmentation and expected response where audience selection and segmentation can be made based on an algorithmic understanding (AI) of previous customer/lead behavior. AI plays further into this model by using the likes of survival models to establish how long clients will continue to transact with you.

AI is able to take the masses of data that is produced by online ads and convert them into tangible insights that advertisers can utilize. One way is taking existing and past ad performance to predict future performance of creative. Using ReFUEL4, Spotify was able to achieve a 40% increase in (CTR) over previous campaigns; 3X app installs over previous campaigns; 4.6M people reached.

Right now, it’s very hard for Google to determine the content of an image from a perspective of true understanding. In other words, they can get a nice array of tags and probabilities, but there’s a lot of error. With advertising, it’s absolutely critical that the computer derives a much more detailed annotation to be fed into the stack, and thus, along to the advertisers in order to bid. We have seen an ROI as far as 12 days over several thousand pages where images are present. Other clients of ours have seen a 4x improvement in their time-on-site from the user, by analyzing visual content to provide related material.

For many marketers, terms like AI, machine learning and algorithms are a foreign language. But a Marketing “master algorithm” could be a key that opens every Marketing lock. Machine learning as the limitless key means the ability to track a segments of customers even as they change, enter and leave segments without having to create a new algorithm each time. The same algorithm will learn to do things depending on the data you give it. It’s all about data.

Research shows the image used in digital campaigns has a 10x performance impact on audience reaction. Yet, most companies fail to find data to guide this decision. Our data has shown that clients who use data science to drive their digital marketing have seen an increase in engagement and reaction from their audience almost immediately. By tuning the voice, attributes, and cadence of new content, a brand’s messaging will reach the intended audience in a way that resonates with them and drives better marketing performance.

We are building a tool that mines social media to uncover narrative structures around brands, people, products, etc. We think it’s smarter for brands to segment people by the types of stories that resonate with them. We take an AI – not just ML- approach in the sense that we build a large measure of autonomy in the application: it is able to detect which stories resonate the most within a specific audience segment then automatically served that story (content) to that audience, measures performance results vs a control group, and layers the results into the models. In early alpha tests we’ve seen performance results of up to 25% click through rates on Facebook ads.