Podcasts > Ep. 122 - Optimize content creation with AI
Ep. 122
Optimize content creation with AI
Christopher Willis, Chief Marketing Officer, Acrolinx
Wednesday, April 13, 2022

In this episode, we discuss how to integrate an AI engine with content tools to provide real time advice on brand, tone, clarity, compliance and consistency. We also explore how to build efficient processes to work around the limits of AI systems when it comes to automating mass updates across thousands of documents. 

Our guest today is Christopher Willis, Chief Marketing Officer of Acrolinx. Acrolinx is an AI platform with a unique linguistic and analytics engine to read all your content and provide immediate guidance to improve it.

IoT ONE is an IoT focused research and advisory firm. We provide research to enable you to grow in the digital age. Our services include market research, competitor information, customer research, market entry, partner scouting, and innovation programs. For more information, please visit iotone.com


Welcome to the Industrial IoT Spotlight, your number one spot for insight from industrial IoT thought leaders who are transforming businesses today with your host, Erik Walenza.

Erik: Welcome back to the Industrial IoT Spotlight podcast. I'm your host, Erik Walenza, CEO of IoT ONE, the consultancy that helps companies create value from data to accelerate growth. And our guest today is Christopher Willis, Chief Marketing Officer of Acrolinx. Acrolinx is an AI platform with a unique linguistic and analytics engine to read all your content and provide immediate guidance to improve it. In this talk, we discussed how to integrate an AI engine with content tools to provide real time advice on brand, tone, clarity, compliance and consistency. We also explored how to build efficient processes to work around the limits of AI systems when it comes to automating mass updates across thousands of documents.

If you find these conversations valuable, please leave us a comment and a five-star review. And if you'd like to share your company's story or recommend a speaker, please email us at team@IoTone.com. Finally, if you have an IoT research, strategy, or training initiative that you'd like to discuss, you can email me directly at erik.walenza@IoTone.com. Thank you. Chris, thank you so much for joining us on the podcast today.

Chris: Erik, thanks for having me.

Erik: But before we get into the business, would love to learn more about your background? I know you have really a deep background working with both larger corporates and tech companies and also younger tech companies in the space. Can you just give us a quick walk through of where you first started to engage with the topic and then how you ended up now at Acrolinx?

Chris: Sure. So I started out of college in an incubator, helping to build small technology companies, grow them to the point of going public, and then moving on to the next. And that actually accelerated me into large company life. I spent two years living in the Netherlands working at KPMG, building the early internet applications at the enterprise level for companies like Rabobank, and [inaudible 02:49] here bringing these companies onto the internet.

So this concept of revolution of new things happening, and that's, for whatever reason, seems to be where I've spent a lot of my time. When I left that company moved back to the United States, we started a company that ended up being the first packaged mobile application company. Pixel mobile was very early in packaged mobile applications.

We built lifestyle applications for, for instance, mutual fund wholesalers. It wasn't a small screen view of big screen data. It wasn't taking Siebel and making it accessible on a Palm Pilot. It was collecting data from all different sources. It was really early composite application that ran on Palm Pilot. Then on Windows CE, then we move to this thing called a Blackberry that looked like it might take off. From there, iPhone and on to Android, while we were doing that building a platform that allowed for that multi-operating system development process.

From there, I went on to a company called Perfecto Mobile that was in mobile cloud testing, helping big companies, banks, insurance companies, manufacturers test their mobile applications in the cloud on physical devices. And when I was there, brilliant company, incredibly smart people from all over the world, my best writers were the product marketing people who, again, very technical, understood our product, understood domain, but wrote in English as a second language.

And the editorial cost for that was very, very high. I couldn't easily edit their content because they're writing technology. And so if I don't understand what they're trying to say from a domain standpoint, from a technology standpoint, changing their writing to be more correct, more high quality is super difficult. And I ran into this company Acrolinx that made a claim that I would be able to push a lot of that editorial back to the first draft to the original writer. And when that content leaves their desk, it would be ready for contextual review.

And as it turned out, I didn't buy the software, I left Perfecto and joined Acrolinx as CMO, because I think that what this company does solves a very distinct problem that I had in my world. Since then, I've dramatically expanded my understanding of what Acrolinx does and we've driven the product forward. But I was drawn here because I wanted to use this. That's different than most people moving to a new software company. But you don't go to SAP because you're like, I can't wait to use SAP ERP rules. But people tend to come to this company because they're excited about what we do and about the use of the product.

Erik: I think my company is probably too small to be an interesting customer at all of Acrolinx. But I can already see the use that we could have because we're in this the same challenge right now of producing research that's quite technical, and then having to convert that into written content to share through channels that works in different channels. And that conversion it's very difficult because we have one group of people that which are basically doing structured research and then another group of people that are writing articles, and the skill sets are completely different in that conversion. So that's at least where I'm imagining right now you fit in it's supporting that conversion.

But before we get maybe too deep into Acrolinx, can you help give just a high level mind map for what the content space looks like? Because I imagine that a lot of people as soon as they hear content, and then software company, they probably have 10 different companies with different business models popping into their minds, everything from SEO optimization and etc. Can you just help a frame where does Acrolinx sit within that larger ecosystem?

Chris: Sure. So it's interesting a couple years ago, we won an award for being the best content marketing platform. And I was sitting at the table at the award ceremony with the CEO of [inaudible 07:09], which is in fact, a content marketing platform. We are not. So as they were getting ready to read out the winner, I'm looking across the table, and this fellow's got to win this because I don't even know why we're nominated and we won. It's hard to understand where to place us.

And I think that from a view from the outside, it falls on what's important to the enterprise at any given moment. And the validation of that award gave us was that content marketing people probably think what we do is important. But that's only one persona that we deal with. There is an existing category in the world called Authoring Assistance. And Authoring Assistance software helps writers in your organization create more high quality work product. So you know a lot of them. Grammarly is an authoring assistant, Microsoft Word with spellcheck and grammar, authoring assistant. ProWritingAid is an authoring assistant, Google Docs has an authoring assistant built into it. I believe we're different than that.

Analysts have been grouping us and companies like us into that category because an author can use our product. But that's only one way that our audience uses our customers use our product. I believe we fall more in a category that's currently being defined around content impact. And what I mean is if authoring assistants help writers create high quality work product, content impact platforms help content achieve or surpass its enterprise goals. So we act on a different noun. We're helping content as a body as an asset inside the organization. And it's a very different place than what companies have looked at solutions like us as in the past.

Erik: So maybe now we can get and just do a quick walk through some of the use cases then. Because I imagine you have a number of different use cases depending on number one the use, and maybe depending on also the channel. What would be the primary use cases and then the user personas and maybe you can look at it from user persona, and then use case or use case in that persona. But how would you break down the usage scenarios?

Chris: Sure. So I mean, for many years, marketers may have talked about the digital shift. Digital shift is coming. And what we were doing was we were trying to scare customers into moving forward towards internet enablement of everything that they do. Because there's going to be a day somewhere in the fiction future, where the only point of contact you're going to have with your end user consumer is over the internet.

And I think we all knew that day would probably never come except that on March 10th 2020, it arrived. Digital shift happened. It's over. We've moved into a new era, where for the foreseeable future, your only point of contact with your consumer is through content, regardless of where in the value chain that lives. And so when you think about the entire customer experience of an enterprise, it's not just marketing content, it's not just the website. It starts in development. It starts in R&D. So I'm creating a product, that product has words on the screen. There is a user interface. And those words matter. I'm creating technical documentation to support that product. That's content.

So all of that technical content, the product manuals that accompany the product that I've released into internal education, I'm creating education content that both my internal staff and potentially my customers will use into product marketing, and the materials that accompany the products that I sell from a marketing standpoint into the content that's created in content marketing, to be bait for demand gen into demand gen and landing pages, and the emails that are sent out through both campaigns and BDR organizations and into the content that's handed to salespeople from a salespeople enablement standpoint.

And then into post sales and support with service desks and support websites and portals that store all that information. And that's not even taking into account the content that's happening in legal and compliance and HR and the whole move towards more inclusive language and accessible language on websites. So really, this lives everywhere in the business. And this has been an interesting thing because you can't sell that way. There are a lot of organizations, most organizations don't have somebody that looks holistically at the words of an organization and says this is how we talk.

But if we go back over that string of folks that we've identified, I need consistency. I don't necessarily need the same tone of voice. Because the way that you talk in technical documentation is different than the way that you talk in outbound marketing, it's different than the way that you talk to a support audience. But the consistency is the same. If I describe something on the screen in my product, I should describe it that way in my documentation and in my product manual and in my product marketing, and probably in my support. The way that I describe it, and marketing materials, all of that matters.

And the way that enterprises run, each one of these groups is its own silo of content creation, they all have their own job, and they're doing it. And they don't really connect very well. There isn't a lot of link up between technical content organization, and for instance, marketing. So I'm not getting the benefit as a marketing organization of the things that are happening in tech docs, I mean, never see that.

And what we're trying to do with large enterprises is help them understand that this entire customer experience represents one of the biggest assets that they own. It's the content that they create. But here's the problem. I can talk to a customer about efficiency, and our product can help you save money creating content. Fantastic, they say. Show me where am I budget I spend money on content. And then you think about it. They don't. It's not a line item.

If you're creating content internally as a byproduct of the things that you're doing for work, so technical documentation, for instance, creating marketing emails. For instance, I don't have a line item in my budget for creating marketing emails. That's what Allison does. And she works here. It's free to me. It's headcount. But I don't think about it as a cost. So the first thing is understanding that content is an asset and it is worth something and it should be treated as such.

A simple question in your team, in your department, and in your part of the organization, I don't know how many pages of content do you own, are in your repository and then we can talk through how much you think it cost per page to create that from a time standpoint, from an editorial standpoint?

And when you superimpose this over a single thing, so if I look at just one website inside a big enterprise, let's take a single subdomain of an enterprise at a company the size of Microsoft, where they get 1 billion visits every year, an average visit duration for that subdomain is four minutes. That's creating somewhere about 4.5 billion minutes of user experience every year. It's a huge number. It represents a huge volume of time spent with the audience. But it's really only a fraction of the total amount of touch points for that business. Like this is just one subdomain.

And when you start thinking about that, like 4.5 billion minutes and the cost that it took to create that, which is in the billions of dollars, it goes from show me where it is in my budget that I spend money on content to, oh, my goodness, we have to do something about this. And the next obvious question is well, of the content that's getting all of those visits, what are you doing from a maintenance standpoint? How much of that content do you look at on an annual basis?

And what we find is that most companies are covering, from a maintenance standpoint, about 10% of their content asset, which means 90% of their content is just sitting from the moment that it was put into production. There isn't a change to it. There isn't an update to it. But these big companies are governed by compliance and regulation and things change constantly. How can we constantly keep up with that across millions of pages? That's where automation has to come into play. That's where AI has to come into play in order to manage these huge problems at scale.

Erik: So you can probably apply some rule of thumb just to get an estimate here and say across any function, maybe 5-20% of people's time is dedicated to creating content that's some sort. But you can use some reasonable rule of thumb and see that it's a large number. So it's a problem we're solving. And then you get to this issue of how do you manage the complexity.

And if I'm thinking through the challenges that I would imagine you're encountering, and maybe you can help me understand these better, beyond the technical ones, I'm just imagining, like you said, you have these different silos and they're using different platforms for creating content. So you have probably a lot of complexity in terms of the different tools people are using to create and they're siloed on people's desktops and laptops and different SaaS platforms and so forth, you have content that's spread everywhere. And then you also mentioned that people have different terminology that they use and you'd like to have that aligned. But you then have this probably alignment issue of function A saying, hey, no, this is the way that we need to talk about it and then function B say, no, no, we should talk about in this way, and having maybe powerful stakeholders that have a preference for how they want to communicate, and then trying to align that on the brand level.

But what are the struggles that companies tend to have when they start to go through this process of looking at how can we more systematically manage and align the content that we're creating across our different functions and business units?

Chris: So I'm going to restate what you just said slightly differently. Every content owner, whether it's technical content team, the marketing team, support team, everybody has an idea of how they want to create content. They know the clarity, consistency and character, the terminology, the inclusiveness of the content. They understand that. It's written on a whiteboard in their office. And it's awesome that it's written on that whiteboard because they can show their management that they have a handle on how they create content. It can be a board level conversation that they have that strategy for content creation.

The problem is that it's written on their whiteboard so most of their employees can't see it. And because most of the people writing content in an organization aren't professional content creators, aren't writers, most of them don't care. Like when you sit down to create an email, you're not thinking about your company's terminology guidelines, you're just trying to get your thoughts down and press send.

And so the first challenge is to take those guidelines and make them actionable. So the easiest way to think about a guideline is top of the food chain, top of the hierarchy, the first guideline is we all spell the company name right. But if you work at a company like American Express, do you write American Express? Do you write Amex? Are you turning it into AE? Are there rules around that you say American Express the first time and then you can abbreviate afterwards? And how do people know? So that's a guideline.

And now, as you create your content, wherever you create your content, to your point, whether you're creating it in a [inaudible 20:23] environment, XML environment, you're doing it in X Metal, you're doing it in Madcap Flare, you're doing it in Adobe products, you're opening up Google Docs, wherever you're creating, you need to be able to be guided to use those rules.

Solutions, like ours provide, in our case, a sidebar that loads into your authoring environment. It's like an angel sitting on your shoulder. And sitting on your shoulder, it will tap you and say, hey, you said test, we say quality, would you consider making that change? And we're not going to force that change. We can automate that process. But writing continues to be as much an art as a skill. And so maybe you meant to say what you said. Maybe you were very specific about the words you chose. But as an organization, we have words that we like. And so would you like to use that word?

We also know our audiences. So we know the education level of our audiences, the complexity of our audiences. Hey, your sentence, this paragraph, this section of your content is way too complex. You're using words that nobody's going to understand. Your sentences are very long. Could you consider simplifying this? And here's some ways that we would suggest you do that. And again, not forceful, just like, hey, we're writing for 10th grade level. Could you make it so people understand this?

And so to your point, taking very technical domain research and making it readable by a general audience, you need that, you need somebody to say, this is awesome, you're very smart. I'm super proud of you for being a brilliant genius. But nobody's going to understand what you're writing. Could you consider making a change?

And the nice thing about that is in the real world, that's an editor with a personality. And in our world, it's just a prompt. There is no subjectivity to it. It's not opinion-based. It's not in a good mood or a bad mood. It doesn't care what shoes you're wearing, so it's not judging you. It's just saying, it's not aligned with our audience. Could you clear this up? You're not using brand language? Could you add brand language to this?

And that allows this team, this department, this organization, depending on the use of this to start to align. And so let's think beyond technical documentation to marketing content. For instance, companies go through rebrands all the time; and that rebrand doesn't just change the color of the logo, it changes the language that the company uses at the top level that the messaging hierarchy that the company communicates with. And so the question from the board, the question from the CEO, after the first month after rebrand is, hey, how are we doing? Are we aligning around the new brand language?

And the only way that most companies can know that is by pulling a bunch of documents, eyeballing them and saying yes or no. Like, I've got 10 documents I'm looking at that were written this month. We're not doing real great right now. But when you introduce AI and governance, I'm able to see from day one, this is how we start, this is our benchmark against this new brand. I can see the process of alignment down to the individual. How are we doing getting closer to writing the way that we want everybody to write? And then once we get to a state of alignment, we never break that, regardless of changes.

So if on day one, our tagline is Gem Lake, no fires. I don't need to know that necessarily. I can learn it once, maybe and the software prompts me. So if I say Gem Lake, no apples, it says, oh, that's cool. But that's not what we say. We say Gem Lake, no fires. If I decide two months down the road that it's now Gem Lake, no camping, nobody needs to think about that. Because when you type Gem Lake, no fires or Gem Lake, no apples or Gem Lake anything, it says, oh no, we change that. It's Gem Lake, no campaign, can you update this? And we stay in alignment going forward. It becomes very difficult for an organization to fall out of that alignment model.

Erik: So, there's a human element and there's an AI element. And the way I'm thinking about this, the human element is determining the set of rules around what terminology we use in certain. Correct me if I'm wrong, but it feels like you would have to have some sense of what terminology people are using improperly in order to refer the right thing because brands use words in all sorts of weird ways that don't align with normal rules of language so you have to somehow program in those unique ways.

And then you have the machine learning activities of then applying that to a tremendous amount of content in real time. Can you help just walk through the actual work to be done for the team that's implementing this? And then where's the human stepping in? And where's the AI stepping into the process?

Chris: Sure. So, in every organization, whether it be a department in a company or the company level, there is a sense of how I want to create this content. And at the scale of company that we tend to work with, they have things like a terminology board, and they're thinking through the language that the company uses, We even have one here because of what we do. And so there is that constant consideration of words that matter to us.

And I don't necessarily need to recognize all of the excluded terms, although that's super helpful. I need to think about the terms that we like, and that we want to use. And then our software can do the rest of that. But so implementation is making this system bespoke to your organization. So, understanding your guidelines, the level of clarity that you desire in your content for the audience that you communicate with, all of that terminology, the style guidelines that you use, so do you use Chicago style, APA style, your own style, and building that into your guidelines, bringing in the level of inclusivity, for instance, that you want in your content, the tone of voice, so are you creating a human voice for your organization, or are you a more formal organization out of the box where like you should be lively and engaging? And that's great, except that not everybody agrees with that.

We have big British banks that use our software, that don't want to use contractions, that don't want to have informal conversations. They are formal. They want to sound formal. And so the ability for the software to capture that tone of voice and then govern the creation of content around that tone of voice. So once all of that has been captured by the product, now, it's about putting that that guideline into play. And there's a couple of different ways to do that. So I talked about the sidebar and the live creation of content. But that's only part of how you would use the guidelines that you now have. Because that's just a go forward model, that's just looking forward at created content.

You also have this repository full of existing content that most likely does not align with the guidelines you've just created. So the ability to go and read through all of your content, it's that at scale problem. If this was just several documents, solutions would not be necessary. But the fact that this is hundreds of thousands, if not millions or billions of documents, means that I need to take those guidelines and overlay them over everything we own. And I'm finding the problematic language the risk in my content, and moving to remove that.

And that's where we go from that 10% coverage from a maintenance standpoint that we talked about earlier to getting to 90-100% coverage, and at least identifying where we have big problems. Where am I going to get sued? In regulatory issues, where am I breaching laws in my content?

Like, when you think about banking or pharmaceuticals, this is governed content. And if I'm not keeping it up, it's likely that somebody is going to grab something out of my repository and go and use it in front of somebody that they shouldn't, not trusting that the repository is the place, but we're not keeping up with the fact that we need to be uplifting this content. And so that's the implementation process, essentially, learning all of your guidelines, and then enabling the organization to now use the tools available to them to govern that content going forward.

The way that we're trying to avoid having to learn every negative term is that our software is taking in this content, this stream of characters and we're extracting context, then content in that and identifying or tokenizing this finding sections and sentences and then adding in linguistic data so that we can identify what we're looking at from a computer standpoint, identifying terminology with varying detection using our proprietary rule set to apply patterns back into that, mixing in rules then statistics and terminology to create tone of voice on top of that, and then pushing all that out in the form of guidance in that tap on the shoulder that says, hey, I found something.

And all of these results in a numeric score on each document. And that score is an objective good or an objective done, which doesn't exist in nature anywhere. Like if I say, hey, I want you to create this piece of content, Erik, and I need you to make it good. That's going to be my opinion. You're going to run into problems with that because I'm opinionated. And it's subjective. Wouldn't it be great if we had a numeric definition of good? For my organization, it's 80. An Acrolinx 80 score means that it's invoice. It's clear. It's consistent. It's using the terminology I care about. It is inclusive to all of our audience members. And it's ready for contextual review.

And that for me, having been doing this for 20 years, knowing that I can review content, for context, not for embarrassment, is a paradigm shift. Because for the most part, for most of my career, if a document makes it to me for review, I'm making sure it doesn't have obscenities in the middle of it before it gets pushed to my website. That's my guideline. Is this embarrassing? No.

Today, though, when I get a document, does this represent my business? Is this what I want to say to the world? And that's a whole different place in the value chain because now I'm looking at why are we creating this content? Going from there, you ask the question, what does your team create? I run a blog team. We create blog pages. Why do you do that? Because I was hired to be on the blog team. No, that's not what I mean. Why do you have a blog? Well, we have a blog because we want to be found for interesting concepts that we think tie back to our business. It's find-ability issue.

But we have conversion points on our blog pages. And so we're trying to generate some leads here. How's that going? Well, my boss says, this year, I'm supposed to increase the conversion rate from our blog section by 10%. We have not even started to do that yet. We're pretty static right now. But we're working on trying to create better blog articles. Okay, so we understand what the content is, we see what the desired impact is, can we lean into this and start driving real value by increasing the overall impact of this site, increasing the conversion rate? And why do we care about increasing the conversion rate? Because that's going to lead to revenue.

So increase conversion on a blog page means more leads that turn into more pipeline that get more opportunities for the salespeople to close business. So I'm a thumbnail away from actual business from dollars. So, if I can increase the impact of all of my content, I'm increasing the value of business and that becomes a whole different conversation for content people. Because again, the first answer of that question was, oh, because I'm on the blog team that's why I create blog articles. No, no, no, that's not why I pay people to build blog articles. I'm trying to drive revenue here. That's what we're all trying to do.

So by looking at, for instance, the best blog articles I have, do I have any blog articles that have 20% conversion? You have one. Okay. What is it about that? Can we find some guidelines associated with this, that make us think that this is why this is resonating with the audience? Can I take those guidelines and overlay that over content I own and drive up conversion on that content? And then if that works, can I do it across all of my content? Can I build a model for content creation that becomes predictive around conversion rates? And that's what our customers are doing with our software right now. And it's amazing to watch.

Because you ask the question unguided, how does your company value the content you create? And they're coming back now with the answer in increase revenue Oh, what? Yeah, we're a big company. People only care about revenue here. And that's what we're driving. Regardless of where we live in the value chain, we're creating content that moves the needle here. It's a whole new world. Content has become the thing that drives business. It doesn't need to be a secret inside the business anymore. People that are involved in the creation of words, which by the way, is all of us are driving the business, whether that's emails you create, the website, marketing content, technical content, support content, all of that is how the business moves forward.

Erik: And I can very clearly understand how this works when I'm creating a new piece of content and I am able to use this sidebar to guide usage of terminology and application of rules around messaging and tonality and so forth. You mentioned earlier that you then have this challenge of large corporations having hundreds of thousands or some cases, maybe millions of documents that then might need to be updated for different reasons, I suppose. So in some cases, we have a terminology change in terms of a brand name, which might be pretty clear. In other cases, it might be because we've done some analysis, and we found out that certain blog posts have a very high conversion rate and others don't. And we might want to adopt some of the learnings from the high conversion blog posts to the lower conversion ones.

How does that happen? Is that then applying a score and then asking a team of people to review the past blog posts and if there's something that has lower than a 70%, you should review and decide whether you accept or reject these? Or to what extent are you able to automate the process of actually updating the language? Again, language is complicated. So I can imagine that you might have an algorithm that would be accurate 80% of the time, and 20% of the time, it does something that just feels a bit strange to people. But what does that updating process look like for content that's already created and is published to the world somehow?

Chris: The software is able to go ahead and make those changes. In most cases, the customer chooses not to, for that exact reason. Because most of the time, it's going to work; some of the time, it's going to create crazy. If you think in terms of one of the ways, one of the mechanisms, you understand the concept of continuous integration, I'm coding all day, I'm checking my content into a repository, that repository is QA at night, I come back in the morning, and I have my results. You do that same thing with content. So content is created. It's dropped into a repository. It's reviewed overnight. You come back in the morning, there are scorecards with scores for each piece of content.

And you can flag those. So you can sort by that in your repository, find your problematic content, go and see what the problems are, and then make educated decisions on whether or not you want to make those updates. But those scores show you content that should be flagged and put on hold. So depending on the content repository you're using, you can actually block the use of that content that falls below a threshold automatically.

And so I'm essentially gating all of that content until somebody gets eyes on and makes those changes or decides that this content is no longer relevant, no longer good, should be taken out of public use. There's also the ability to just command line prompt checking. So just going in, digging into an existing repository, or drive, looking at every piece of content in there, and then coming back with just a master list of scorecards for all of my content, and again, flagging the ones that are problematic for review and letting the ones that aren't problematic go through.

The interesting thing about this is that humans still matter. So we are an AI solution, you get a lot of who does this replace? We don't position it that way. What we think in terms of is there are mechanical jobs that you don't need to do. And the people that do those today in most organizations are highly skilled folks that went to school to create amazing things. And somewhere along the way, because of their expertise, put into an editorial role where they're doing very mechanical tasks, that take a really long time, over the course of a lot of days.

If we can clear that mechanical stage up, that sets these people free to be creators, to come up with the ideas that lean into creating more and better content. It's not so much a replacement. It's a move in the organization to where value exists. So if you think in terms of support tickets, like why do I have support people writing tickets on my website? It's because I want to try and avoid people calling in, because calling into the call centers more expensive.

If I have better support tickets, does that mean I fire all my call center people? No, because they're the ones that know how to answer the questions. So they're going to be able to answer more questions online writing than they can on the phone. It takes less time. If there's answers, solve problems, better than less calls in more call center people writing, it's a circle to get to value. So I'm saving money on the call center and my people are creating better answers and better customer experience, which results in customer retention which results in more revenue. It all comes back to that.

Erik: Can you walk us through what this would look like for a company that's just getting on? What I'm thinking about is, this is obviously it's an enterprise solution. But I'd say mostly, it probably doesn't make sense for a subsidiary to be doing it. It's something you have to do at a corporate level. So who are the stakeholders that make the decision to deploy this?

I imagine there has to be some kind of centralized management team that's finding the rules and maybe in a lot of corporations, this already exists, and others, maybe they have to define that. So who are those people? And then you roll this out to a large group of abusers who could sounds like be almost anybody in the organization. And I guess that is more like using Grammarly which is a small ask for people. Just walk us through the process of onboarding the solution and then who would be the stakeholders along that process?

Chris: So if we look at something like tech docs, that's an easy one. The first conversation usually going to happen with a senior tech editor or the person that owns the technical documentation organization inside R&D. So it could be the head of product, the head of R&D, the head of development. And they're looking to solve a problem. They're trying to get a technical documentation process to keep up with the speed of software development. Because over the course of the last couple of decades, we've moved from waterfall to Agile to continuous integration to continuous delivery. And the development of software is continuing to accelerate.

But writing is still writing. And it still takes the same amount of time. Editorial takes the same amount of time. So how do I keep up with the speed of the business, the speed of development. So I'm looking to solve that problem. That organization has existing guidelines. They're very structured. Those guidelines are documented. But that's largely all they are. They don't have a governance system. There is a book. This is how we write content. And it is incumbent on people to know that. And my editorial team needs to essentially bang up against that book manually as part of their editorial process to make sure that we're following those guidelines.

Acrolinx will consume that set of guidelines. And that now that rollout is to all of those technical content creators that are creating that content and they're given live guidance as they write. And then the editorial process, again, becomes contextual, does this, say what we want it to say. And so it becomes very straightforward. If you want to now overlay that over an organization, over a broader enterprise, it probably starts in brand, probably starts with somebody that cares about all the words in the business. It could start with the CMO. It could start with the head of customer experience. But brand seems to understand this the best because they're setting the feel of the organization. And they too, have guidelines, but their guidelines are hierarchical above, for instance, technical documentation team.

Tech docs has their voice. Their voice isn't going to change regardless of what brand says. And this is a challenge that I learned myself here. I've built out our tone of voice, our clarity levels, the way that we communicate for our front office, sales, marketing, customer success, and it resonated. So I said, cool, let's overlay this over everything. Support Team, please use this set of guidelines. You know who doesn't think that are, are fun and light and friendly tone of voice is interesting? Our support customers: they just want their questions answered.

So brand up at the top of this hierarchy can't set the voice for everybody. But they can set the top of the food chain. So this is what the name of the company is. This is what the tagline is. This is how we want to communicate. And if you think about a company like Apple that does this really well, like Apple understands at the top level, how they communicate as a global enterprise, as hundreds of thousands of people, but there's lots of flavors of that as it drops down into product groups into technical content into the way they market.

And so that hierarchical drop of rules is the basis of how we build this out across the organization. It's oftentimes we're going to start with that senior technical editor. That's going to be the first person we talked to. Or it's going to be that head of digital marketing that we talked to and we're going to have to move up ladder to go back down ladder because somebody needs to be interested and somebody needs to be the first buyer.

But the beauty of this is when the organization as a whole starts to participate and that's when you see that corporate level terminology board, the people at the top of the food chain creating audience guidelines. You see HR building in diversity language, accessibility language into the way the whole company communicates. And that's where this really gets exciting, because then you start to see that alignment across the whole business. And that's where things really start to take flight and a company sounds like a company.

Erik: There seems to be an evolutionary process here, where when you first roll out the solution, maybe there's some gaps and then those are filled in and refined over time as the different functions get involved in. And then maybe three years into the solution, you've made an improvement from 60%-90%, in terms of where you want to be in the language. So I imagine there's an evolutionary process on the human end of people refining the structure and the approach. What about on the ML end? And is there a process where the ML is X% accurate on day one, and then on day 100, it's X+Y, and on day 1,000, it's X+Z. So does the ML improve the accuracy for each organization over time? Or am I thinking about this in the wrong way in terms of how that would work internally?

Chris: No, you are, because you've already identified that people will be able to identify whether or not the guidelines that are being fed to them are correct. And so we don't let the user change the guidelines. But we do let the user inform the guidelines. So part of the use of this is the constant monitoring of user feedback. You gave me a guideline, and I feel it's a false flag, could we update this?

And so that review of okay, a lot of people are saying that this isn't how we want to communicate, should we make this change or are we very specific about this plays into this? And the software will get smarter and better and you'll see less and less guidance that isn't relevant to your business. But that's only part of it.

The other part is we're coming up with these guidelines, ourselves, creative people who have that as a job, develop this strategy that they want to align their business around. But again, you're inventing that. You're making it up. Where it gets more interesting is in the misalignment between your AcroLinx score and the performance of the content because you're learning something.

I think that a number of years, and I remember a customer saying, hey, I've got an AcroLinx 90 score, and I put this out in the world and it didn't perform. I think AcroLinx is broken. That's not how software works actually. That's not the case. AcroLinx isn't broken; you made up guidelines that don't resonate with your audience.

And so what we've done most recently with the product is integrated the product with postproduction analytics, with Google Analytics, with Pardot, with Marketo, with support systems, where we can see what consumption looks like. So your AcroLinx 90 that you say is awesome goes out into the world and it doesn't perform. What can we learn about that? What can we take away from that to reinforce the process, to reinforce the guidelines so that we're creating content that resonates with our audience? And that's where the system gets markedly smarter. Because now we're able to identify how our audience wants to hear us, how they want us to communicate with them. And we can better align with that audience. So it's moving from this concept of internal strategy-aligned content to audience-aligned content. And that's where this gets really exciting for me because it's allowing you to be more aligned with the people that you're talking to.

Erik: So I think we've covered a good deal of territory here. Anything that we haven't touched on, Chris, that would be important to cover today?

Chris: No, I think that's a lot. I hope this is interesting to folks.

Erik: Yeah, I think it will be. Certainly, it was interesting to me, so thanks for taking time today. What is the best way for folks to reach out if they're interested in learning more about AcroLinx?

Chris: So our website is www.acroLinx.com. I can also be found at LinkedIn @CP Willis and I am always happy to talk, so feel free to reach out.

Erik: Great. Thank you, Chris.

Chris: Thank you.

Erik: Thanks for tuning in to another edition of the IoT spotlight podcast. If you find these conversations valuable, please leave us a comment and a five-star review. And if you'd like to share your company's story or recommend a speaker, please email us at team@IoTone.com. Finally, if you have an IoT research, strategy, or training initiative that you'd like to discuss, you can email me directly at erik.walenza@IoTone.com. Thank you.

Contact us

Let's talk!
* Required
* Required
* Required
* Invalid email address
By submitting this form, you agree that IoT ONE may contact you with insights and marketing messaging.
No thanks, I don't want to receive any marketing emails from IoT ONE.

Thank you for your message!
We will contact you soon.