Linked Data, Intelligent Content and Utility on the Web – a talk with Aaron Bradley
Today I am grateful to the genies of search, poststructuralism and marketing for granting me access to the one and only Aaron Bradley and allowing me to present you with his valuable thoughts on intelligent content, commoditizing data, linked data, knowledge graphs and writing.
Yes, writing – that big thing between soliloquies and product descriptions. I also want to warn you about something: there are too many conceptual thinking parts here. So please, for best experience onboard, let sink within reasonable timeframes. The reason for these conceptual heights is that within this 1-hour conversation I made an attempt to wrap my head around content marketing, linked data and a knowledge graph strategy at once. Fortunately, in this risky endeavour Aaron Bradley took my hand and lead me out of the munutiae right into the takeaway: in a multiformat, omnichannel universe, content needs to be portable.
With that, enjoy reading Aaron (or listening to him) below:
The Intertextual Animal: A Conversation with Aaron Bradley [Video Transcript]
Hi everyone I’m Teodora, and you are with the Intertextual animal series, this is my second episode with the one and only Aaron Bradley who is a Knowledge Graph strategist. And if this is not cool enough, he has been working as a Technical Services Librarian and his educational background is in English literature with a specialization in postmodern literary theory, meaning he has been dealing with semiotics and the science of signs for a while. So without further ado, Aaron – hi! Aaron, would you like to introduce yourself with some different words from the ones that I chose?
No, I think that’s an apt description. Thank you very much for having me, Teodora, it is a pleasure to interact with you one on one on this. For viewers should know that we’ve known each other online for a long time, the first time we’ve actually exchanging information in a live format. So very excited to be here. And yeah, I’ve been struggling with meaning for all my life, both personally and professionally. So the quest continues.
Wow, yes, it continues for all of us. We are together in this journey.
And if it doesn’t, there something wrong. If you think you know it all? Well, you know…
Let’s start with a question that I’m sure a lot of people are thinking about. And I will ask you this way. I know, we all know that a lot of us enterprises are tapping into the power of linked data. The Swedish library does it, doesn’t it? And this is powerful a technology. However, how do I, in my tiny writer’s life, implement linked data, provided I don’t have a knowledge graph strategist nearby? Can I do that?
As a writer, you know, as a writer, you’re already dealing in a world of linked data to a certain degree, because, well, not to a certain degree. In, in a large measure, because, you know, depending on your approach to knowledge, and your understanding of epistemology, because what any writer does is link facts, link impressions, bring knowledge to bear that you have that others have, question knowledge that you may not have. So any writing pursuit is, in some measure, an articulation of the world we find in the Web, we think of Linked data as it has been on the Web.
That’s where the correlation between writing as a pursuit of, you know, to be seen, and linked data as a pursuit to be on the Web are kind of different.
If you’re speaking simply of writing conceptually, as the linked data, you know, you need to actually somehow interact with the Web in order to bring that to bear as, as strictly speaking linked data. But the principles are the same in the approach of the Web for linked data, and the approach of a writer to the universe that they occupy. The things that they’re writing about, because there’s a series of backs and connections that need to be uncovered in both cases, or rather, that from a link data perspective on the web, that it facilitates uncovering information about those things. And most importantly, I think the thread that ties all this together is connections, right? Linked data is all about connections in a linked data universe, the connections between things are as important as the things themselves, and I guess, taking that, you know, conceptually a step further to apply that to writing, you would see that it’s not just what you talked about, but how the things you talk about are connected. If that makes any sort of sense.
It does make a lot of sense, is it the case that when I’m writing, I can for a while, forget about schema.org and about mark up? Do you think I can just do only with very well-structured and thought-of content and writing?
That’s a very loaded question, it depends on the intentionality of your writing too., right? If it’s not in a market environment, you know, you can write a tale or even something expository and it just has the structure that you provide to it. If you’re writing where your intention though, is, to
play a part, especially a marketing part in the Web, then structured then does become definitely a factor of play, it may not be a factor in your writing so much in the way that your content is packaged, and delivered, you know, this whole notion, I’m a big advocate of the notion of intelligent content – content that has structure and meaning. But that’s applied at the different level, right. The containers into which your content is placed, and then how you leverage that. And in an intelligent universe, intelligent content universe, you then do need to begin looking at structure of your writing as it relates to the content model in which you’re writing is going to be placed. And you can say, I don’t have any content models, what the hell you’re talking about Bradley. You do. I mean, even if you’re writing a blog post, there’s semantics and there is structure involved with that – you have a title, you have description, you know, those typical pieces of metadata that help provide meaning, you may have tags associated with a blog post that also semantic layer, the aboutness of your piece, but then the way it’s, you know, laid out, it may be a textual blog, but there are different elements within that your lead paragraph, videos that may be embedded, images that maybe embedded, sections that even in HTML have some semantics, you know, in terms of headings and paragraphs and list, there’s all structure to that, that you can start to think about as a means of creating your content so that it’s going get the biggest mileage on the Web and can link together different concepts.
So if you then translate that into a more formal linked data and intelligent content environment, you may be dealing with predefined content structures, that also gives you leverage and greater depth. So what you may have in WordPress as a blog posts may be further decomposed, in a localized content model so that there’s more structure rather than just a blob of content that makes up the body of your writing, there may be sections, there may be embedded components that you can start to view as modular and useful in different circumstances. So it starts to provide a mechanism by which the way in which a writing artifact is fluid, and can be presented in different ways in different environments.
That’s one of the primary benefits of having that structure. And also where meaning can be attached to a piece of writing on a more granular basis, not simply as that blog, but even pieces of that blog, and the references to which those points in terms of starting to link data from a piece of writing. And I’d be remiss to say that the killer app of the the entire Web is the hyperlink. And I think sometimes in the linked data world, we overthink or link and when you’re creating a hyperlink, and then you you’re creating a connection between the artifact that you’re presenting and some other piece of knowledge or some other piece of content, and that’s extremely powerful.
You mentioned intelligent content and a content model.
I am pretty stream of consciousness, though.
So I just wanted to combine something out of this and ask you, if a content model, the content model concept, the content intelligence, or the intelligent content concept and SEO were linked with typed links? What would these be? How are these working together?
Not precisely sure that I that I fully understand that question
Do you want me to ask again?
Sure. I mean, are you asking about the how content models and intelligent content impact SEO and approach to it?
No. I was excited to because I, I very much connect you to search engine optimization. And I’m very happy to hear you talking about intelligent content. And that is not something a lot of SEOs talk about, am I right?
You are right,
It’s a sea change, I think, for most search engine optimizers, and I don’t think they’ve, they’ve really understood this to a large degree, because one of the primary benefits of intelligent content is that it allows your content tend to appear in different endpoints, and to appear sometimes in different ways in those different publishing environments, depending on their capabilities, and how you style it for those different environments. So intelligent content, a fundamental concept is that you separate the presentation from the data layer. So that makes your content much more portable and much more fluid. So that if you have something that we might, you know, conceive, again, we’ll just keep using the example of a blog post, how do you port that to a watch, how do you represent that, and when trying to provide that to a voice assistant, right. And by having that structure in those that meaning attached, it gives you that ability to have your content adapted to both present and even feature environments, because it’s just data that’s separate from the presentation layer. It has structure and semantics. And that means that you can push that data to whatever endpoints you want. And I said that this is kind of a foreign concept, I think, still to most SEOs because they tend to think of the Web.
And especially, they tend to think of a purpose of SEO, if not, the primary purpose of optimizing content is to drive users from search engines to a web page. But a web page is just one of those publishing endpoints. And as you know, should be quite clear, and this at first occasion, by the birth of mobile and the importance there, you know, what Google said, a webpage is not the same as a mobile experience, right, the representations there may be different, then, you know, what we want to be able to provide content from that piece of authored content directly in the search results in the form of a featured snippet. We want to be able to provide users, Google users with the capability of having some portion of content read back to them in an assistant. So bringing this back to the kind of trajectory of traditional SEO, and why that is so radically disrupted by
intelligent content is that you’re no longer optimizing for website traffic you’re optimizing to have your content surfaced in as many environments as you can. And you know, I think, still, most SEOs really conceive of the primary success metric for a Google query as being driving traffic back to a website, that’s cool if your stated goal is to have website traffic. But of course, business goals are normally to sell things to engage users to have them, you know, participate in whatever funnels or actions that you know, a business wants to participate in. And you should be leveraging the change in environments and the possibilities of things like, linked data and intelligent content to say, No, I don’t care if I get traffic to my website, what I care is that I get that sale, I get that the the lead gen that I was looking for, doesn’t matter that that happens through an assistant or on a web page or on an app, no, but you’re going to be constrained without the adaptability of having a intelligent content, fuel that creation and publication of that content. And it’s a it’s a different outlook. And I think most of all, you know, the, the single most radical change in search technology, since the invention of Google was the introduction of the Google Knowledge Graph in 2012, where instead of, you know, Prior to the introduction of the knowledge graph, Google was a repository of, it was an index of link data summaries, I mean, of web page summaries, right.
In 2011, if you made a query into Google and got what you got back were the famous 10 blue links, and what those ten blue links are the names of just linked, you got a link title, and link description, it wasn’t returning data to you, it was giving you a mechanism by which you could check out the resources that Google thought might be pertinent to your query.
With the introduction of the Knowledge Graph, instead of being an index of documents, Google became a repository of facts, and then that gives them the capabilities or it started to give them the capability rather than returning linked document summaries in a search result response to provide facts, whether that’s in the form of featured snippets, whether that’s even though that’s not from the Knowledge Graph. But as a construct where data is provided directly into the SERPs building on featured snippets, or rather rich snippets, which started around 2009, you start to see data directly associated with those linked documents summaries in the SERPs. And the terms of things like user reviews, or product information where you don’t need to leave your browser, you leave your search results in order to determine the blue widget cost 9.99 and has a 4.5 rating and that with the introduction of schema.org in 2011, from there being able to create more and Google creating more and more data types by which again, data is provided directly into the search result. A recent one, that’s a good example or job postings, right, where the trajectory before whenever those were introduced a couple of years ago, now, I think I would search for a job. And hopefully, Google would find a match at some job board or some job aggregator board. And I would go and check out that job with the introduction of the job posting schema, same thing, right, where I can look for, linked data specialist 🙂 and find no no jobs :), no, I can enter something in, you know, a job related query into the search results and have data about that job returned directly to me without leaving the SERPs.
That’s that’s a massive sea change. And I don’t think many in the search industry have really wraped their heads around it, they mostly see that as an incursion of Google into their traffic that they think they deserve back to their website. And again, that’s fine, if you think the purpose of the web from a business perspective is driving traffic to a website. But it’s generally not in the case of a job posting, you know, you have different players that have different goals there, right. So if you’re an actual hiring organization, you don’t care if someone goes to LinkedIn or Monster or wherever it is to see the job if it’s exposed to a user as a directly in a search result. And that leads to you being able to hire qualified candidate, you’re all good, right? from a job aggregator perspective that’s different, they absolutely want you to visit their website and, you know, participate in whatever conversion funnel they have there, or, you know, get the benefit of being able to say that we refer so many candidates and why our job board is different than others. And that’s disruptive, right. And increasingly, and this pertains all the data types for which we see those sort of in SERP results are how much is a job board necessary, right, except as a repository for the data that send us the fuel other applications. And I think hopefully, that brings us back full circle to the notion of intelligent content and its benefits. If you’re hiring organization, and you’re trying to get qualified candidates, you don’t care about traffic to any website, you care about having that job description made available to users in as many environments as you can. And the environments where where they are, whether that’s on a website, like LinkedIn, whether it’s directly through search results, whether that’s through a mobile app, whether it’s through asking Google Assistant, hey, I really need a new job, right? That’s where you want those results to surface and it like, and that’s facilitated ultimately, by having that piece of data. You know, job posting becomes a content type, which has structure which has meaning. And that’s what makes it portable and you start to think and comparison of that WordPress posts that’s essentially a blob, I think it’s easier to understand structures and pretends to something like a job posting, salary, work experience, requirements, start date, all of these things are data points that can be represented differently in different environments, but are are the same no matter where those environments are.
If the data surrounding the meaning of of a job post is what allows it to then appear and Google search results. Yes, that’s a result of the structured data markup that’s attached there. But what you’re doing is just providing annotations on a posting. So that’s a mechanism of structuring that it’s a de facto transformation of flat text into a piece of intelligent content because it then acquires structure and meaning.
Okay, that’s very well put, because I wanted to ask you, within that context of the text becoming from a flat thing into an intelligent, dynamic structure, how do content writers ideate and think about and create their content, whether they are people who are writing for themselves, or they’re working, for example, for you as a Knowledge Graph strategist? I don’t know, do you work directly with content writers?
You bet. Yeah. All sorts of people involved in content operations. And I guess, there’s only so much a content creator can do?
Yes, because I’m sure that you have a lot to say about your process of working with, with content writers, because they are weavers too and their structuralists too. Do you know what I mean?
I do, but especially at enterprise scale, you know, it’s a lot different, having, you know, an individual, you are blogging where it’s all down to you, at the enterprise level, though, there’s a notion and I know, you spoke to Cruce Saunders of [A] (simplea.com), the other day, the proponent of these things that you’re really dealing with the realm of content operations, right? So fundamentally, a writer’s job is to write it’s not to design structures in order to bring that writing in a sensible way to the web. And I mean, the web read large, not to websites, but again, to to the internet, right.
Things like user interaction designers, and content modelers, and those involved in content operations really need to facilitate the assignment of meaning and the structuring of that content. So let me give you, you know, let’s go back to the job posting as an example, because somebody actually writes those, right, they don’t appear out of nowhere. In that case, it’s not a writer’s responsibility to figure out where the structure, you know, to modify the structure so that they can put a job title and a job description and a start date, right, those are fields that then exist in the structure of a job posting. And by providing that structure, you, you know, it’s fairly evident in the case of a job posting how a writer needs to leverage that, in the case of something like a blog post, if all the structure that you have is, you know, that blog, you’re not going to be able to do too much with that, aside from make kind of fundamental assumptions about how things appear in the web, you know, where you’re, for example, your lead paragraph is extremely important, because it’s likely to be the first thing reproduced and limited context environments, right, where you’re not going to get the full post, but you’re going to get the first couple of hundred words of that. But if you work with a content engineering team, where that lead paragraph is actually a separate structural element, and that can be swapped out, you know, depending on the environments that appears, where you can have something like, I think you do this in your blog post, Teodora, where you have a pull quote that you can tweet out, right, that’s a structural element that you can then plan around, but you can’t, if you’re a writer, and all you’ve got is a blog. So there, you know, you you put that widget or however, you know, that was constructed in the CMS in order to facilitate that. So structure requires, from an engineering point of view, also maintain the capabilities of providing that structure. And that’s the structural side. And then on the meaning side, you know, in a blog world that’s mostly thought of as tags. But that’s, that’s the semantics of it, right? What is this piece about, what are its topics, to which entities does it refer,
And again, there’s engineering work there, that can either just be wrapped into your writing, or that can be formally stated in the form of things like tags are in the form of things like linked data, right? In terms of structured data, where, again, most writers aren’t going to care about that, and it shouldn’t be their responsibility.
But if I’m writing about, I don’t know, Samsung Galaxy s5 there’s a way in which I can convey to search engines and other data consumers exactly what I’m talking about through the use of the same aspect, you know, that can be attached to any piece of content. Schema.org gives you the about tag and the about tag its range is a thing. So you can say, This article is about Galaxy Samsung s5. And here’s the hyperlink to it in Wikipedia. And that makes it clear to a data consumer what you’re talking about. So there are mechanisms by which the semantics the meaning of a piece can be conveyed to data consumers, that’s independent of the way in which it’s conveyed to humans writers, I mean by human writers in the actual text. So in summary, I guess what I’m saying is that the propagation of intelligent content, because it is both structural and semantic relies not only on how you approach writing, but the containers into which that writing falls. And to a certain degree, and this is a, you know, a gap I see in this rapidly changing environment, we’ve tended to separate the world of content engineering, which is, you know, mostly thought of as operations surrounding a content management system and content strategy, which has tended to focus on what content should be about.
A more forward facing content strategy also brings in the, you know, it’s not just strategy about the topicality of an article, say, but also its structure and how it manifests itself in the broader universe. And they’re something of a gap because traditional content, you know, operations, people that run content management systems, like haven’t been one designing content mark, you know, typically designing content models, and figuring out how content is represented in different endpoints, it’s more, okay, we need to add tags to, you know, AM or WordPress so that I can mark this up.
I think that organizations that are going to be increasingly successful with having their content surface in the modern internet are going to have to have a dual focus not only on content marketing, which is what I really think a lot of content strategy is about, but also the strategic manifestation of that content. So not just it’s topicality. But it’s representations. Does that makes sense?
It does. I just had an a-ha moment, thinking about what if content marketing was done not that right? I mean, what if content marketing, a big part of content marketing, can be just data done dum? right? Correct me but is the case that when you create content, you feed your algorithmic audiences, and then they reach to the different endpoints you talked about?
Exactly. And I think it’s also shifting and this this is the really hard thing, I think, in terms of being disruptive, where it’s the utility of the information that you’re providing, rather than it’s the destination that you’re driving people towards, in this, you know, increasingly is just going to cut out a lot of content. If it’s not usable in an abstract environment, right, then it’s not useful just because it resides on your on your website. And content marketing has been used, you know, and I’ll use the word content marketing relevant content strategy, there has been, you know, used as a hook to drive people to websites, so that they can then start on a conversion funnel. In a multi format universe, where information becomes more important than the place where it resides, it will turn out that a lot of that information simply isn’t useful. You know, if you look through, you know, I am an obsessive collector of information about these things in the web, I tweet a lot of links. So I have a feed.ly setup. And one of my categories, this content strategy. And again, you know, it’s really largely content marketing. And you see, like, endless posts were five, five ideas for your next content, post right, interview endless variations on these things.
Sure, that’s good if you want to create some hooks to drive traffic to your site. But your content strategy should be predicated on the needs, interest and desires of your users. And so that should reveal itself, you shouldn’t need to, let’s come up with an idea for the blog post on Tuesday.It should be driven from the other way around, what are our customers missing that we can provide to them, and especially that we can provide to them that nobody else can. And it turns out in a lot of cases, that’s pretty slender. You know, if you think of this from an e-commerce perspective, and this is perhaps the biggest example of that sea change that came with the introduction of the Knowledge Graph.
Prior to that products, really, you know, in the non shopping world, that is to say, not in Google Shopping, but in the organic SERPs really weren’t products, they were documents summaries, right? So Acme Blue Widget, you query that, you’d get 10, blue links, all of them documents summaries for Acme Blue Widget, what that actually is, it’s a product with all these product characteristics and things, right, but Google didn’t know that in the pre-Graph world. As you’re able to fashion these not as documents summaries but as products, acme blue widget, guess what, for all of those 10, blue links, it’s the same product. And you don’t really need to go to any of those pages to find out information about Acme blue widget. And you can see now for a lot of products on the Web, you’ll get the kind of the equivalent of a knowledge panel on the right hand side that captures the key characteristics of that widget. You know, if it’s a camera, it’ll list things like the aperture range, and the pixel image resolution, and land size, all of that. So at one time, I used to have to go and visit those documents to get that information. But because it’s a product that is mass produced and sold by these different vendors, in 10-blue-links land, I don’t need to go there to get the data about it. So what’s unique about that product that I can showcase and get people to purchase it for me, rather than somebody else. And that’s where you have the utility of things like visa reviews, or extolling the virtues of your customer service, because it’s better than somebody else’s. But the days where you can expect to leverage, especially for mass produce goods to kind of send people to your product details page, because it’s better optimized than somebody else’s product detail page are disappearing, that’s still going to be the case, you know, the 10 blue links are there, but what do your users care about, they care about the product details. And if those become widely available through aggregation, then they don’t, you know, your potential customers don’t need to go to your site to see information about that product. So what are the mechanisms that you’re going to sell that
As data becomes commodified.
And as the most important thing that a search engine can return, that’s what Google wants to return to users, particularly in a mobile world, where it doesn’t want to send people to web pages, which may or may not be mobile optimized, and which may or may not contain the information that you’re looking for, it would rather extract the information from web pages and provide that to you directly. And where that information isn’t actually derived from web pages, but instead, from other fact repositories. In the case of a product, it could come from Google manufacturer’s center, or from the structured data included with that the de facto structured data of Google Shopping. And by the way, Google Shopping just opened up to organic users. You no longer need a Google Shopping account, to actually sell your, you know, to use Google ads in order to have a Google Shopping account. Because Google wants to feed the XML feed of your products, which is equivalent, exactly equivalent, in most cases, to the sort of product attributes that you can mark up with structured data. So in both cases, what Google wants to harvest information about your products, that that it can then reproduce on its results. So it can provide them to you directly, rather than forcing users to go through a webpage. And I don’t think this is fundamentally about stealing traffic, or keeping people in the ecosystem. I mean, maybe part of that, but it certainly a better user experience. And in some cases, it’s, you know, having the data available, rather than a bunch of text is required to fashion that experience. And I think, especially if voice speakers where, by one information, better product returned by Google Assistant, it needs to be provided to Google assistant in such a way that it can say, you know, hey, Google, what’s the maximum aperture of a Nikon T151000, right? Try and kind of retrieve that from text.
But if you have a structured field that says “aperture range”, right, then Google can provide that. And that’s certainly going to be the future, we’re just going to see more and more data points being exposed on the internet, and then can be used flexibly, right. And you think, to extend that to comparisons, where you know, you need like data for that, if you say, which, which camera has the bigger range of apertures. And you name two different cameras, right? That’s very difficult to do. If you’re just dealing with unstructured text, it’s much easier if you know what the concept of “aperture range” is, and you know, the entities that are represented by the two cameras that you name and you’re able to, then rapidly if you Google “create a comparison”, you can see this, I urge anyone listening to Google GDP of France, that’s one of my favorite examples where, because that’s a bunch of data, right?
And indeed, what we’ve seen with Hummingbird and Rankbrain and a lot of the post-Google, post-Knowledge Graph release work by Google isn’t in an understanding the query, right? It’s not about providing results, but it’s understanding the queries that you can provide the right results. So if I say, GDP of France, Google does a whole bunch of stuff that, you know, I know what that means.
Well, GDP means gross domestic product, I then reconcile that to, you know, as a concept in the graph France->country-> cool, I know what that is great. I can give you the GDP or France. And if you were to make that query, it’s also going to give you the GDP of other nations, because it knows, you know, it’s got that structure, GDP is an attribute of a country, and therefore, other countries can have GDP, oh, and we know through queries that there’s other attributes of countries that people frequently search for, and we can return those as well. So I think that that where you’re also going to see things like population of France and unemployment, right. And you can compare that against those different attributes of countries as well. And the other interesting thing about that query is, and I may be wrong in this, it may be a single source, but often increasingly, in knowledge graph, power queries like that, you’ll see sources include Wikipedia and the World Bank, right.
And that’s when it it becomes increasingly difficult to maintain the notion that you’re always going to have a link back to your site if you have a source of data because Google is capable of amassing data. And reconciling that from heterogeneous sources, and increasingly so. So what is the source of you know, the fact that a Nikon D 5100 has a maximum resolution of making this up, 20 megapixels. Is it one source know, it’s probably several sources where Google has compared and contrasted, made an assessment that got an authoritative source and another authoritative source, they agree on this, and you’re not going to have a simple path of provenance there that lead you back to a single expression.
I know that a lot of SEOs have been complaining lately that featured snippets or, you know, what we think are kind of sort of features snippets and feature snippets for everyone listening just to, quote, common misperception are not generated from structured data that are generated by Google parsing websites and extracting data. But what if it starts to extract that data from multiple sources and bind it together in some sort of data narrative? That singular link back to a site is going to become increasingly rare, particularly in terms of anything that’s data oriented rather than more, I don’t know, subjective, you know “How to change a tire?”, you’re probably always going to get some sort of nicely structured list of steps of changing a tire from somebody. But as you move into things that are kind of less subjective, you’re going to increasingly see as Google’s capabilities of interpreting that as data and returning that as data, fewer and fewer of those links back to a single source, because they don’t exist, which raises epistemological problems that we won’t get into. In a recent article, I forgot the author, we will check later, mentions that Google is increasingly saying this is in our, this piece of data is in our graph, just trust us, right? And that is problematic because Google may have provenance information about that piece of information, but the rest of us don’t. And I think that that’s going to increasingly become a problem.
And I see that increasingly as a problem gain epistemologically so you can verify the sources of data and make assessments on it, rather than the SEO concern in that case, which is you’re not providing a link back to my site, and therefore, you’re stealing traffic to that ladder concern. I think you’re just going to have to come to grips with that and start rethinking the way you approach SEO and again, putting value on the quality of the information, the usefulness of the data that you’re producing, rather than on its seeming ability to drive traffic.
Okay, you’re saying rethinking the way people approach SEO? The truth is, after hearing you, I’m rethinking the way I’m approaching content writing. And is there such a thing as content writing in the universe you described?
Maybe not, maybe there’s just writing, right. The same way that it used to be.
You know, keeping in mind that the web created content marketing, right? I mean, there’s a little bit of the that, say, in the print world, but there was a much clearer divide between advertising and what we would now call content, you know, people just didn’t create content in the same way that we do, which is as a lever to sell something. And when they did, and interestingly enough, perhaps, then it was valuable, right.
You would have Robin Hood, a flower manufacturer, produce a recipe book, right? And that, you know, provided branding. But it was useful because you got recipes out of that, and recipes in the pre web day, you have to, you know, open a book, or pick up a card from the grocery store, or whatever it is that was actually a useful commodity, to have those Robin Hood or Betty Crocker or whatever recipes.
And there’s a direct correlation on the web today, where the same thing happens where recipes are used. And this is this is maybe a good example of content marketing, where the data is useful in terms of a recipe. Recipe may lead, you know, a content creator, or, you know, a spot, you know, whether it’s a Food Network, or Betty Crocker or whatever, oh, there’s ingredients that I sell or that I broker that you may be able to, as a result, see through this, there’s advertising that you’re going to see when you visit the recipe page as a monetization path. But the reason why the recipes work is that they’re actually valuable data in themselves isn’t some stupid listicle in order to drive someone to your site so that they can go “Man did I really just spend four minutes of my life looking through this lame list?”, it’s “No, I wanted a chicken cacciatore recipe. And here it is”.
And user needs satisfied, publishers needs satisfied in terms of generating revenue from advertising or selling ingredients to that user. So recipes are a good example of where you can come out of, you know, commoditize data when the data is useful and valuable to your users.
Okay, and my last hope is that corporate blog writing is not dead…
Um, yeah, I don’t know, it depends on on the blog post, I guess, right?
Of course I am talking about my blog posts 🙂
Well, in terms of your blog posts, you’re talking about ideas, right?
That’s a different, you know, that’s a commodity that isn’t mass produced product information, right? It’s something unique, and that users aren’t going to be able to get anywhere else. And from a corporate perspective, think of that in the same way, when you’re producing your corporate blog.
If your words, or the words of insight from your CEO are kind of useful, because he or she has unique perspective on the business and may be able to offer some things there that are valuable to users… It’s a type of information.
But as we all know, a lot of the information that gets placed on corporate blogs is window dressing, and never has window dressing being less important than it is now. And it’s going to become even less. So content has to have purpose.
And again, I’ll just again, just endlessly underline the point it has to be of independent value to users, if it’s simply a mechanism to get them to a place where there’s other things that might be useful to them. You use traditionally blog content marketing to drive users to a website, so that you can set them down a conversion funnel so that it can purchase something, maybe purchasing something is useful to your users. In that case, that’s what you should focus on.
You are to ensure that you expose a pathway and information about products as broadly as you can. That’s the other big switch, in the context of the intelligent content I was speaking about. A world that’s married to a data-driven world in Google. It’s the value of that content, because it can then be ported to different endpoints.
The more structured the content, the more meaning in it, the more places that it’s going to appear. And a lot of the traditional kind of SEO thinking has been around restricting the flow of information: I want to bring users back, I don’t want to have that duplicate content thing. And that is absolutely legitimate from a search engine indexing point of view, but you should want to, speaking more broadly, not in terms of duplicating your content across two or more web pages under different URLs, that’s fine, you know, work towards canonicalization, but you should want your content to be exposed as many places that you want, you should want to duplicate that in the sense of not duplicating it over to web pages, but having the same content available to users on a bespoke app as on their webpage, as is on a Apple Watch, as is on a Google Assistant.
By approaching content as intelligent content, you have that capability. If not, this is the challenge that a lot of content operations people find, especially working with legacy content management systems that have been designed for the Web. It’s hard to port that to all those different endpoints. And you have to create specific structures, or, the most commonly of all, you end up copying and pasting things from, you know, oh, this, this works as a blog post, but in order to get this into our app, we need this in a different data format, so we’re going to copy that over. And then some of the content is there. And some of its not. And we made a change to this piece, but it wasn’t ported over to that piece. So the more in which your content is styled for a single endpoint, the less portable it becomes, and then is exposed to users, and fewer and fewer environments.
So going back to my earlier point there, don’t be afraid of duplicating your content, be afraid of duplicating it in a way that’s going to confuse the search engines such as the same content under two URLs. But definitely, you know, exposure content to as many environments as you can, and and where your users are, right. So your users may not be on websites. They may be in different environments, like apps, or interacting with a voice assistant or looking at their watch, right? You You don’t want to make the users come to you. You want to go to them.
I have so many questions to ask you. And we’re so many minutes over what I had promised to myself to keep this short.
As you can see, I have no problems talking. Viewers still watching. Thank you for hanging in there.
Okay. I’m thinking about the question I was just curious is about. I will ask it as an epilogue: Where do you start creating a knowledge graph strategy from?
Great, great question! The fundamentals of that, I think, are fairly straightforward. Start to think about the entities that are important in your domain of knowledge, and how they’re connected, because that’s ultimately, then then the first step in construction, right, is to build it’s to build an ontology. And that’s what an ontology consists of. And that’s the underpinning, you know, there are several things that go into a knowledge graph. And maybe on a subsequent conversation, we can get into what knowledge graph are, and why it’s so confusing. How to this point, there is no singular definition of a knowledge graph. But it’s fairly commonly accepted that one of the key parts of a knowledge graph is an ontology. And what an ontology describes are the objects, the entities, the things in your universe that are important, and how they’re connected. And it’s the how you’re connected part that differentiates a knowledge graph from more traditional database-like approaches to the capturing and presentation of knowledge and the Web.
Because in an ontology, and then by extension, in a knowledge graph, information about the connections between things are as important as information about the things themselves. And that’s the enormous power of LinkedIn as suggested by the word linked, that allows you to start to connect those different entities and in meaningful ways, and even to generate new knowledge on the basis of surfacing connections that you didn’t even know previously existed, right? So knowledge graphs are all about the connections between things that’s about the fundamental fuel that that that makes them a valuable technology.
Another thing I cannot leave without asking you again, I admit, I’m kind of sad that you said that, as a writer, I’m not designing structures. And I just want to talk through this a bit because you mentioned capturing representation of knowledge. I do that, don’t we writers do that. And aren’t we the points in the portals of collision of different domains? And is this useful for for this structured formalized world you described?
Yeah, again, that depends on intentionality. Right? And, and what the, you know. Writing is a big word, it, it can be a product description, or it can be a thesis on some obscure philosophical subjects so you can’t approach in the same way. But actually, I think here’s kind of a middle of the road suggestion in regard to writing for the web and, and commercial environment, we just discussed featured snippets, right. So yes, in order to avail yourselves of the benefit of intelligent content, you need content models, which define structure and meaning at a level that you’re not going to be able to do as a writer. So cool, I need to get content operations involved and do the best I can there. But at least currently, because featured snippets are generated from the content of web pages rather than our database. So, there’s combination, you know, it’s still trying to ascertain meaning of what’s being talked about and return that that summary. But fundamentally, they’re based on simple HTML structures, right? So yes, if you if, if you want, you know how to change a spare tire to have your result appear as a featured snippet on the web, their structure that you add to that, and that’s fairly well-known. I create a list and more specifically, I create a list using, you know, standard HTML markup for a list.
If I want to have a summary table presented in the SERPs and link back to my content, then you create a table and you create a table using HTML tags for that, because that’s the structural clues that Google is currently processing in order to create featured snippets. So there’s tons of ambiguity there. But it is something that a writer with knowledge who has the intention, let’s say, again, from a marketing perspective, where you want that content to appear in search results above all others, then independently of whether or not there’s a content model associated to that, you know, even if you’re just writing a WordPress blog, sure, you can say “Oh, we know that Google likes the sort of structures for this sort of featured snippets, and you can design your, the structure of your writing accordingly”.
But to the, to your broader point, writing is a means of expressing thoughts and information, it’s not a technology in itself. So you need to be able to bind the two in order to make that sensible, you know, writing is the fundamental activity of humans. That is the way in which we express ourselves: everything that is everything from product information to soliloquies on the nature of love, right, that’s, that’s both writing, but the way in which you approach that it’s different depending on the the intent of what you want to accomplish with that.
And generally, people are, you know, have a different intent when they write a product description of a Blue Widget then when they’re creating a thesis on the meaning of love
Point taken, I’m just very into this idea that linked data is a lingua franca. A language that you can express yourself, no matter whether you write a love poem, or a product description, and maybe this is why I’m asking so many questions. This is a conversation to be taken to another level. We need to meet again.
Yes, we do.
Is there something I forgot to ask you?
Oh, I’m sure there’s many, many other things. I don’t think you’ve forgotten anything, I’m sure there’s many other things we could talk about. And I’d enjoy the experience of meeting again.