Artificial intelligence generates lots of interest and more than a little bit of fear among agency owners. How will machine learning, AI, and all that super-technical stuff change agency life? Will it make agency work irrelevant?
Not according to my guest, Paul Roetzer from PR 2020 and The Marketing Artificial Intelligence Institute. In this episode, Paul shared how his agency is leaning into AI because of the power and possibilities he sees in terms of agency efficiency and profitability. He makes a strong case that AI has the potential to make agency work more intelligent and even more human.
AI is one more way agencies can leverage new technology and new tools to serve our clients better, to help them grow their businesses, and to more profitably, efficiently, and effectively grow our own agencies.
I’m sure that some of you find this a little scary to even contemplate. But just like we’ve embraced all of the technologies before AI (the internet, mobile, programmatic media buying, etc.) we’re going to have to wrap our heads around this one too.
One of the best aspects of owning an agency is that we constantly get to evolve and re-invent ourselves to better serve our clients. AI gives us all the opportunity to scale and grow in ways we couldn’t imagine. AI isn’t about robots stealing jobs. It’s about the potential to eliminate the boring, repetitive tasks so we can spend more time thinking creatively.
Paul always sets his eyes toward the horizon. He’s continually wondering what will happen next in our industry and how he and his agency can be at the forefront of that. So, I wasn’t at all surprised when Paul and I were talking a few years ago and AI started to creep into the conversation.
In the last year or so, Paul has doubled down on that, not only in terms of what he’s doing with his own agency but also through his new organization, the Marketing Artificial Intelligence Institute. Later this summer, the Marketing Artificial Intelligence Institute is presenting MAICON, an AI convention for marketing leaders (use discount code McLellan19 to save $100 off the registration fee). Its mission is to make AI approachable and actionable for modern marketers so they can use this technology to build a powerful competitive advantage.
What You Will Learn in this Episode:
- How intelligent automation will continue to make repetitive agency work easier
- Why AI isn’t after your agency job
- How machine learning can help you share data with clients in a cost-effective way
- How agencies can understand AI and be a learning resource for clients
- How to develop use cases for testing AI in your agency
- Why small and mid-sized agencies are well-positioned to pivot into AI
Subscribe to Build A Better Agency!
Ways to Contact Paul Roetzer:
- MAICON: https://www.marketingaiinstitute.com/events/marketing-artificial-intelligence-conference-2019
discount code: MCLELLAN19
- Website: https://www.pr2020.com/
- Website: https://www.marketingaiinstitute.com/
- Resources Page: https://www.marketingaiinstitute.com/resources
- New Intelligent Automation Solutions Post
- How Do We Prepare College Students for the Age of AI? Post
- Book: AI for Marketers: An Introduction and Primer: Second Edition
It doesn’t matter what kind of agency you run, traditional, digital, media buying, web dev, PR, whatever your focus, you still need to run a profitable business. That’s why Agency Management Institute started the Build A Better Agency podcast a few years ago. We help agencies just like yours, grow and scale your business, attract and retain the best talent, make more money and keep more of what you make. [inaudible 00:00:25] has 25 plus years of experience as both an agency owner and agency consultant, please welcome your host, Drew McLellan.
Hey, everybody, Drew McLellan here with another episode of Build A Better Agency. I am super excited to bring my guest to you today and to raise this topic, which I think has been floating around on the edges of agency consciousness for a while, but I don’t think most of us are thinking about it with the clarity, and the emphasis that we need to, and that is the whole topic around how does artificial intelligence impact our world?
So, just like in the days when, for those of you that are really long in the tooth like I am, remember that when there was a time when you worked at an agency, and there weren’t websites, or there wasn’t access to the internet. And then remember when we were just manually sending out if we had to do any mass marketing, and then the electronic way, it was by fax, or we were manually keying in email addresses and sending them out before the Mailchimp’s of the world came, Constant Contact came into play.
And then we thought, “Okay, that’s the Nirvana.” And then all of a sudden, SharpSpring and HubSpot, and all of the marketing automation software showed up. And all of us have retooled our business around all of those technologies. And all of those technologies eventually made the work that we do better and easier and more efficient, programmatic media buying. There’s all kinds of, if you look back on the history, just the recent history of how long we all have been in the business, there have been some dramatic changes that technology has brought to our business and artificial intelligence is just that, it’s another way that we can leverage new technology and new tools to serve our clients better, to help them grow their business, and to more profitably and efficiently and effectively grow our own agency.
There is no one on this planet that I think knows more about this from an agency perspective than my guest Paul Roetzer. So, Paul, many of you know Paul from… he owns a PR shop in Cleveland called PR 20/20. Lots of you have probably read his books Marketing Agency Blueprint, Marketing Performance Blueprint. One of the things that is true about Paul is, and I’ve known him for years, he’s always wondering where the puck is going next. He’s always wondering what’s going to happen next in our industry? And how can he and his agency be at the forefront of that.
So, I wasn’t at all surprised when I started talking to Paul a few years ago, and he started really leaning into artificial intelligence and how we can use machine learning and all of the tools around that, to better serve our clients. So in the last year or so, Paul has really doubled down on that not only in what he’s doing with his own agency, but he has created a whole new organization called the Marketing Artificial Intelligence Institute. The whole purpose of that organization, which is a separate organization from his agency, is to educate us and to probably provide resources, I think down the road, for agencies and brands to really capture the power, that is artificial intelligence, and use it to change and evolve our businesses, as these new opportunities present themselves.
And so, many of you, if you’re on the E-newsletter list for AMI, you have been seeing me talk about a conference, the MAICON conference, M-A-I-C-O-N. So the marketing artificial intelligence conference, which is going to be held this July, it is July 16th and 18th, that is coming out of Paul’s organization. It’s going to have a separate agency teaching agencies, the one-on-one, the beginner level of what is artificial intelligence, how is it already showing up in our business more than we recognize it or not? And what are the opportunities for us as small and mid-sized independent agencies to learn more about this and to leverage it inside our business?
If you have an interest in going to that conference, we’ll include a link in the show notes, but basically, it is, the conference is M-A-I-C-O-N. That’s the website as well is never mind, don’t worry about the URL, I will include it in the show notes, or just google MAICON, M-A-I-C-O-N, or the marketing artificial intelligence conference. July 16th and 18th, if you want to go, there’s a discount code you can use, it’s mcclellan19. So M-C-L-E-L-L-A-N 19. And you’ll get $100 off of the registration.
I will be there, I’m going to be running a panel, talking to folks about how agencies can begin to harness the power that is AI. I think a lot of us are worried about it, we’re scared about it, we’re worried it’s going to change our business that’s going to knock people out of jobs. I think that’s all what I want to talk to Paul about is what does this mean for us as agency owners? What does it mean for our agency in the future? So, without any further ado, let’s get into the conversation.
All right. So without further ado, Paul, thanks for joining us today, this is going to be a fascinating conversation.
It’s always fun to talk.
It is. So, I gave everybody in the intro a little bit of your bio, but walk us through your agency, and you have taken an interesting evolution. So, you start an agency, and then you basically, when I think of you and I think about your agency, I always think of you as somebody who’s looking for, what’s the next thing? Where do we have to evolve to before everybody else gets there? I think that the books Marketing Agency Blueprint and Marketing Performance Blueprint are that where you saying, “Here’s where the puck is going, if you will,” [inaudible 00:06:38] Wayne Gretzky, “here’s where we need to go.” It feels like this most recent shift is part of that. So, walk us through… I guess what I’m asking you is how does your brain work?
Oh, man. We don’t have time for that.
[inaudible 00:06:51] a lot of agency owners, in that you are constantly looking to reinvent yourself.
Yeah. So I think it’s a blessing and a curse. The tagline in our logo is look beyond and I feel like, subconsciously, I’ve always lived that. So, in 2000… Well, I launched the agency, it was standardized services, preset pricing, it was how do we change the dynamic of the pricing model in the agency world, make it more transparent, make it more outcome-based? So that was the original, outward-looking how do we move the industry forward? Then in 2007, we were the first HubSpot partner. And it became, “Wow, marketing automation, we can bundle our services around this thing. And this is going to be big.” And if Brian Halligan and Dharmesh Shah are successful at disrupting SEO and PR and all the industries they wanted to not necessarily take down, but definitely improve, there’s this massive opportunity.
So we went all in and we built the first inbound marketing agency, and I wrote a blog post called Dawn of the Inbound Marketing Agency in September of 2008. And that became the internal rallying cry at HubSpot to build this ecosystem. Then 2011, I wrote the agency blueprint. And then, unfortunately, for me, or fortunately, I guess, IBM Watson went on jeopardy in 2011. And that set me down the path of what is this? How does this artificial intelligence technology work? And how is it going to transform what we do? And I just went on a period of discovery, of reading everything I could about AI and trying to put the pieces together. And while the world was pouring billions of dollars into marketing automation tools in 2012, to 2014 to 2015, I was looking at those tools and saying, “But they’re preschool level. You just have to do all the work.”
And so for seven years now, I believed that we were going to arrive at a day where Intelligent Automation would actually change what we fundamentally did as an agency. And so I hope, I’ve now after all these years, gotten to the point where I can stop and sees what’s actually here instead of… because I think AI is the thing he was all leading to is like we saw it coming and now it’s here. And now we can really build on this instead of me now like, “Okay, what comes after AI?” It’s like, “I’m good for the next 10 years. Let’s just do it, yeah.”
So recently, you published a blog post called Dawn of the Intelligently Automated Agency.
We’ll include a link in the show notes for everybody to check this out. But give us a 30,000 foot level of what you were trying to do with that blog post, and what we as agency owners should take away from it.
Yeah, and it’s obviously a bit of a, I just mentioned in the Dawn of the Inbound Marketing Agency, it’s a bit of an ode back to 2008. So 11 years later… So for the last two years, I’ve been doing talks around the world on artificial intelligence. And I’ve said, “I believe that at least 80% of what we do as an agency will be intelligently automated to some degree in the next three to five years.” And here we are two years into that period. What I meant was, there’s all these repetitive data-driven tasks we do every day. If you’re in content marketing, it’s figuring out what to write about, when to publish it, what channels to push it through a social, how much paid media to spend on things.
There’s all these things we do. If you’re in social, it’s what to share, when, what hashtags to use, what images to use, when to publish, when to republish. I’ve just looked at all these things and said, “A machine can do all that better.” I know it can. And there’s technologies built to do each of those things better at scale than humans. And so we finally just got to the point where it’s like, we need to start doing this. And so that’s what the intelligent automate agency was about. It’s like, our path to get to this point and our first steps to actually build a suite of services around it.
So I think, for a lot of agencies, this is scary stuff. It feels like they’re going to be extinct soon. And that wasn’t the sense that I got from what you wrote or all the conversations you and I have had around this. So frame that up in terms of what does this mean for the average agency?
Yeah, my belief, there’s certainly dystopian views of what artificial intelligence will do to marketing, to the industry, to the agency world. I tend to find the net positive outcome side that it’s going to take a lot of the things we don’t like doing anyway, and automate them. So, for example, there are very few agency owners, marketers, people with journalism backgrounds, who like analyzing data and building pivot tables and trying to extract insights from that data, or digging into Google Analytics, like, “Why is the blog post traffic down?” And trying to actually find those answers, it’s just not what you enjoy doing.
So imagine a machine much like a Google assistant in your life. It’s just surfacing things for you. So as the agency owner, as the marketer, it’s coming up with some insights and saying, “Here’s what I found. Is this interesting to you?” And now, instead of building the pivot tables, I’m just looking at some insights from them and trying to figure out what to do next.
So our belief is that it’s actually going to make us better, it’s going to make us enjoy our jobs more because it’ll take some of the mundane, repetitive things out of it. And it’ll enable us to be more creative, more strategic, more empathetic, more human, which is really, what would make us all happier anyway? So, it’s like looking through the rose-colored glasses, I suppose, of what I think is going to happen. I’m not pretending like there’s no downside to all this. But I think the long game is it’s going to make us better agencies.
Well, as I listen to your talk, I think about for many agencies, the things that they have a hard time articulating to a client, why they’re so labor-intensive, and why they cost so much? Every client wants reporting frigging every day, but they don’t want to pay for it. Agencies are trying to figure out how do I staff for this? How do I have somebody who is articulate and a good writer, but also understands the numbers do this? And how often do I do it? I know it’s going to be a loss later in most cases because if I actually charge a client $150 an hour, which is the average agency hourly rate these days, to do this, they’re going to have a cow because they don’t understand all the work that goes behind it.
So, as I listened to you, and as I read some of the things you’ve been writing, I guess the optimistic version of me is like, “Well, you know what, maybe this is a way for us to actually get paid for our intelligence, but not the manual labor, it took us to get smart.”
Yeah, there’s a great book called Prediction Machines, that’s written by three economists. And they look at the future of work. And they say that… because machine learning, which is the primary type of artificial intelligence is all about making predictions. That’s really what it does, it makes predictions on future outcomes based on historical data, and then it keeps learning, that’s the name. So it’s literally machine learning. It’s like Google Maps recommending a better path, it’s learning ways to…
So they say that the two jobs that will exist is telling the machine what to predict and then using human judgment of what to do with those predictions. That’s what we’re trying to get to. It’s like, I don’t want to spend all my time trying to figure these things out. And like you’re saying, we bundle strategy and performance, like the performance reporting into our service packages. So when someone’s on a monthly recurring revenue model, and they’re paying $8,000 a month, or $13,000 or whatever it is, we give the performance reporting away, which is crazy, because in our opinion, it’s probably the most valuable thing, but they don’t value it. The reports don’t.
So, we have actually, for years, bundled it into our pricing model to just do that. And then our strategy, we need to talk about it because they don’t want to pay for that. So, for us, it’s always been about well, how do we do those things more efficiently, and create greater value? And that’s what we did with our automated performance reports, which is one of the things we recently introduced. So we trained a model to take 12 core questions you would want to know from Google Analytics every month. The machine answers those 12 questions. And so on the first business [inaudible 00:15:04]
Give us an example of how that would work. So on the first business day of the month.
Yeah, we go into the software that we’ve taught to write that report, and it literally just pulls data in through the API from Google Analytics. And it writes human-sounding narrative to say, “Was traffic up or down? How is that compared to last month, last year, same month? Was the blog post traffic up or down? Were people more engaged? What were the top blog posts? Our goal conversion.”
So the standard 12 things that as a marketing leader I would want to know, it just write some and literally on the first day of the month, because what we used to do is you would have people who aren’t analytics by trade, usually our staff is journalism, communications majors. So we have to first teach them how to look at analytics, then teach them how to extract insights from it, then they would spend 6 to 8 to 10 hours building a deck with all these charts and screenshots and trying to come up with insights, we just automated that whole process.
So something that used to take 6 to 10 hours, literally takes 6 to 10 seconds now, per client. And then we have a senior strategist that looks at that and says, “Are there any anomalies? Is there any insights we can extract from this? Is there any value we can add that otherwise, the client was paying six hours for potentially?” Now they don’t have to. We charge $150 a month or something for the basic report.
Right. So, where does the line stop? And so, you gave the [inaudible 00:16:38] example, it gathers all this data, but it’s not predicting, or it’s not recommending what to do based on the data or does it?
Not in this case. So all this is doing is writing a narrative of what happened. It’s not saying necessarily why? We do have some of it that actually does answer the why question, but it doesn’t drill down really far. And it doesn’t tell you what to do as a result of it. That’s still where the human piece comes in. However, if you have access to Google Analytics, you can see Google’s attempt to do that part for you.
So if you go into the insights in your app, and just click on insights, is that little circular thing at the top that usually has a little bubble you looked at ever, it’ll make recommendations to you. It’ll surface things and then when it does, it says, “Was this helpful? Yes or no?” That’s machine learning. You’re actually telling the machine, “Show me more things like this or don’t.” And it learns, and it evolves what it shows you.
So we’re moving in that direction. But you and I’ve talked before about this Intelligent Automation, if you think of autonomous vehicles, and you go from one to five as the standard rating, five being a fully autonomous, there’s no steering wheel, it just goes. We’re trying to get to a two. We’re just trying to take all manual, and do some machine-assisted stuff that maybe uses a little bit of AI, but it really is irrelevant. It’s elegantly automating a process that otherwise was really inefficient. And that’s what we focus on.
So, I’m thinking that a lot of the listeners as soon as you said, “And then we taught it to gather all of the data and write this stuff,” they just went, “What do you mean you taught? You taught who or what?” And then I want to get into how you moved into the institute and all of that in a second. But what does that mean? Who What did you teach how to do that?
In essence, what we did is back in 2015, we started a skunkworks project internally called Project copy scale. And I basically said, “Can we scale content creation through machines?” Because I had sat in on a panel at South by Southwest where the Associated Press told the story of how they were using machines to write earnings reports. So I basically said, “What is that technology? How does it work?”
So we went and found what’s called natural language generation technology. In essence, it’s just software like a HubSpot, you tell it what to do, you create a branching logic, just like you’re building an email workflow, like they downloaded this, do these things, except in natural language generation, you’re saying traffic was up 10% so it skyrocketed. Goals were below 5% of goal, they plummeted. You’re teaching it how to write and then you’re creating a human-written template that has all these variables. Basically just pulls data from a spreadsheet.
So if you can write a story, and envision branching logic, you can actually teach a machine. If you have the right software, you can teach it how to write it. It’s almost like you write one press release or one report and you turn that into a template to then write 1,000 of them.
So, what drove you to this experime