Episode 10, Season 2
Content Marketing Strategy: Utilizing Artificial Intelligence to Optimize Content
Jeff Coyle, Co-Founder and Chief Strategy Officer of MarketMuse, joins host Kriste Goad to share his expert insights on AI-powered content strategy optimization. Along the way, Kriste and Jeff discuss the impact of AI and machine learning for healthcare marketers, the evolution of smart voice technology, and how to create – and optimize – content through the lens of a subject-matter expert.
What if marketing teams could boost efficiency and speed while simultaneously creating more space for creativity? With artificial intelligence, the possibility is promising.
Jeff Coyle is the Co-founder and Chief Strategy Officer of MarketMuse, an industry-leading content solution that brings data scientists and engineers together with content strategists and SEO experts to make AI accessible for content teams of all sizes and skill levels. Jeff has had his finger on the pulse of content marketing and strategy optimization almost since its inception, and in this episode, he shares his expertise on how to optimize workflows – at every stage of the content lifecycle – by harnessing the power of AI.
In this episode, we discuss:
- Jeff’s journey with MarketMuse, from early-adopter to co-founder
- How MarketMuse helps writers, editors, and publishers create content through the lens of a subject-matter expert
- If AI “posers” exist and, if so, what to do about them
- What is artificial intelligence, really?
- The evolution of technology, including AI and smart voice tech
- Is AI a threat to healthcare marketers?
- Two ways to train clients to optimize existing content
More From This Episode
(1:03) Jeff explains his journey to MarketMuse and how he became a content strategy and optimization expert
(6:56) Jeff explains AI-powered-content research, intelligence and writing
(8:44) Jeff and Kriste discuss MarketMuse’s capabilities, including how they work with agencies, what that looks like and how they benefit both agencies and their portfolios
(19:14) Jeff shares what it means to optimize existing content
(22:26) Jeff and Kriste discuss “posers” in the artificial intelligence space
(32:31) Jeff and Kriste talk through the evolution of technology, including smart voice tech and artificial intelligence, and the quick-to-judge mentality of consumers
(38:13) Jeff and Kriste discuss the impact of AI and technology on healthcare workers, particularly healthcare marketers
(41:49) Jeff dives deeper into how to get started optimizing and updating current content with two methods
[00:00:04.300] – Kriste
Hey, welcome to How It’s Done, a podcast for curious marketers. I’m Kriste Goad. I’ll be your host and I’m really glad you’re here.
[00:00:15.400] – Kriste
Today, we’re talking to Jeff Coyle, who’s co-founder and chief strategy officer of MarketMuse. MarketMuse offers AI-powered content research, intelligence and writing. Welcome to the show, Jeff.
[00:00:29.110] – Jeff
Thanks for having me. Should be a great discussion.
[00:00:31.630] – Kriste
Yeah. How are you doing today?
[00:00:33.430] – Jeff
I’m great. I’m looking forward to talking about all things content strategy and getting into artificial intelligence and being on your show!
[00:00:47.050] – Kriste
Thanks for coming on. So tell us a little bit more about MarketMuse. How did it come into being? You’re a co-founder – tell us a little bit more about your background. And of course, I need to know what the heck is AI-powered content research, intelligence and writing?
[00:01:03.580] – Jeff
OK, I’ll start with the short version. So I went to school at Georgia Tech for Computer Science.
[00:01:14.260] – Kriste
[00:01:14.260] – Jeff
[00:01:15.100] – Jeff
So I started working in the search engine space – I did usability theory as my specialization for school – but that was when search engines were enterprise search, right?
[00:01:27.073] – Kriste
When was this?
[00:01:28.960] – Jeff
It was trying to figure out how to search a Web page. That was ’99, 2000.
[00:01:33.310] – Jeff
I’ve been doing search and trying to figure out how to get websites to rank, and whatever the active search engines were, for about 22-23 years, which is kind of frightening.
[00:01:44.080] – Kriste
You’ve been there from the ground up.
[00:01:46.660] – Jeff
[00:01:47.290] – Kriste
[00:01:48.310] – Jeff
I started working with a company called Knowledge Storm. The story makes sense of why I tell it this way, by the way, when you get to the end.
[00:01:55.600] – Jeff
So I started working for a company called Knowledge Storm, and we were the first company trying to get people to put content on the Internet for B2B technology. So we’re trying to convince, for example, IBM that content marketing is a thing. You want to get your Web pages on the site.
[00:02:11.470] – Jeff
You want to get your PDFs – I mean, talk about scanning brochures so that you can generate leads with them.
[00:02:18.130] – Jeff
So at its peak, we were generating millions of leads for B2B technology companies per month. And I was managing both the product and also all the traffic, paid, organic, whatever, testing. And we were acquired by a major publisher in 2007 who was also our major competitor. And I stayed on and worked for them for another seven years and change, almost eight years, and grew their traffic and grew their team tremendously.
[00:02:47.710] – Kriste
Who was that? Can you say?
[00:02:49.450] – Jeff
Tech target. They’re one of the largest publishers for B2B technology, everything from healthcare – I managed huge communities for both personal electronics as well as healthcare as well as B2B tech. Any piece of traffic that came through and what happened to it, whether it was conversion rate optimization, paid or organic, our team was managing.
[00:03:11.290] – Jeff
The thing I learned, though, and the reason why this is important is because when it was just generating leads with brochureware and PDFs, you didn’t have writers, you didn’t have editorial. When it was an amazing publisher who also had those assets, it was, “oh wow, we’re working with an editorial team.” And I learned that those workflows that those writers and the subject matter experts worked with are painfully manual and very subjective. But that’s great in a lot of cases where you have these experts and artists and beautiful writers and storytellers. The standard practice for search engine optimization professionals, at the time, was doing things that would actually make those people very, very unhappy and stunt their creativity.
[00:03:59.020] – Kriste
Because you’d be like “you’ve used this word 16 times…”
[00:04:03.880] – Jeff
“… Pile of keywords,” or “we should be doing this” or “we should be doing a trick here or trick there.” But I always did it the right way. I was always looking at hyper-high quality, comprehensive content before that was a thing. And I was working on these workflows that would take forever with writers.
[00:04:22.930] – Kriste
Why would they take forever?
[00:04:24.250] – Jeff
I mean, manually researching a keyword. Manually researching a topic for comprehensiveness. Trying to figure out what it means to be an expert on a topic without the expert or interviews and it’s very subjective measures. What should we write? You ask five editors on the same beat what they should write, you get five different answers. There was no way of doing anything with data-driven reasoning in that field. And we built out manual processes that were just rough.
[00:04:54.640] – Jeff
But it worked a little bit better and a little bit better. And towards the end of my time there, I found my now co-founder… this is the punch line… Who was starting to – completely separate from me – he had built a technology that would read thousands, hundreds of thousands, millions of content items and tell you what they’re about and tell you what it meant to be an expert. And I’d been dreaming about something like this forever, and I became one of the first customers at any reasonable level of the original technology behind MarketMuse. It worked.
[00:05:29.280] – Kriste
What year was this?
[00:05:29.580] – Jeff
This was 2014.
[00:05:33.690] – Jeff
And so it worked. I mean, the first test plan I put in place, it just blew me away.
[00:05:38.700] – Kriste
Were you like, “oh, my God, this is awesome,” so excited?
[00:05:42.240] – Jeff
30 hours to four minutes just to put it in perspective. And I was like, “oh, my gosh.” So then I left TechTarget in spring of 2015. I was working for a private equity firm and my now co-founder reached out and said, “hey, Jeff, we’re really going to start taking this to market for real. Do you want to join as a late co-founder? Because you’re really the only person that knows what the heck this does.
[00:06:05.580] – Kriste
How it’s done!
[00:06:06.660] – Jeff
How it works, how it’s done, right? From practice and from the manual labor.
[00:06:10.770] – Jeff
And I was like, so I could either stick with this private equity firm or I can go jump off a cliff. And obviously I jumped off a cliff. If you know me, you know that I would jump off the cliff like nine times out of ten. And now we are the thought leader in the space of content strategy, doing it with technology. Amazing team. I’ve been able to transition to my new role as chief strategy officer where I’m looking at competitive horizon scanning, but also I’m in charge of some of our new innovations, bringing them and productizing them.
[00:06:42.010] – Jeff
But before this, I ran the product team: marketing, engineering, data science. When you’re a co-founder of a startup, you basically touch everything. Your job is to not touch everything – that’s like your transition in the majority. So that’s how we came to be.
[00:06:56.040] – Jeff
And you said what is AI-powered content research, intelligence and writing? It’s the ability to break down your content lifecycle into pieces and try to figure out which of those workflows are manual, miserable, not optimized, not efficient, and figure out the ways that you can make those things more efficient.
[00:07:18.030] – Jeff
And so in detail, we look at research, planning, prioritization, briefing, the actual writing process, editing, publishing and post publish optimization or enhancement. And everything you do at all those stages can be aided by technology, by artificial intelligence for every reason. So we try to touch each one of those workflows and build them into streams that can help people no matter where you want to start, no matter what your typical workflows are.
[00:07:52.170] – Jeff
And we turned all that into a software platform that can appeal to writers. It can appeal to research people, keyword research, search engine optimization, subject matter experts, editors or content strategists, because we want all those people to work together because that’s how you really win. And what we have, unfortunately, is teams where those teams work in silos. They have different KPIs and they do tremendously inefficient things. And so that’s where MarketMuse came from.
[00:08:22.680] – Kriste
Well, I can say from experience, there is so much inefficiency in that process. You’re right. We have some clients who I think would be all on board with what you’re talking about and others just might not get it, even though they are grossly inefficient in a lot of these areas.
[00:08:44.190] – Kriste
So do you actually work with agencies like fuoco? Would we partner with you to bring you onboard to work with us with clients and automate what can be automated?
[00:08:55.890] – Jeff
Yeah, absolutely. And what you said, it’s so common. I like to think about, we built a content marketing maturity model. You can actually look it up and you can see where your team is on that chart.
[00:09:11.790] – Jeff
By the way, just to be blunt, it hasn’t been easy to figure out how to work with agencies, but we just kind of cracked that nut over the past year. What we were doing before made agencies have to pick and choose which client they were going to test it out on and go through that and it had to be a shared thing because you were kind of paying per drink.
[00:09:34.560] – Jeff
Now what we’ve done is we made it easy for agencies to bring on their entire portfolio, their entire client list. And where we excel, our really unique, differentiated value on the planning and strategy side: you can give any agency the ability to have a scalable strategic practice or planning practice. They used to just be focused on execution or teaching that you’re going to do a yearly inventory or a yearly audit or quarterly audit where what we can bring is on demand, auditing and inventory and quick strategic insights.
[00:10:14.510] – Jeff
So, for example, we can make it so that you can deliver a monthly or weekly if you wanted, content plan, and it’s not all that painful. And you can explain why.
[00:10:27.890] – Jeff
What I would stress for agencies is we’re the why behind your content strategy, because a lot of times it’s like you’re doing this because the client told you you want to do it or you’re doing it because you’ve got a hunch. You know, we’re giving you the why in the form of data that is easy to interpret. And that’s a big difference.
[00:10:44.240] – Jeff
But yeah, we work with tons of agencies. Agencies that have portfolios of sites as well as clients. Really just focused on if you deliver content, if you deliver content strategy and not the actual content, regardless of where you sit in that, or you’re just an SEO agency that provides recommendations or updates content, these are going to be solutions that really change the way you think about what can be done with technology.
[00:11:17.450] – Jeff
A lot of things that people didn’t realize you could do, we’re doing. One of the most important ones is evaluating content quality and comprehensiveness objectively. There is a standard we have built for content quality and comprehensiveness, so there’s no reason to ever do that subjectively anymore except as an addition or amplifier.
[00:11:40.580] – Kriste
Why do you think people still are? Is it just because they don’t know you exist or – I don’t know if you’ve seen the latest Harvard Business Review, but there’s a whole section dedicated to AI-powered marketing and it talks quite a bit about the challenges and the risks. People and organizations are inherently risk-averse oftentimes. And change is hard. So I’m kind of curious about your take on that and is that at play? And then I’ll have a second part to that question, I’ll let you answer that and I have a second part to that question.
[00:12:21.130] – Jeff
Yeah, absolutely. I know what you’re referencing and it covers a lot of different wings of marketing, just like artificial intelligence is. How many branches of AI are there? Over one hundred, I believe. It covers a lot of the different elements of marketing automation – chat bots and other types of…
[00:12:42.850] – Kriste
I was looking and you’re not even on that list. They don’t even know about you.
[00:12:47.310] – Jeff
Yeah content strategy and content marketing – natural language processing is the first branch level of where we work and that’s in topic modeling. So it’s to say if I were an expert on a topic, what would that mean? So that basically says if I’m going to write an article or build a website that covers content marketing, for example, or content strategy, and it doesn’t talk about buyer personas and it doesn’t talk about target audiences, it’s not all that expert driven.
[00:13:17.560] – Jeff
So we can basically tell you whether, if you’re going to cover a particular topic, you also need other coverage of these other things to be legit. I always like to use a recipe example: if you’re going to make beer and you don’t use water, it’s not going to taste very good, right? So those are some obvious ones, but topic modeling works with that.
[00:13:40.670] – Jeff
The other angle is where because we can analyze text and analyze a lot of content and tell you what it’s about, we can also do on-demand gap analysis and things like that. The other new frontier for us is generation. So that’s where we can actually generate content or build outlines. So we build outlines for writers on the fly.
[00:14:01.540] – Kriste
So help me understand that a little bit. So instead of me sitting down and going, “OK, I think here’s my outline,” your machine learning is going to do that for you and it’s going to say here’s where to focus your efforts?
[00:14:17.050] – Jeff
Absolutely. So we can basically tell you on the fly our content brief offering – by the way our first content briefs were made manually seven years ago. We were delivering them at scale, partially assisted by our technology. But now, we’ve automated those so you can actually go in – we have a self-serve model – and build your own brief and we can basically tell you who should this content be for, what intent? What are some title considerations or headline considerations? Things that are table stakes, but also to differentiate. And that’s one reason why MarketMuse is unique.
[00:14:53.480] – Jeff
We also can give you subheaders and sections to consider, all the topics you want to consider and then internal and external link recommendations. So as you’re writing, you can build out this outline so your writer gets a candidate or inspirational meets minimum and it puts your writer, if they’re outsourced or in inside the team, it gives them a standard source of truth, basically, so that they’re never going to write something that completely is off the reservation.
[00:15:24.280] – Jeff
Then you get it and you’re like, “no, this is not what I asked for.” So it’s going to be at least this base outline and then they can add their own expertise and add their own creativity, build it into a story, build their voice into a story. And that’s wonderful. We also do, though, take it one step further and build drafts. We build draft content.
[00:15:45.460] – Kriste
What? Are you trying to put us out of business over here?
[00:15:48.140] – Jeff
Absolutely not. Absolutely not. We’re building it so that you can also have that same experience and use that as inspiration of how one might consider it and then take that and build quicker and quicker and quicker. We have teams that are able to put out five X and they’re happy with the outcomes of the content. And that’s where there’s a lot of things that you can work on.
[00:16:13.180] – Jeff
But the coolest part about it is – and this is not even covered in this piece that you’re referencing – you start here: what is your current content efficiency? How many content items do you create every N and publish? How many do you update or optimize? How frequently are those successful?
[00:16:36.510] – Jeff
If the team can’t answer that, that’s issue one. If they do, it’s probably about 10 to 15 percent. So let’s just say it’s 10 percent. That means your average cost of content is 10x what we think it is. That’s the true cost of failing content. It’s that you’re publishing 10 articles to get one to work. So you’re actually spending 10x.
[00:16:57.240] – Jeff
What if you could turn that into 40 or 50 percent success rates? That’s what our planning and prioritization solution does. It takes teams that are operating at 10 percent efficiency to 40 and also speeds them up at the same time. And so we’ve got these multiple wheels rolling. So a team that’s not ready to adopt a generation in their workflow, totally understand. That’s out here in left field. That’s a little scary, right?
[00:17:23.580] – Jeff
You may want to bring briefs in and build your briefs for your writers and at the same time improve your prioritization so that more of your at bats turn into hits. When you do that, everybody’s happy. And those are the types of wins we get and then you kind of work down the funnel. But the goal is to make subject matter experts and editors and writers the most valuable assets on a team. And that’s what we feel we do with these add ons.
[00:17:53.340] – Jeff
And it makes it so that writers don’t have to do the stuff they don’t want to do. Ask an editorial lead if they want to do keyword research manually and find me that person. Nobody wants to do that. You know, they spent 20 years working in a field to be a subject matter expert, to be a great writer. Writers want to write. Writers want to add their subjectivity, to add their expertise to things and be creative. What the technology world has done, until the last few years in this specific space, is kind of beat down their creativity. And what I’m trying to do with artificial intelligence is actually give them more space to be creative. And I feel like when we embrace that and harness that, these teams are crushing it. I mean, and they are. It’s fun.
[00:18:41.820] – Kriste
[00:18:43.470] – Kriste
So are you allowed to talk about that a little bit more when you say they’re crushing it? Like, do you have…
[00:18:49.290] – Jeff
[00:18:51.060] – Kriste
… One in one example in particular, you know, that folks would be like, oh, OK, that makes sense. Or I can go look at their content. And, you know, it all becomes crystal clear.
[00:18:59.460] – Jeff
On our site, we’ve got piles of case studies. I’m so happy and grateful that a lot of people let us do it. We also had a customer conference last month with over one thousand of our customers.
[00:19:14.880] – Jeff
We had hands-on workshops from some of our clients where they showed how they’ve been successful. How cool is that? They want to tell everyone. We work with a ton of cool examples where people are really able to – even if they’re only using the most basic use case. The most basic use case from our community is optimizing existing content.
[00:19:44.210] – Kriste
OK, tell me what that means.
[00:19:47.490] – Jeff
You’re looking at a page and MarketMuse gives you advice as to ways you could have made it more written from the lens of that which an expert would have written. So an example is we record this podcast. It’s about content strategy, right? You take the transcript, you can feed it into MarketMuse, it says we don’t talk about buyer personas for some reason. So you can add your own.
[00:20:09.670] – Kriste
Except you just did.
[00:20:10.840] – Jeff
I know. I just did. I blew it.
[00:20:14.460] – Jeff
So you can do that with your existing content, you can do that with drafts, to basically give you a sense of how to make it better from the lens of an expert. You also then can compare it against your competitors to say, what is everyone doing? That’s table stakes. If I’m going to talk about how you get bees out of your garage and I don’t talk about, you know, checking the rafters, I don’t know.
[00:20:39.540] – Jeff
But it can tell you.
[00:20:44.250] – Kriste
Have you had bees in your garage?
[00:20:45.240] – Jeff
No, I didn’t. I just was looking at a brief that a client recently wrote and it made me laugh because it was how to get bees out of your garage.
[00:20:52.680] – Kriste
I had this friend though, she did. Her house was infested with bees. The rafters, yeah. They were like they had made this whole crazy colony… anyway.
[00:21:05.100] – Jeff
No matter what level of specificity – and we also will build-out and tell you the intent behind great content on that. So is it early stage awareness? Is it comparison? Is it transactional? And so the biggest gaffe in marketing today and writing content, is you think it’s one word, one page. That’s not worked for over a decade. And the only people that still tell you that are affiliate touts that aren’t plugged into what it means to run a business anymore.
[00:21:37.740] – Jeff
What you really need is coverage across the entire buyer journey or patient journey, and you need to cover post-purchase troubleshooting, post-purchase advocacy and really give yourself full coverage on the topic. You can also give it flavors for industry, for level of expertise as well. So what we’re trying to give you is a sense of what it would mean to be an expert, but also you’ve got to cover all the variants and all the intents that match what a typical buy cycle would.
[00:22:10.110] – Jeff
With artificial intelligence, you’re giving yourself the ability to understand a concept like never before. You’re not just writing the word and it shows you other words that have the same word in it, which is what has stood for keyword research for a really long time.
[00:22:26.850] – Kriste
OK, so I’m going to ask you a question. Some people might market themselves as AI-powered, fill in the blank, because they’ve done the keyword research and that’s the buzz word and that’s, you know. But it sounds like you actually are AI-powered. Do you see that?
[00:22:47.210] – Jeff
I love this question. This is a great question.
[00:22:51.420] – Kriste
So are there the posers out there?
[00:22:55.180] – Jeff
Oh yeah, there’s posers.
[00:22:55.900] – Kriste
And how do you catch them? I’m just curious, is this how you ended up positioning your own company? You leveraged your own tools.
[00:23:11.540] – Jeff
It’s a great question. And so what I’ll say is people are confused about what artificial intelligence is. So a lot of the judgmental sense is because they don’t really understand what artificial intelligence is. There’s still people that don’t really understand. They think it means robots are doing something. They understand some branches of artificial intelligence, but not that all of them are under one umbrella.
[00:23:42.670] – Jeff
But there’s still a lot of naysayers out there that will say that’s not real AI. Well, there are components that aren’t actually doing anything related to artificial intelligence, but a lot of them are. And the naysayers are inappropriately dogging them. But think about computer vision, for example, which is being able to assess what an image is. with the computer in various applications, it’s all about the applications and what’s behind them.
[00:24:18.260] – Jeff
The thing that is what I would call “poser” today – poser, such a good word.
[00:24:23.750] – Kriste
It’s a great word.
[00:24:25.460] – Jeff
The thing that’s sort of “poser-y,” I’m not going to call anybody a poser, is going to be when – so we have our own data science organization. We have our own office for our data science team, the M4 lab in Montreal. We built everything ourselves because that’s a decision we made as a business very early on. We want to be as data-independent as we can. That’s why we have the largest consumer database that exists, we have the largest vault of data that relates to this of any company.
[00:24:57.200] – Jeff
We also built our own natural language generation platform. It’s not open AI. Open AI is our leading competitor for natural language generation. And they built a model called GPT3, which is very popular in this generation x. They also have a single reseller right to Microsoft. So yes, going up against Elon Musk and Microsoft, really good idea, Jeff.
[00:25:22.790] – Jeff
So the key thing that I would say isn’t necessarily artificial intelligence. You may be powered by it, but that’s like saying that you built a car because you drive it. Or you built a car because you’re the Uber, that’s the driver or someone else rides in it. So what we’re seeing a lot with these technology companies is they’re buying reseller rights or API access to commercially available datasets for machine learning or trained language models. In our field are train language models. So a lot of software that’s out there that will process data, they may put a beautiful interface on it that solves a user problem.
[00:26:12.920] – Jeff
So they are effectively enabling users with an AI solution, but they didn’t build it. So the risks of that are what happens if the provider of that data changes? What happens if the reseller rights change? Well, then this solution that you bought is not exactly what you thought it was. Whereas if the company has actually built it, then all the innovation is going to get to you where there’s no middle person. And that’s really the big difference that we’re seeing.
[00:26:45.740] – Jeff
So it’s like the equivalent of, you know, if you want to go use Google’s NLU, Natural Language Understanding API, you can pay for a code and build and use it. It costs a few bucks at the most. That doesn’t make you an artificial intelligence expert, it just means you know how to use software. It’s like saying that you’re a grammar and linguistics expert because you use Grammarly, and that’s not true.
[00:27:17.600] – Kriste
What’s coming to my mind as an analogy, maybe for marketers, too, and tell me if this is completely wrong and dumb. But, you know, it’s like buying access to a database, a list of names. It’s only as good as – a lot of people just resell the same data, but then some people actually have their own proprietary database and they are doing their own thing to keep that fresh. Is there an analogy there at all?
[00:27:48.320] – Jeff
Yeah, I think so. A lot of great ones. And you can provide a lot of value, right? You can provide a lot of value by looking at data and interpreting it and then delivering a creative solution. And that’s great. You can be a beautiful, useful solution provider that takes a large database and you can even apply artificial intelligence practices or machine learning practices to someone else’s data and that can be very valuable too. A great example of that is there’s a company called Mad Kudu, if you’re not familiar with them, you should be.
[00:28:28.850] – Jeff
They’re amazing. They’re a predictive lead scoring application. They’d be very relevant for the Harvard Business Review article. They can grab all of your CRM data and even open database data and build out models that look at demographic and demographic data, but also look at user behaviors on your site. And you can combine them to build models with various characteristics, the predictable quality of that lead. It’s also dramatically deadly accurate.
[00:29:06.800] – Kriste
[00:29:08.030] – Jeff
Yeah. And so it can actually automate and give you an aggregated lead score. I’ve been trying to do a lead score for my entire career. I delivered millions of leads in my career. I always thought that was something someone could do. I always thought it was going to be an if-then or grade each segment component. So you get a lead that’s got these components of this, these components of this, and the true test of artificial intelligence from a training model perspective is when it tells you something that you wouldn’t have guessed with your rule-based architecture.
[00:29:43.610] – Jeff
So think about that. It told me that this person from this company was actually a good lead. But if I had used my rule architecture, it would have told me it was a bad lead. Once it does that and that is reliable, you know you’ve eclipsed what you could have done with your own brain. That’s a great example. That’s not their data, they built the technology that allows you to predict better than you could on your own.
[00:30:13.760] – Jeff
And so that’s great. I don’t mind that case. The cases where you’re not adding anything is where it’s like, this is something that you can buy off the shelf. You’re effectively a reseller. And that’s a little bit different.
[00:30:31.520] – Kriste
[00:30:32.307] – Jeff
[00:30:34.190] – Kriste
I’m going to check it out. We’re going to include that in the show notes too for people.
[00:30:38.840] – Kriste
This is great. This is opening up whole new worlds and my mind is just racing right now. Let me ask a very technical question back to that HBR article. They frame-up – I’m going to assume this is accurate and true – the four kinds of marketing AI. And they talk about four quadrants: standalone task automation apps, standalone machine learning apps, integrated task automation apps and integrated machine learning apps.
[00:31:10.730] – Kriste
I feel like that’s exactly what you’ve been talking about just now. Where would you guys, where would MarketMuse fit into there, and do you think that that’s an accurate and good way to look at it?
[00:31:23.270] – Jeff
I think it is. And the challenge being that there are hybrids and things where the models used to automate tasks were created with machine learning components. My background is in product management, so I always like to look at these from the level of product market maturity. That’s the key, I believe, to understanding AI and its likelihood of yielding value for your business.
[00:32:00.440] – Jeff
First of all, if you’re familiar with Paul Roetzer from the Marketing Artificial Intelligence Institute, one of his key things is: pick something that’s relatively mature for a workflow and run a pilot. Best way to get started.
[00:32:17.420] – Kriste
Great advice. We’re going to put that in the show notes too.
[00:32:20.510] – Jeff
For sure. And if you haven’t looked at the Marketing Artificial Intelligence Institute, go check it out. I have a course on natural language generation that you can grab too.
[00:32:31.220] – Jeff
So that’s my key. It’s not necessarily about “is it a standalone application, or is it integrated behind the scenes, or integrated in a way that as a user, you understand,” but it’s about how mature that workflow is and how understood it is. And the example I always like to use for us is the first time you ever used a spell checker, it probably was alien and it seemed like it couldn’t possibly work and it kind of was OK.
[00:33:01.010] – Jeff
But now spell checkers are just ingrained, like when would you ever write something that didn’t have a spellchecker on it, right? And then we went through the same thing with grammar and with tone of voice. The first time you saw that, it was kind of clunky and it was like, “is this ever really going to work?” Now, it’s like, oh my gosh. Like I can almost do ‘accept all’ on all of these.
[00:33:21.290] – Kriste
Except for auto-correct somehow never seems to get it right.
[00:33:24.890] – Jeff
For some reason, auto-correct always thinks that I always want fir trees, F-I-R, and I never have ever wanted anything that was ever spelled F-I-R when I write ‘for’. That’s an example of the other side of it. But it’s also like Alexa, right? When you first used that, it was terrible. I mean, it was terrible. Terrible. Now they’re right most of the time.
[00:33:47.060] – Kriste
Listen, Siri is much more reliable than Alexa for me. And I think it’s because I have Siri on my iPhone. So she’s heard my voice a lot more… Just saying.
[00:33:56.960] – Jeff
I mean, my kids can control the Alexa and they’re two and a half and four and a half and that blows my mind. They’re like, Alexa, play super wings. And it just does, right?
[00:34:10.520] – Jeff
So really it’s about the maturity level. So now think about the ones I was just talking about. How comfortable are you that there’s technology that can tell you how to have written something as if you were an expert? When we first started, no one was comfortable with that. Now, most people are. Most people are cool with something, giving them recommendations for enhancement and optimization. Are you comfortable getting draft content written yet? Nope. Nobody is. Nobody is.
[00:34:40.490] – Kriste
OK, that was going to be my next question. Keep going.
[00:34:43.400] – Jeff
No, the first versions of what we delivered was in Google Docs and they were finished content. And people would look at them and they’re like, “nope.” People’s brains are human brains, they don’t do well with unnatural experiences. They don’t do well with net, new, novel, immature experiences. So they looked at it, and their first thought was, “I need to judge this to see if it’s good or bad,” because that’s what people do.
[00:35:16.550] – Jeff
When you land on a brand new website you’ve never seen before, you judge whether it’s good, very quickly and you leave, you decide. Over 80 percent, I think, is the mark of that metric where you decide whether a new site you’ve never been to is hit or miss within the first two seconds or something like that. It was a study done by Content Science by Colleen Jones. She does it every year. It gets worse and worse every year. We become more judgmental every year.
[00:35:42.870] – Jeff
So if you read this content, you immediately judge it. You’re not naturally thinking that you should use this to enhance and speed up your content creation workflow. That’s not what you’re thinking. You’re thinking: judge it. And so over time, that’s basically the journey for every workflow that uses AI. Predictive lead scoring – great example. Maybe you look at it you’re like, “oh, no, this is stupid because it surfaced this lead that I don’t even think we sell to this demo.”
[00:36:18.703] – Kriste
And you’re like “well they should.”
[00:36:20.310] – Jeff
But maybe you should and here’s why. And here’s the model as it was trained. And so that’s the journey. Everyone goes on, they’re judging good or bad, and then all of a sudden it flows and flows and flows. And then you just make the –
[00:36:37.830] – Kriste
All of a sudden there’s a shift.
[00:36:39.210] – Jeff
You just think, “oh wow, it actually works better than I or a team of my making or building could possibly do.” And then you change. And that’s the experience.
[00:36:52.460] – Jeff
Self-driving cars, a great example. Still on the edge. “Hey, this isn’t ever going to work. Oh, what about this? What about that?” Obviously you have ethical implications in all of these. Ethical ethics in AI, another topic for another day, I think.
[00:37:08.970] – Jeff
But think about it. When it first came out, not comfortable, not comfortable, not comfortable, not comfortable. Are you going to be comfortable probably three years from now with it? Probably. Why? Enough data is provided. It gets better. There’s less percentage of failure rates, there are ethics discussions. And then you’re like, “OK, I can be OK with this.” Computer vision is a great example. Captures. You go into all these, you can get an example. All of them go through this maturity model.
[00:37:47.940] – Kriste
It really makes me think of two things. One: yesterday I follow Stoic on Twitter, and it was like, you don’t have to judge. You don’t have to. But our natural inclination as humans is we think we have to have an opinion about something, even if we don’t have enough information to actually have one yet, and that it’s actually OK not to judge until you know more.
[00:38:13.770] – Kriste
So that was number one it made me think of and this is a great example. And then number two, since I’m in healthcare marketing and healthcare, as you probably well know, is notoriously risk-averse, slower to adopt new technologies and emerging technologies, and there’s a ton of conversation out there in healthcare about, you know, I think there’s a fear factor around artificial intelligence and machine learning. And mostly it is, “they’re going to take my job away. That’s going to take my job away. I won’t have a job.” But the reality is, to your point, no, it’s just going to free you up to do the job you actually have the training to do rather than all this kind of stuff. Like my reaction earlier, “what are you trying to put us out of a job?” No, we’re trying to help you be better at what you’re really good at.
[00:39:06.620] – Jeff
In healthcare specifically, there’s reasons to be concerned. I work extensively in healthcare, whether its Pharma, life sciences or traditional marketing doctors, marketing hospitals, as well as drugs and such. There is a reason to because abuse of generation technology could impact one’s life. I mean, that is absolutely and positively a risk. Generated content is not fact-checked by default. So proper nouns, numbers, people’s names, things like that are not guaranteed to be right.
[00:39:48.410] – Jeff
And that is where the power is in the user. You, though, your business is responsible for words you put on the page, just like if it was an outsourced writer, would you ever take something in healthcare from an outsourced professional and not have it proofed, not have it fact-checked? So why would you do it with AI?
[00:40:06.680] – Jeff
And so it’s the same logic path. But what it can do, it can tell you unbranded patient journey research. I need awareness. I need unknown need-aware, I need known need-aware, I need pre-purchase comparison, I need treatment diagnoses comparison, I need post-purchase adherence content. So it can actually go through that journey and tell you what intents you’re targeting. Oh yeah and by the way, we have two personas: we’re shooting for patients here, we got health care professionals over here. It’s two completely different content strategies.
[00:40:46.580] – Jeff
And so it can actually build out models that can build the right thing for a completely novice person who just has a particular ailment versus someone who’s trying to make the decision between drug A and drug B, which are two completely different things but they get smushed together in traditional keyword research. And that’s why healthcare workers have it so bad, because it’s so glon together. You’ve got to separate a lot of these things. And so AI can help those types of really tough things and make healthcare marketers smarter.
[00:41:21.740] – Jeff
Mike Sweeney, RightSource Marketing is a great example of someone who does this well and I feel like really gets it.
[00:41:33.320] – Kriste
Well, how do we help more people get there?
[00:41:35.120] – Jeff
That’s a great question.
[00:41:36.970] – Kriste
You had a great suggestion. Do a pilot, right? Just do a pilot. And that probably converts people to believers pretty quickly for you.
[00:41:49.280] – Jeff
Yeah, I think it’s about knowing your workflow. The way that I like to do it at the strategic end is really with that efficiency discussion. How much content are we making? How much content are we updating and optimizing? How frequently are we doing inventories and audits? First of all, inventories and audits should be an always-on thing. And then of the stuff we publish, how frequently are we successful, and are we happy with that? Are we happy with how much we’re getting out the door? Is there stuff that just keeps dragging on? Are we able to do these things surgically as far as like, what happens if something changes in the market? Does it throw our entire plan out the window?
[00:42:36.170] – Jeff
So you’re getting into these kinds of conversations. Do you really know the true cost of content? Do you have risks? Every plan falls into one of four categories. In most cases, you got quick wins: what are the things that I could write today that will have the biggest impact on the business or update today? I might have topics like I want to own this topic that might be more my goal-based. Risks where I’ve got competitors breathing down my neck.
[00:43:01.850] – Jeff
Or I’ve got a page – this is where everybody gets really red in the face – I’ve got a page that gets far too high of a percentage of my traffic on my site and I don’t know how to support that. I’m afraid to touch it and I don’t know what to do about it but I know it’s a huge risk to my business because what if that gets cut in half? What do I do? These are the types of things that we run into. No matter where you are, the way you change is to know why you plan today, what your efficiencies are and what maybe you’d like them to be. And then pick one workflow to improve or do the easy one. Go update five articles.
[00:43:41.450] – Kriste
OK talk to me about that, because I’m like, OK, this is a no-brainer. But so many clients don’t do it or they don’t want you to focus there. Talk about why that’s the easiest place to start and the best place.
[00:43:56.540] – Jeff
Updating existing content for – finding multiple reasons to update: is it out of date? Do you have gaps? A few gaps, one would be based on comprehensiveness – it wasn’t as good as it should have been. The other one is a fun one. This is an easy one to train a client how to do.
[00:44:19.250] – Kriste
[00:44:20.180] – Jeff
It’s called intent mismatch.
[00:44:22.010] – Kriste
[00:44:23.630] – Jeff
Everyone understands this.
[00:44:25.970] – Kriste
Oh my God. I’m learning so much from you today, Jeff.
[00:44:28.400] – Jeff
Here’s the one.
[00:44:29.910] – Kriste
[00:44:30.710] – Jeff
You have a page. It ranks for something. The page doesn’t follow through on that promise. Every brand understands they do not want that to happen. So you’ve got a page. It’s about desk lamps. And when someone searches for “how do you screw in a light bulb,” they get to your desk lamp page. It doesn’t tell them how to screw in the light bulb. They get to your brand. Their first experience with your brand is disappointment.
[00:45:01.180] – Jeff
That page needs to be updated to show how to screw in a light bulb or point to a link to a page that tells you how to screw in a light bulb. Intent mismatch is the easiest way to sell an executive on updating existing content because it’s people who are going to your site and leaving unsatisfied. And that’s the thing that everyone who works at your company and executive leadership should never want to happen. It’s like walking in the front door of a Porsche dealership and there’s no cars. And no one would ever want that.
[00:45:37.240] – Kriste
Is the frustration there mostly they can’t figure out why it’s happening in the first place or they know why it’s happening, but they don’t know how to address it? Or what do you find?
[00:45:46.030] – Jeff
I have a need. I found this page that, for one reason or another – I can get into all the reasons why this happens – it doesn’t answer my specific need today. And so I leave and I go find a competitor or someone else…
[00:46:01.720] – Kriste
That does tell me that. Maybe it’s another desk lamp, but they have a payoff to…
[00:46:06.760] – Jeff
Actually specifically answer this question. The other version is where your page, when you landed on it, isn’t satisfactorily exhibiting expertise.
[00:46:20.230] – Jeff
So in other cases where you can show someone’s journey, they search for this term and they land on this page. “Hey, are you comfortable that this page represents our entire team’s expertise? No? All right, let’s fix it, let’s make it so that it does exhibit our expertise.” And that’s why I think a lot of CEOs of the past are really losing their cachet is because they for a long time have been telling people to build pages that the entire team isn’t proud of.
[00:46:59.560] – Jeff
We’ve got to build pages that do both, and make everybody proud. Let’s say you do win: “hey, we’ve got a great ranking!” And then the page stinks. “Oh wonderful we’re ranking number 10, five, three for this!” And then someone gets to the page and it doesn’t satisfactorily represent my business’ expertise. Well, now, did we just do something good or bad? Well, if we don’t update that page, we’re in trouble because we’re just not doing right by our business.
[00:47:29.380] – Kriste
And so optimizing existing content for intent mismatch or low quality or thin content are two of the easiest workflows to introduce to a company who hasn’t historically been comfortable with any of these workflows.
[00:47:45.370] – Kriste
[00:47:46.420] – Jeff
And MarketMuse tells you exactly what pages to update, tells you the intents to focus on that are mismatched, gives you all this data.
[00:47:55.960] – Kriste
So this is MarketMuse. This is one piece of MarketMuse.
[00:48:01.390] – Jeff
One piece of MarketMuse. You can go in, it’ll tell you the top five pages you should optimize today to have the biggest business impact. It can also then build the outlines for new pages you want to create, tell you how to optimize the existing ones, tell you where you got missing internal links, what questions should you answer elegantly in this piece, really building out that so that you don’t have to do that research and it’s all there and you can just use what you got in your brain to amplify and accelerate.
[00:48:32.830] – Kriste
[00:48:34.570] – Kriste
OK, say I want to start working with you tomorrow, which I do. What do I need to do? Talk to me about how folks can find you, get engaged with MarketMuse, all the good things. How do they reach you?
[00:48:50.170] – Jeff
I’m on Twitter, Jeffrey_Coyle. LinkedIn, I respond to everything. Jeff@marketmuse.com. We have self-service offerings you can just buy on the site if you don’t want to talk to anybody. If you are an agency or portfolio owner or mid-market enterprise, which is our historical focus but we do now have a walk-in-the-door for one person, soul openers and single person practices we have a self serve. Shoot a form in, shoot a note to me specifically tell me what the details of the agency are or whomever.
[00:49:26.950] – Speaker 2
Build out some examples and make it make sense. Our mission is to set the standard for content quality, that’s the actual party line. But I like to say it’s to rid the world of bad content. So that’s the shadow mission, is to rid the world of bad content. I can get behind both of those.
[00:49:53.320] – Jeff
Give me a shout at firstname.lastname@example.org or on Twitter.
[00:49:56.800] – Kriste
Awesome. And then you’re out there on the Speaking Circuit podcast. Tell us maybe a few places that you’re going to be or even some of your favorite places to go for inspiration or podcasts you listen to. Any or all of those things.
[00:50:11.470] – Jeff
Awesome. Well, I mentioned a few already.
[00:50:15.220] – Kriste
Yes, thank you for that.
[00:50:16.360] – Jeff
But we have a contest strategy webinar series that’s integrated into our community called the Content Strategy Collective. If you shoot me a note, we can get you an invite to that community. We have two webinars per month in that community that really focus on specific content strategies. Some of the ones I’m looking forward to actually traveling again in this world, I’ll be at Pub Con in Miami in the beginning of August. Also, I’m a big advocate of the Rhodium community of entrepreneurs. Go check that out on Facebook.
[00:50:49.720] – Kriste
[00:50:53.140] – Jeff
Yeah, like the metal. R-H-O-D-I-U-M. Go check them out on Facebook. They also have an event in September I will be at and speaking and then Content Marketing World, which back in Cleveland, live. If you haven’t booked that, please do.
[00:51:09.130] – Kriste
I’ve never been to Cleveland.
[00:51:11.860] – Jeff
It’s cooler than people think.
[00:51:14.140] – Kriste
That’s what I’ve heard.
[00:51:15.070] – Jeff
I like it a lot. I surprisingly like it a lot.
[00:51:19.540] – Kriste
When is that?
[00:51:19.540] – Jeff
I want to say it’s the last days of September fading into the first part of October. But yeah, I’ll be rocking orange sunglasses on stage like everybody else there. So, yeah, come on down to that. It’ll be a good old time.
[00:51:36.270] – Kriste
[00:51:37.530] – Kriste
OK, well, thank you so much. I’ve learned so much today, I really appreciate you being a guest on how it’s done. That really wraps it up for today. Unless you have something else you want to add or share. Anything we have failed to cover today, Jeff, otherwise.
[00:51:53.160] – Jeff
No, I think those are great examples. Go check out our blog as well, which has a lot of content strategy concepts. blog.marketmuse.com
[00:52:11.370] – Kriste
Awesome. OK, thanks so much. Our guest today has been Jeff Coyle, who is co-founder and chief strategy officer of MarketMus, which offers AI-powered content research, intelligence and writing. Thanks a lot, Jeff.
[00:52:27.390] – Jeff
[00:52:28.260] – Kriste
That’s it for now. Thanks so much for listening. We’re looking forward to keeping great conversations coming your way as we grow this podcast. There’s even more great content from our conversations on our blog. Be sure to check it out at growwithfuoco.com. That’s grow with fuoco, F-U-O-C-O, dot com.
[00:52:51.840] – Kriste
Stay tuned until next time and no matter what, stay curious.
- Harvard Business Journal: How to Design an AI Marketing Strategy
- People, companies and tools to know:
- MarketMuse website
- Find Jeff on:
- Email Jeff: email@example.com
- Find Kriste on:
More From The How It's Done Podcast
Want to be a guest on a future episode?
Share your thoughts directly with our show host, Kriste Goad.