Posted by
Radomir Basta
The ability of Google’s algorithms to extrapolate data from human speech and other implicit signals in order to understand a user’s intent behind a search query has given birth to an entirely new era of web content creation. With the semantic web greatly changing the way we find and consume web pages, new rules were imposed on companies and brands aiming to provide insightful information for their online readers.
Essentially, they are required to formulate more comprehensive, versatile and useful ways to organize online data and enable both search engines and end users to find it easily. More importantly, this means adopting techniques beyond engaging in a bit of ‘subtle’ keyword/synonym stuffing and actually understanding the ways the semantic web works before producing any type of content.
Semantic Search: Background and Development
Exploration of the semantic web may be said to have culminated in Hummingbird, but most experts would probably agree that this is a bit of an oversimplification. While it’s true that Hummingbird draws heavily on the existing data and constantly improves semantic web infrastructure, it is far from being a definitive answer to the question of how content should be created for the semantic web.
Hummingbird is the best answer we can currently give to the changes in the way people perform search and the accuracy of results they are getting. While some of the algorithm’s features do address the needs of people making queries from mobile devices, their convenience is not even close to being the only motivation behind the final formulation of the algorithm. The semantic markup, entity-based search model, as well as all patents, innovations, and research underlying what we now know as semantic web, haven’t been invested with so much time and effort to only facilitate the search process for the people who browse the web on a smallish screen.
Instead, this approach tries to organize an exponentially growing chaos of human creation (uniformly acknowledged as the least organizable chaos of them all) known as the World Wide Web into a structured system where every idea, relation, person, or object are defined through a single uniform resource identifier, making their properties clear to the machine tasked with retrieving them.
Possible implications of such a technology extend far beyond making voice search more reliable and, if nothing else, the release of Hummingbird has brought much attention to this area of inquiry. The main idea here is that search engines increasingly understand not only the explicit signals a user sends when searching the web, but also a variety of implicit ones.
For webmasters, the content should be optimized in such a way to anticipate the needs of end users who might be looking for a specific piece of information in a number of different ways.
Optimizing Website Content for the Semantic Search
One of the goals of people working on the semantic search is to enable the engines to make use of all the digital data available so as to display the right information. However, webmasters too need to find a way to help the SEs in this task by optimizing content properly through XML sitemaps, clean information hierarchy, semantic markup, and the like. Obviously, if a search engine is to retrieve relevant results, it must have a firm grasp on a user’s intent, whereupon each new relevant explicit or implicit signal helps enormously towards this goal.
In terms of individual websites, this means that webmasters need to provide the exact sets of data to enable search engines to identify website contents properly and recognize them as meaningful units. This is particularly important for businesses that aim to differentiate their company names, especially those that are comprised of common, generic nouns, from all the other items that already exist on the web. As Google starts adding more value to brand mentions on external pages and social activities in Knowledge Graph, it becomes clear that businesses need a smarter strategy to accurately represent their area of work to search engines.
Therefore, apart from building a brand around a specific keyword or phrase, which has always been anything but easy in certain highly competitive industries, organizations also need to be mindful about how these phrases are distributed around the web. Coupled with location-based personalization, this becomes an even more complex strategy that primarily needs to rely on a user’s intent. For brand managers, this means:
- Investing more time into creating pages filled with the right pieces of overt and meta information (through meta descriptions, semantic markup, Knowledge Graph, etc.)
- Examining possible user search behavior, i.e. taking user intent into consideration when planning a keyword strategy
- Planning an online content strategy that would ensure search engines recognize the relevant context around the brand name
While for unique brand names such as Coca Cola this is relatively easy, those like Amazon need to work more on making sure search engines will recognize them as a brand and not the associated personal noun or any other business with the same name. This is where triplets and entity based search come into play, and where we try to cross into the ‘actionable advice’ part of the article.
Transforming a Brand Name Into an Entity
If you know how triplets work, you know they are roughly based on the linguistic model of subject, verb object, where subject and object are entities, while the verb denotes their relation or interaction. If you haven’t already, make sure to check out Schema’s hierarchy to get a clearer idea of what this looks like in practice.
So in order to come up as a result for a particular query your brand name must not only be established as an entity, but also meet the conditions prescribed by the rest of the query. We’ll try and take a closer look at both of these requirements, but before you go further, you might want to check out this great post by Barbara Starr that clarifies the difference between implicit and explicit entities, and generally, gives you a broader perspective on the subject.
Let’s take a look at some of the things that you can do in order to establish yourself or your company as one of the prominent industry authorities:
- Semantic markup, but this deserves its own paragraph so we’ll get to it later
- Distinguishing yourself on any platform that Google considers a reliable source of information and extracts data for their Knowledge Graph from – Freebase and Wikipedia being some of the more popular options (however, they are mostly suitable for establishing a person as an entity, and don’t work quite as well (at all?) with brands or companies).
- Verify your identity to help the search engines make the connection between your various online assets – i.e. recognize everything that is a part of the entity you are trying to define. One of the ways of doing this is finding venues where you need to verify your data, like reputable local directories, guest posting on industry recognized blogs, or anything along those lines. Different social networks, especially G+ are also a great way to give Google a clearer picture of what you’re all about, especially after recent changes to Knowledge Graph.
Becoming a Relevant Entity
Once you are decently established as an entity, you want to ensure that you are relevant for queries that the demographics you’re after are likely to use. Here’s (in some fairly broad strokes) how to do that:
- If you wanted to be considered a relevant source of information some time ago, all you had to do was ensure that your pages mentioned the desired keywords enough times, and perhaps that you had enough inbound links with the right anchor. Today, you are not trying to get your site to be relevant for the keywords that users might choose, but for the actions they wish to take, i.e. for the triplets they might use to get the desired result.
In order to do this, you must understand how your potential clients might formulate their query, what other entities would you benefit from being linked to, and ensure that you are not only covered when it comes to explicit signals (entered keywords), but for implicit ones as well.
- Naturally, much of what was once true stills stays in effect – you’ll still want to get connected with industry influencers (just make sure that the connection is obvious enough to crawlers) and relevant websites; and you’ll want to provide high quality, relevant content which will, if written properly, not only naturally contain the keywords you’re focusing on, but also all the needed synonyms and other co-occurrence triggers as well.
Implementing Semantic Markup
When considered in the context of the semantic web, semantic markup may seem like a revolutionary new idea, but the fact is that it is basically an elaboration of an existing practice. When you think about it – metadata didn’t come into existence with semantic markup. It was there in form of page titles, meta descriptions, and link alt text for quite a while. Remember, all those things that search engines used to use to determine the properties of content or a website before markup came along?
Semantic markup simply takes this approach further, to a micro level if you will, applying machine readable descriptions to page or content elements that were previously left undefined. One of the things that this allows us to do is provide search engines with additional information on different kinds of media that were previously difficult to describe, like images or video as explained by Bing’s own, Duane Forrester.
While pretty much anyone with any kind of authority claims that other languages will stay supported for some time to come, they also recommend going with microdata, a format used by Schema.org. If you need help with figuring out which one is the best for you, this page provides a great breakdown of specifics of either kind of syntax.
Implementation Challenges
While semantic markup does indeed seem to be the bee’s knees, the fact that people are rising almost unanimously in its defense makes us want to trash it at least a little.
Sure it can bring structure into the cacophony of voices shouting in unison about their cheapest products, five great ways to… or whatever it is that they want to share with the world. However, many people find semantic markup too complex to implement properly and are worried that they’ll only mess their site up. While this might be a bit of an issue, with the help of Google Data Highilghter, their Structured Data Testing Tool, and this great article on identifying entities on a web page, you shouldn’t have too much trouble detecting and fixing any potential issues with your markup.
Finally, perhaps the only real and (currently) unavoidable problem with semantic markup is the fact that while it may make it easier for Google to find your site and label it as relevant for a given query, it also enables it to simply scrape your content and directly provide the person making the query with an answer without them ever having to actually visit your page. Therefore, while you might get increased rankings, supplemented with those nice looking rich snippets, you might actually end up losing some traffic.
How All This Applies to Your Brand
With all the “science-based” data presented above, it is not really easy to segment the audience that should care about this talk the most. However, one thing is for sure: with search engines becoming ever smarter about the multiplicity of word meanings and contexts they appear in, brands that aim at establishing authority online should not only follow the how-tos listed above, but also work on developing a specific strategy around making their name meaningful to search engines. Of course, this may only be possible if both Google’s and the users’ intentions are taken into consideration, which yet again points to the importance of understanding the target market and the ways SEO works. Those that are capable of implementing such knowledge in their SEO strategy stand more chances of making their brands recognizable in the wide cyber space.
So, what kind of businesses should care the most about adding semantic markup to their HTML? Well, ultimately, all of them will, but until we’re there, the ones that should make this a priority include:
- Locally oriented businesses. Providing search engines with as much verifiable and consistent information on your service area and location as possible, will not only help in establishing you as an entity, but supply crawlers with a set of invaluable implicit signals that will ensure that you are considered for a wider array of relevant searches. It also helps businesses with multiple phone numbers or addresses specify which piece of information is relevant for which location or branch, preventing their various offices from competing with each other or confusing the search engines.
- Video marketing. We’ve already mentioned sites with an abundance of materials that were previously difficult to label, especially videos. Applying markup to this kind of media helps your material find its proper place in the grand scheme of things. This way, if the fame of one of your videos outgrows that of your site or brand, you’ll be able to harness more of its success and give viewers and crawlers a much clearer idea of where the credit is really due.
- Businesses that know the value of increasing conversions and click through rates and lowering bounce rates. Properly implemented markup provides users with additional information about your site, which means that they are able to make an informed decision on whether they want to go ahead and click that link, allowing you to increase the amount of targeted traffic you’re getting.
Conclusions
Undoubtedly, data structuring is the future of the web. Even though the favored syntax as well as the benefits for webmasters could change, it is currently the most efficient approach to making the ever-growing oceans of data more meaningful than what we have. The way things are now, chances are data structuring is only to become more important as people and brands continue to build their online selves and as mobile search becomes more prevalent.
About The Author
