Jessica Miller, VP and Managing Director of PR20/20, comes on the show to talk about artificial intelligence: how agencies can define what it is, how to talk about it with clients, how to identify real applications of it, and how it impacts service delivery.
Hi, folks! My name is Kevin Dunn, and welcome to "Agency Unfiltered", a biweekly web series and podcast that interviews agency owners around agency operations, growth and scale. Nobody knows how to scale agencies better than those that are already doing it, and they're happy to share an unfiltered look into what has worked and what hasn't. With us today is Jessica Miller, V.P. and managing director of Cleveland-based PR 20/20. She digs into artificial intelligence, specifically, how she would define AI, how to identify applications of AI, when she would look to AI for solving client problems, and how she's seen it impact client engagements. Prepare to have artificial intelligence de-mystified. This is: Agency Unfiltered.
KD: Jessica, welcome to Agency Unfiltered. We are super psyched to have you. I think this will be a topic that's incredibly interesting and obviously, with your background with MAICON, and what PR 20/20 does on the artificial intelligence front, I think you're absolutely a perspective that we want to capture on it. I think the best place to start is that when I say artificial intelligence, I think folks have a hard time, the real world application or definition of it. So if I was a client or anybody asking what is artificial intelligence, how do you go ahead and define what it is?
JM: Yeah, we would definitely say to you, our client, this audience, there are a hundred definitions out there, so it's okay that you're confused. The one that we love the best is from Demis Hassabis, founder of DeepMind with Google, a really cool project if you want to look into that, but he says it's the science of making machines smart. So, the way I like to break that down for our clients is is the technology getting smarter on its own? Can you give it a goal and can it solve it? More than an if/then type of algorithm that we're used to today.
KD: And then it just does it do so more efficiently over time as well? Or does it becomes more effective at its ability to do so?
JM: Does it become smarter on its own, so does it learn, can you give it feedback, does it learn based on data without your feedback, just some of those types of questions is how we filter AI versus not AI.
KD: That's great. And then what are the applications of artificial intelligence that agencies should be cognizant of or are there opportunities? Is it just finding the right software that says they're AI-powered? So how does artificial intelligence kind of fit into this agency sphere?
JM: Yeah, I think a lot of what we're doing is super data-intensive. Some of it's really repetitive, and a lot of it does have that end goal in mind, so achieving a goal. Those are the three questions we ask when we're thinking about can we use AI to do this? So, that spans across the marketing services we offer our clients, the sales goals they're trying to achieve, and then agency operations as well. So a lot of times, we're just breaking down your day into is this data-intensive? Is it something that's repetitive? Does it have an end goal in mind? And if the answer is yes to that, then we can look for an AI type of solution. And I think the way we're approaching it too is with that long-term vision for what's possible and where the industry will be in who knows how many years it'll take to get there. It could be next year, it could be 10 years.
KD: Who knows for sure?
JM: I honestly don't know. But then we're also thinking in a shorter term with how we can help little pilot projects to prove either more value, more efficiency, like you said, or just better personalization, those types of quick wins simultaneously to learn how we're going to use technologies, and get there to that longer-term vision.
KD: Do you have any examples of the things you've been able to operationalize with AI-powered tools at all?
JM: Yeah. A lot of our services right now are either that very high-level consult, so giving companies like Roadmap to define their use cases, how they can think about what to do AI on, what that pilot project will be, and then helping them find certain technologies for their pilot project, whether that is tagging content more effectively so that it's searchable and findable by your audience, but also by the AI techs, thinking about data both in that long-term and short-term, how to organize, how to collect, how to make it usable for technologies. But yeah, a lot of it is that consult to get to the pilot project, and then thinking about data just in a long-term type of way. And then some of the more specific things we've recently introduced, I wouldn't call them AI but I would say they're making our services more efficient around content marketing. So we've done a lot with language generation, if we have data-driven content for a client. We had a client a couple of years ago who did ad reports every quarter. Here's what the top retailers are spending, here's where they're spending it. It was really data intensive.
KD: Pretty boilerplate and templated, but the data inside changed.
JM: Exactly, exactly. So if there's a template for the formula that you're writing, even if you go to a news-writing class, my background's journalism, it was, here's your title, here's your lead, your dateline, your intro sentence. Even if you can get that formulaic about your content, there are opportunities to do it faster with some natural language generation that's out there today.
KD: That's great. So it makes your ability to produce these reports more efficient. How do you position it to clients? Are you just saving time and effort on the back end, or are you actually leaning into this positioning of like, hey, it's actually machine-driven or natural language-generated reporting. Is it something you put out to the clients or is it just on the back end to make things more efficient for you guys?
JM: We're really transparent with however we're producing work for our clients. And I think I have never personally leaned into our tagline for the agency is Look Beyond. I've never fully understood that until a couple of years ago when we started doing this AI stuff, because I'm like okay, this is literally “looking beyond”. And I do think the type of clients we attract and the type of employees we attract are really into this whole what's next type of thing. So we've had a lot of success positioning our clients as this forward-thinking change agent within their company. They're so excited to see behind the curtain and take this to their enterprise and be like, look at what we're doing here. It's saving us. When we pitch new stuff to them and the investment that goes along with it, we're saying, hey, here's what you're probably spending per year having a senior level person write these personnel releases for example, something that's very formulaic. And this isn't even AI AI yet, but it's doing things more effectively and automating a process that if you're paying someone six figures to write releases, think of all the things they could be doing. And that's what we try to position it as as well, if you can kind of use machines to help with some of the more time-intensive, repetitive or data-driven tasks, you can then focus on that huge brand launch, that client meeting you might not have time for, or really the empathy, the relationship building, the trust. Those more creative pieces that I think we all got into marketing to do.
KD: Just being able to just show time savings, cost savings, right?
JM: Yes, true.
KD: And I think, so it sounds like you're attracting the right type of clients that are open for this, so I think it's a certain type of business or a person that wants to look beyond and test some of this stuff. You mentioned earlier that there's pilot programs to see other ways to be more efficient the way we do things. Do you roll off those pilots on clients? Are they open to it? Or is it more like experimentation on the agency itself when you're validating new options?
JM: I say both. For a client doing something brand new, we definitely have a set price. We might say this is a huge 55-point project because it's going to save you down the line so much. But for your pilot, we're learning this with you, let's do a 13-point version of it where we're eating some of that learning time because we haven't done it before. So that lets our agency do cool stuff that we can then talk about and show them the value for, and then they're willing to take that risk further on. And that is the cool thing about, I would say, the marketing AI institute part of the agency is we're talking to all these founders who have this amazing technology and they're giving us a look at it so we can really have that front row seat to be able to make recommendations for our clients, but to also kind of get our hands on some of these cool techs.
KD: When we say artificial intelligence, I feel like every tool, every software, every company is saying that what they do is AI-driven. Are you starting to see more and more? Do you have any tips for folks on how to validate when things are actually AI-powered, or how do I validate that a tool is actually using artificial intelligence? How do I cut through the noise in that regard?
KD: Yeah, I would ask, what's HubSpot's point of view on AI? Go to your existing vendors that if they are saying, we're AI-backed, we're doing this with AI, ask them, okay, what's your vision? What's your strategy? Ask them how it works. Just plain out, how much data is required to inform your solution? How much data is aggregated in a black box to inform the type of outcomes that are going to be recommended? If it's an email subject line or a CTA type of tester, I would just ask quantity, how much, what's happening? Is it a black box? That's kind of a little bit of a red flag. I want to know, okay, well, even if you don't know exactly what's happening right now to form these recommendations, who built it? How'd they built it? What type of strategy went into setting it up? What type of data is informing it? And how are you monitoring the adjustments, the recommendations it's making? And then, I would ask for success metrics. I would ask, give me or show me a couple customers.
KD: Like social proof?
JM: What has it done? Everyone's kind of putting AI on the sticker on their products because I think there's something we read that said you can sell 50% more if you just say this is powered by AI. So we are asking, that question we went back to with the definition, is it becoming smarter on its own or is it an engineer or developer saying, well, if this, then that, over and over. So I think is it smarter on its own is the big question but ask for quantity of data, how it was set up, use cases.
KD: And then yeah, success stories or anything like that. That's great. That's like the perfect second layer of questions, right? So you can actually go to this person, like, okay, how do you define AI? How do you explain it? And actually try to call 'em out on it just in case.
JM: Yeah! It's not even calling them out on it, it's just there's so many definitions out there, it's really interesting to hear their point of view.
KD: And does that align with what we're looking for, right?
JM: Yeah, exactly.
KD: So my next question is, so we talked about natural language generating reports and things like that, but are there any other ways that you've seen artificial intelligence manifest itself in the way you provide services to clients?
JM: Yes. And I would say it's such a variety because there's so many solutions out there and ways you can approach it. So the way we're approaching it with clients is either a use case-based model or a problem-solution type of model. So I have a challenge, I have this email audience and they're not engaged. That's one way to think about it and there's a lot of solutions out there to help solve that specific challenge for the client. To get them to even create that problem statement, you might enter with a use case model approach where we're saying, here's the things your team is doing on a daily basis. How often are they doing it? How long is it taking? Is it dated back? Is there a goal to achieve? And is it repetitive? So then we're helping prioritize and kind of score their use cases to say let's try this. And so then, we're showing, is there technology out there that can solve it? Would it be a build? Try to get their budget in place, and then actually help them tackle it. So I would say yes, there are a lot of things we're helping roll out and they just vary because marketing varies from strategy to form supporting.
KD: That's just not boiler-plating, right.
JM: Yeah, to operationalize, writing. There's just a lot out there in the marketing menu.
KD: That's fair. All right, hard-hitting question for you: are we going to be replaced by robots?
KD: Okay, so last question for you, and I ask all my guests this, but what is the weirdest part of agency life?
JM: I think agency people, in general, get to work hard and play hard. So I don't think it's weird, but I do think by Friday, everyone's kind of just at their limit. Your brain has been working so hard for a week, five days in a row, and you're just ready to go crazy. So it's fun when you have that team and the clients and the culture where you can just go to happy hour. But yeah, there's—
KD: Those Friday afternoons, who knows where this is going.
JM: Yes, Friday is good. So they get weird. Yeah, and then even being at INBOUND, I've never been to a conference where you have a big ball pit and there's a bunch of people taking selfies in it. The snaps they send home to my family are definitely weird. They're like, what are you doing here?
KD: All your friends are, what is inbound? Do you know?
JM: I know. They're like, where are you?
KD: No, there are some installations. It's awesome, yeah.
JM: Yeah. So I'd say that's the weirdest part of agency life. It's all the fun that marketing brings.
KD: Awesome. So I say that's the last question, but maybe this is. We usually end there, but the final final question. So what is the absolute next step if I am somebody that just heard you share what you did about artificial intelligence, where do I go to learn more? How do I get started? What's my next step after this taping?
JM: Thank you. That's a good question. We started a blog, Marketing AI Institute because we were just learning, we were just finding this really interesting and we wanted to share what we were learning there. We're marketers so we blog about it. I would say go there. There is this cool resource, it's like 100 plus resources to get you started. My favorite on that list is as a team, we took a Coursera class by Andrew Ng, and it just put us all on the same page and got us speaking the same language and seeing the potential of what AI can do. So as I go to the institute blog and then take that Coursera kind of courses too for you.
KD: I think I'm due. I think I need to take it. I'll do it with everybody else.
JM: It's easy, yeah.
KD: Yeah, awesome. Well, I think we're out of time, so I really appreciate you coming on and joining us. It's been fun.
JM: Thank you, it's been fun for me.
Thanks for tuning in to another episode of Agency Unfiltered. If you like what you saw, heard, or read, make sure to subscribe to our playlist on YouTube, our podcast on Spotify or Apple Podcasts, or our newsletter on AgencyUnfiltered.com. Alongside episode launch notifications, the newsletter also comes with a ton of other helpful, strategically curated agency content from yours truly. If you want to keep the conversation going or provide a counterpoint to this episode's discussion, tweet me at @Kevin_Dunn. I'll see you again in two weeks. But in the meantime, keep it unfiltered and let's all grow.