Episode Overview: Optimizing hundreds to thousands of pages of content may seem an insurmountable task if you’re an enterprise-level business. Keeping content fresh site wide is key to maximizing your returns on your various content pieces. Developments in AI are the key to simplifying the content optimization process. Join host Ben as he speaks with Acrolinx CEO Volker Smid about why AI is the optimal solution to implement content optimization strategies.
- It’s best to apply AI solutions as you begin the content production process and not to published archives. It saves valuable time and eliminates missed optimization opportunities.
- When applying AI as a solution enterprise-scale businesses need to ensure it’s applied company-wide.
- Support site content is best supported by AI, such as user ratings and comments, as the data acquired by users can help retool the page.
GUESTS & RESOURCES
- Volker Smid: Website // LinkedIn
- The Voices of Search Podcast: Email // LinkedIn // Twitter
- Benjamin Shapiro: Website // LinkedIn // Twitter
Ben: Welcome to the Voices of Search podcast. I’m your host, Benjamin Shapiro, and today we’re going to talk about content optimization technologies. Joining us is an old friend, Volker Smid, who is the CEO of Acrolinx, which is an AI platform that uses a unique linguistics analytics engine to read all your content and provide immediate guidance to improve it. The Acrolinx technology is transforming how the world’s biggest brands create high performing content. Yesterday, Volker and I talked about when is the time to prioritize content optimization. And today we’re going to talk about why AI is the right solution for enterprise brands to optimize their content. Okay. On with the show. Here’s the second part of my conversation with Volker Smid CEO of Acrolinx. Volker, welcome back to the Voices of Search podcast.
Volker: Thank you, Ben. Good to be back.
Ben: Yeah, well hey, always a pleasure, It took us long enough. He actually had to sponsor the podcast, leave and now we’re having you on, so we work in a backwards fashion here on this podcast. What can I tell you? Yesterday we talked about the right time to prioritize your content and you know, for the enterprise brands, it seems that content optimization fundamentally is more important than content production, where most of the time you have this gigantic archive of content and making sure that that is performance and getting the visitors that you’re reaching or the time on site up is probably more important than net new content generation. Obviously it’s different for every brand, but for the large enterprise brands with huge content basis. Talk to me about why artificial intelligence is the right solution to be able to focus and implement content optimization strategies.
Volker: I think this is a great question, but long before I talk about artificial intelligence, let’s start with human intelligence. Sometimes overload.
Ben: I’m a podcast host. I don’t understand what that is.
Volker: I think we’ve all seen in the last two decades, a rapid evolvement of the existing of content. I think today, every day we enjoy it’s 3.7 billion Google searches every day. That’s a lot of searches. They search for content.
Volker: A large enterprise Fortune 500 company probably hosts about one to 1.5 million pages on their website. And if you ask the enterprise, okay, how many content repositories do you have? The novel responses? We have a content management system that is Roadtrip, Adobe what have you? The reality is that probably in every enterprise you will find minimum 30 to 50, maybe sometimes 100 content repositories. And that’s the one that, you know, that exists.
Ben: So before you go on, talk to me about what you mean by content repository. Is this just content that lives in a different section of the site or something that is being managed by a different tool.
Volker: It sits there for a different purpose. A good example is that for technical documentation, which is part of the user experience, a good example of a content repository is don’t be surprised GitHub. So if you have a software company producing software and creating documentation for that software, that documentation sits in GitHub, by the way, a very good content repository. And so you have GitHub have them connect to, to a content management system. You have SharePoint, you have Google Drive that hosts a large amount of content. They all have ties into the content management system for sure, but they all exist on more on a departmental level. They are managed on a departmental level and they are then finally connected by other content management systems. But the work needs to be done on the basis of an existing content repository.
Ben: Okay. So by content repository, you don’t mean the end content that is being published on the site. You mean all of the different tools that are being used to create and manage content before they get to the content management system?
Volker: The publishing is only define an animal. So number one, you need to create a workflow in your company that says, we need to comprehensively understand where we host content. Then if you subscribe to a platform that helps you to optimize them to govern content, then AI comes into play. But if you only apply AI to what exists on the publishing side, he would probably miss the million mark.
Ben: It seems like the solution for integrating AI into the content production, not necessarily the published content archive, is really about making sure that your content is optimized before it’s published. Am I making a bad assumption or?
Volker: Yeah. And it’s a complex workflow. As I said yesterday, sometimes you have an agency creating content, then it’s being shipped. Normally it has an xFi and a ton of stuff inside the xFi. Then it’s being scrapped sometimes to have a workflow that even includes legal before something can even be published. And in all of these steps, content optimization is one factor to make that step more efficient and the outputs more performing. So it’s not enough to just apply what you have to the existing site. You need to catch the piece of content in every step of your workflow. And then I think AI plays a vital role because we are talking enterprise and we are talking massive scale. If you want to establish this quality score for 1.5 million pages that exist in, let’s say 100 different repositories actually was the application of AI. That’s virtually impossible.
Ben: So, the standard enterprise has 30 different systems that are producing content. Hopefully all being fed into one content management system, everyone’s systems and platforms and processes are going to be a little different. But on average, we’re looking at a fair amount of different sources to produce content. Talk to me about how Acrolinx and you know, other AI systems that evaluate content work, how are they able to understand what is going to be seen as effective and valuable content to the end consumers?
Volker: Yeah, I think the most obvious application of AI has probably little to do with our callings. It has a lot to do with say Microsoft and Google, and this is on the fly grammar, correct? Let’s not forget that a massive amount of content is being produced from non-native speaking people. So you always have a grammar issue. So all the correct when you rise has almost become a commodity. When you think about Microsoft and Google applying these services in depth, that stops. But then there’s another factor which I would call the comprehension of content. This is not about a single row or two words in a row. This is about taking the entire content and find out whether the content is actually scannable.
Ben: So what do you mean by scannable?
Volker: Scannable means that you and I, human beings, we go where we find a piece of content that normally we would like to get to the bottom of this thing in less than five to 10 seconds, we don’t read 15 pages.
Volker: So can somebody, a consumer of the content scan and less than 10, 15 seconds, the major output of this content? Yes or no. So is this scannable, is it readable or do you have highly complicated nested sentences in your content that makes it very hard to comprehend? Third aspect is the content emotional or does it have to be emotional. You have to set a target. This content needs to be emotional. It is emotional. Yes or no.
Volker: Tone of voice.
Volker: This has less to do with AI, but are you avoiding terms that are blacklisted, right? If you think about an enterprise was 1.5 million pages. I actually assure that all of your pages avoid all the black lists of terms that you have created as an enterprise. Have you established a terminology for your enterprise? A terminology that always uses the correct product name in all content that you create.
Volker: So, and again was out the application of AI models that have been trained over years. All of that is virtually impossible, but it all starts with number one, you need to do this comprehensively as an enterprise, not in isolation in that department. Number two, you need to think about the value of your content. If you have 1.5 million pages and the production cost of each page is about, let’s say $1,000 to make the math easy, your investment in that estate was what?
Ben: $1.5 billion.
Volker: $1.5 billion, and you have invested your time to create that content. Now, does it perform? Does it speak the right language? Does it hit the right tone? Is it scannable? Is it readable? Is it emotional? All of these factors make that investment more variable.
Ben: Is there a way that AI, Acrolinx or other tools can evaluate the effectiveness of a piece of content, whether it provides the desired outcome or answer to the consumer. And I’ll preface this with in a lot of help content. You’ll see a, you know, I got to this page, was this page helpful or not? And there’s a smiley face and a sad face. And hopefully the consumer clicks one most of the time. It’s the negative one. Is there any sense of evaluating the content effectiveness?
Volker: Yes. Depends on the purpose of the content. Technical content is normally connected to NPS. NPS is connected to retention. On other words, churn that content has a number of output factors that we know very well. Number one, if you have bad technical content, the number of calls that you will get on your hotline on your physical hotline increases and each call costs, if it’s between $13 and $68. So optimized content helps you to reduce the call rate into your call center. It’s a direct relationship between an economic output of that effort. Number two, if you do NPS relation and MPS that content normally leads to bad scores, normally leads to high churn or low retention. So the connection over time that we have accumulated and the knowledge that we have gathered is there’s a severe connection between the quality of technical content called effects and NPS related to churn.
Volker: On marketing, a lot of the content it has to perform. You have a piece of content, the consumer finds the content. That’s not enough. You want this content to convert for the next level, whether this is a capturing of lead data or whether this is a direct transaction, doesn’t matter. But content that does not act is again a waste. So it really depends on what you’re trying to achieve, right? And sometimes as I said, content needs to be looked at in a way that avoids risk on enterprise. If you’re a bank, an insurance company, or a life science company, your view on content needs to be something that is in the mix between performance, NPS calls but also the avoidance of risk.
Ben: It seems like most brands think of content as something that once it’s produced it’s finished. There’s the notion of ongoing content optimization. How much does technology play a role in understanding what content needs to be optimized or are you just purely looking at the KPIs that you laid out, if NPS is low, then we need to go back and look at our help desk content to make sure it’s effective. If marketing conversions are low, we need to go back and look at our lead generation content, are brands using technology to figure out what content to optimize? Or they just waiting for the trailing signals.
Volker: Yeah, depending on the enterprise to be honest, let me respond in a different way. The first thing that we do Acrolinx when we come in is we establish the quality scores. So let’s say you have half a million pages and you establish a threshold of, let’s say 80 from a quality score perspective. And you find out that you have 5% of your content at a score below 10, and you have 15% of your content at a score between 75 and 80. My recommendation would always be go for the bottom where you have the lowest scores, change the content, optimize it, and try to get it as fast as possible to score 50 to 80.
Ben: That’s funny. I thought it was going to be the opposite. I thought it would be.
Ben: Go to the bottom, press control A and delete, and then focus on the content that is almost to the threshold you want it to, and they’re going to find faster results there. You’re saying go to the broken content first.
Volker: Yeah. Yes. Thank you. It was time to look more broken content than in the bucket of 75 to 80% already.
Volker: And so it becomes a task of a, let’s say several months, quarters and years to get the bottom up to something that is appreciated and performs a lot better and avoids risk.
Ben: Okay. So last question for you. Having gone from working on an SEO focused company at Searchmetrics now focusing on content optimization, you’ve really seen both sides of the technology platforms that helps not only optimize content as after it’s produced, but also while it’s being produced. When you think about the combination of content, production and SEO and how those technologies are becoming more advanced, what are you most excited about? You know, in terms of the collaboration between SEO and content production teams?
Volker: I honestly believe the combination of two platforms is the simplest form. They are one plus one equals three. If you can get all the data power into the hands of a content editor, when he or she creates content, the leverage is enormous. If you wait for the content to be produced and published and then find out it’s not performing, this cycle time that you need the extra money that you have to spend, the fact that you are not getting to the revenue as fast as possible is so big that my intent would always be, get the data into the hands of the editors as fast as possible.
Volker: So this game can only be one if we see tight integration between platforms to increase the level, because nobody will log into 15 different platforms that have similar to optimize content, it needs to be integrated into the platform that the editor uses when he or she creates content. And that integration is probably the biggest liability or the biggest opportunity of the market of say, the Searchmetrics, the percolates, the composts, the actual links, all of them need to work together to create a platform that creates more value. Because the scariest thing that I had to keep looking at is what’s the name had been, you know.
Ben: Scott Brinker MarTech, 5,000 list which is now a good 7,500 companies wide.
Volker: Yeah. Yeah. Okay. So now you’ve put yourself into the shoes of a CML. He or she looks at the list of 7,500 companies and thinks, “Where do I begin?”
Ben: Just throw darts.
Volker: Haha, yeah.
Ben: Well, Volker, I appreciate you coming on and talking to us about the importance of content optimization a little bit about how technology plays into that. And I totally agree with your last sentiment, which is content optimization starts in the content production process, not after the content publishing process. And you know, if you’re doing your homework and, from the SEO perspective, you’re realizing what are going to be the most effective topics to write about in your bringing in data during the content production process, you’re going to be more successful and you’re not going to have to redo your work down the road. And that wraps up this episode of the Voices of Search podcast.
Ben: Thanks for listening to my conversation with Volker Smid, the CEO of Acrolinx, we’d love to continue the conversation with you. So if you’re interested in contacting Volker, you can find a link to his LinkedIn profile in our show notes, you can contact him on Twitter. His handle is Smid Volker, S-M-I-D-V-O-L-K-E-R. Or you can visit his company’s website, which is acrolinx.com, A-C-R-O-L-I-N-X.com. Just one more link in our show notes. I’d like to tell you about, if you didn’t have a chance to take notes while you were listening to this podcast, head over to voicesofsearch.com, where we have summaries of all of our episodes and contact information for our guests.
Ben: You can also send us your topic, suggestions, your SEO questions. You can even apply to be a guest speaker on the Voices of Search podcast. Of course you can always reach out on social media. Our handle is voicesofsearch on Twitter. And my personal handle is Ben B-E-N-J-S-H-A-P. And if you haven’t subscribed yet, and you want a daily stream of SEO and content marketing insights in your podcast feed, we’re going to publish episodes every day during the work week. So hit the subscribe button in your podcast app, and we’ll be back in your feed tomorrow morning. All right. That’s it for today, but until next time, remember the answers are always in the data.