• Is sentiment analysis something you’re considering? If so, take a page from Ryan Watson, VP, Head of Media Platforms and Marketing AI at Kraft Heinz, who has built a team around this nascent practice, which entails using technology to parse consumer messages sent through social media, email, and other channels. 

    Listen to the podcast below or read an excerpt of our conversation:

    RampUp:
    What is sentiment analysis and why does it matter?  

    Ryan:
    Sentiment analysis involves extracting meaning and understanding from unstructured customer communications like emails, commentary on social media, or a phone call to customer service, then mapping that to specific categories, brands, or products, and doing it at scale.

     RampUp:
    Can you give us an example of how it actually works? What does sentiment analysis look like, not just from a technical standpoint, but also from a staffing standpoint?  

    Ryan:
    As an example, if a consumer posts on Twitter that they hate cream cheese,  I can see from Twitter’s raw data feed that particular user said, “I hate cream cheese.” We then pass the verbatim of that text to an API endpoint and tie it to that consumer’s profile on our side, without ever having Google understand who that person is.
    Now that we know that this individual said something negative about cream cheese, we want to figure out how to leverage that to better serve content and marketing and messaging to them. In this case, we might create an internal blacklist for a period of 90 days. If we’re running a cream cheese campaign programmatically on the open web during that time, we’ll exclude that individual from those audiences.
    Regarding staffing, for us it makes a lot more sense to rely on external partners that have off-the-shelf plug-and-play technology that is purpose-built to do this. We have some strong technical talent in-house, but right now it doesn’t make sense for us to create our own sentiment analysis models. But our number one concern is always making sure that consumer/user privacy comes first. You always have to remember these are human beings—individuals who are sharing their thoughts, sometimes in a very public forum like Twitter, but also in more private ways, like calling our customer service department. What we’ll never do is pass any PII—anything that’s personally identifiable. We always make sure that we’re doing up-front data engineering work to strip out anything that could possibly identify an individual. It’s important to at least have in-house staffing to make sure that anything sensitive is stripped from your data whenever it leaves your ecosystem.

    RampUp:
    What I really like about that example is it really brings together the two aspects of your title as Head of Media Platforms and Marketing AI. How has your ability to apply AI-derived insights and analytics changed Kraft Heinz’s approach to media buying and media planning?  

    Ryan:
    Everybody can see the writing on the wall that over time, it’s not going to be a binary 1 to 0 shift where today, everybody is marketing like this and tomorrow everybody’s marketing in a different way. For the last nine or ten months since we started on this journey, we’ve focused on laying the foundational capabilities and elements that we need to be able to ultimately bring something like this to market. Now that we’ve done that, we’re shifting into activating and executing on the back of this data and these insights.
    I always suggest to people is to start slow. Crawl, walk, then run. It’s easy to say we’re going to build this amazing whiz-bang capability or system or whatever you want to call it, but in many cases, that can take multiple years to do. I don’t see a future where any company is going to say, “it doesn’t make sense for me to personalize. I don’t think my customers want to see ads for something that aligns with their likes, wants, and needs. I think we can all agree that that’s not the future.
    Consumers want to see content that properly reflects how they think about the world. I’d love to say that we’ve cracked the nut and have figured it out, but it’s not an easy task. We do, however, have a clear idea about where we want to go and why, but we’re realistic with our goals and try to manage expectations as much as possible. It takes time to get there.  

    RampUp:
    What is your advice for companies at the beginning stage of selling in a  far-reaching two- to three-year project? And also for those at the stage you’re currently at, where you’ve done some of the foundational work and now you’re starting to put out pilots. What information or reporting are you sharing that assures the other teams that yours is moving in the right direction and that this is something they should continue investing in?  

    Ryan:
    The first priority is being very specific about the business problem you’re trying to solve when talking to broader stakeholders and people in non-technical roles to get buy-in. Strip away the tech buzz words and behind-the-scenes details and focus on the challenge or problem at hand and how you plan to solve it.
    The second thing is to put up wins on the board quickly. Bring the organization along with you by getting feedback and evolving as you go so your actions aren’t viewed as some dark corner where people are doing a bunch of interesting but opaque things and no one knows the value it drives.  Be very clear, bring the organization along with you, and get some wins on the board.  

    RampUp:
    To your point earlier about stripping out PII and making sure that all the data you’re using is privacy conscious, what part did working with your legal counsel and your privacy teams play in selling this project?

    Ryan:
    It played a very important part. I have a recurring meeting every few weeks with our Chief Privacy Counsel and the people who are responsible for our regulatory affairs. We obviously want to protect the company and avoid financial penalties, but regardless of regulatory issues, it’s very important, ethically, for us as a company to make sure we’re trusted by consumers and that we’re protecting that trust.
    We’re actively trying to focus our bias towards first party data, precisely because I don’t think any of us know what’s going to happen to non-first party data (purchased data, social media data, etc.) from a regulatory and a technological standpoint. We want to focus on the things that are deterministic, authenticated, and protected, and then selectively layer in third-party data in a way that’s reversible, so if new regulations are passed, we can change course and pull back.  

    RampUp:
    It all sounds very sophisticated and something that I’m sure a lot of companies are striving for in this regulatory landscape that we’re in right now. I look forward to hearing more about how this project evolves and perhaps a little bit more about how Kraft Heinz is moving more towards personalization and how that is solving the original business challenges that you laid out. Thank you for your time. 

    Ryan:
    My pleasure. Thanks for having me.  

    Subscribe to RampUp

    Subscribe