Back to Searchscape

Machine learning in 60 seconds

Machine learning is a field of computer science that uses algorithms that iteratively learn from data. It has many applications for the web, including image recognition, email spam filtering, sentiment analysis and, of course, improved search engine results. Machine learning is one branch of Artificial Intelligence.

The topic is becoming a more and more important part of everything Google does, but can seem quite inaccessible to learn about. While understanding the technical side of machine learning is not something most marketers will need to do, understanding the use cases and the future outlook can put you in a strong position. Another reason that machine learning is hugely important is that it is so vital to the development of a number of our other predictions for the future of search, such as natural language and compound queries.

Google now has an extensive machine learning infrastructure, built around its Tensor Flow AI system, which is open sourced for public use and experimentation. It's so ingrained in the future of Google that there is no doubt that machine learning and artificial intelligence will continue to shape the way we use the internet.

Tom Anthony's Brighton SEO presentation offers a great introduction to machine learning.


"Machine learning is a core, transformative way by which we’re rethinking how we’re doing everything,"

Sundar Pichai, CEO of Google

Intelligent Personal assistants in 60 seconds

An intelligent personal assistant is another Searchscape topic that has been around for some time, but for various reasons (primarily technology) they are becoming increasingly important in the search industry. In the past, they've been seen as clunky, but with the rise of machine learning, the newer breed of software is considerably more advanced than their predecessors.

The main players in the intelligent personal assistant space include Apple's Siri, Google Now, Amazon Echo, Microsoft Cortana, along with up-and-coming competitors Hound and Facebook M. To take Google Now as an example, it recognises actions performed by a user to provide relevant information, such as event reminders, how long it takes to get home, restaurant reservations and much more.

However, it's what's yet to come that is most exciting. Google has stated repeatedly that it wants to become the ultimate personal assistant. We can see this happening laterally with things like their Self-Driving Car project and Nest. We need to stop thinking about Google as a mere search app and focus on the bigger 'intelliegent personal assistant' picture.

As Google and other personal assistants become more and more sophisticated, and have access to even more data, we will increasingly search both personal and public corpuses using just one interface. This interface will be whichever personal assistant can provide us with the best answers.


A selection of some of the most popular intelligent personal assistants.


"So much of what Google do makes so much more sense when you think about it in the frame of reference as them being the ultimate personal assistant."

Tom Anthony - Head of R&D, Distilled

Compound queries in 60 seconds

A compound query is the process of submitting a search query and receiving an answer from Google, then submitting another query, whereupon Google takes cues from your first search to provide you with a better answer.

There are several types of compound query. An "intent revision" query for example, might be if you started by asking "show me cookbooks", then revised the response with a second query of ‘show me the vegetarian ones' which inherently refers back to the previous query. In the previous model you would have instead run a second, independent query, such as ‘show me vegetarian cookbooks’.

Another example of compound queries is "chained queries", which work in a similar way, but rather than revising the initial query you instead run additional queries around the same topic. For example if you first asked "how tall is the Empire State Building?", you could follow that with a chained query, such as ‘When was it built?’. You’re not revising your query, but it is dependant upon having asked the first query.

We’re currently only seeing this in conversational search, but we expect it to migrate to desktop in the future. Compound queries are hugely important for the young generation of seachers who’ve never been trained to search in a keyword-oriented way.


The updated compound query model.


"The move towards compound queries will mean it becomes more natural for people to use Google to 'interact' with data in an iterative process; Google won't just send us to a set of data somewhere else but will help us sift through it all."

Tom Anthony - Head of R&D, Distilled

Implicit queries in 60 seconds

Users of Google have long been used to the idea that the words and phrases typed into the search box would directly match words and phrases on the web pages that appeared in the search results. Over time, this has become less true. So, if this time-honoured process isn’t happening, what is? Implicit queries.

As Google has become more sophisticated, it’s begun to take more and more factors into account in your search, many of which you, as a user, never explicitly say or type. These are called the implicit aspects of the query. They can include the likes of what device you're using, your location, search history, mode of transport and more.

The main reason this has become a more prominent trend and will, we believe, continue to be so, is down to Google having more data, which enables them to give better results. This is primarily due to the number of smartphone users, and Google's increasing ability to understand the growing list of data points.

In the near future Google will almost certainly continue to give further weighting to the implicit signals of your search query. Google Now on Tap is an example of this process happening right now. It will be exciting to see how far implicit queries can go.


Examples of implicit and explicit query aspects, along with the trend towards more implicit factor-heavy queries.


"We predict that as the search engines have access to more and more information, and as their ability to use that information intelligently improves, we're actually going to see users relying on this more, and using shorter explicit queries."

Will Critchlow - Founder & CEO, Distilled

Natural language in 60 seconds

It's likely if you're reading this, you're part of the generation who grew up trained to search in a very particular, keyword-oriented way. The thing is, this isn't how we interact with other people, either on the web or in person; it's not a natural way to speak or ask questions. We were forced into this way of searching due to the limitations of Google and other search engines. However, as search engines have gotten smarter, and have much more sophisticated natural language processing (NLP) capabilities, we're now presented with some exciting opportunities and interesting challenges.

If you look at how children use search engines, or how new users interact with voice search, you'll see natural language processing in action. For example, a desktop search for finding out the weather in London, you might simply search "weather in London" or even "London weather". On Google Now the same intent may result in the query "Ok Google, what's the weather going to be like in London?". This second example is much closer to what we'd actually say to another person.

As more searchers grow up having never been 'taught' to search, and existing users learn the capabilities of natural language searching, we will see queries become more complex. While it's difficult to predict exactly what marketers should be doing in this regard, structured markup is an excellent place to start, because search engines will need to understand entities associated with your content, so they can provide users with direct answers.


A humorous example of natural language searching from a 5-year old. Image source.


"We haven't yet seen the watershed moment in regards to natural language. That will come when more people realise that you can search in this way and that you can do more complex queries."

Tom Anthony - Head of R&D, Distilled

Site speed in 60 seconds

While site speed is far from a new topic - it’s been a ranking factor for some time, as well as an integral UX consideration - there are a number of reasons that the importance of site speed will begin to increase.

As users have become used to much faster internet speeds, they’ve also become accustomed to slick, lightning-fast web experiences. While we, as marketers, might not be expected or able to directly fix site speed, we can still measure our site to spot problems and keep abreast of all the new technologies, many of which will appear in marketing news.

Tools like URLProfiler let you run Google Page Speed tests en-masse across your site. Running comparisons against major competitors can help create a good benchmark for the kind of performance customers expect in your industry

There are also a number of technologies on the near horizon that you can be reading up on in preparation. Server-side enhancements, like HTTP/2, which will improve download speed are beginning to be adopted and Google’s Accelerated Mobile Pages (AMP) project is due to go public in 2016, and is aimed at making it easier to build fast mobile websites.


The Accelerated Mobile Pages (AMP) project in action on Cosmo. Image source.


"We expect to see users value site speed more and more, especially as they get used to things like Google AMP, Facebook Instant Articles and Apple News. This could mean speed will become even more of a ranking factor."

Will Critchlow - Founder & CEO, Distilled

App indexing and deep linking in 60 seconds

App indexing is the process which allows your app’s content to appear in mobile search results if it is relevant to the searcher’s query. The result would link directly to that content “view” within the app, using “deep linking”. At the moment the main player in app indexation is Google, but other search services such as Apple Spotlight are implementing processes to do this for their search results as well.

Deep linking is the term for setting up links which, when clicked, open content “views” (like webpages) inside the app, rather than simply being directed to the opening screen of that app and having to navigate from there.

The main reason app indexing and deep linking are becoming such a major part of search is that standard web crawling extends poorly to apps, which have been traditionally seen as ‘walled gardens’. The introduction of deep linking, and in particular the option to support HTTP deep links for in-app content has seen app indexation become a viable process for search engines.

The importance and application of these two processes will continue to grow. For example, Google has just released app streaming testing, which allows for a seamless journey between web and app, by negating the need to download the app in the first place.


With app indexing, Google can display a link to a specific page within your app. Image source


"As app and mobile web usage continues to rise, the ways in which people access this content is beginning to converge, which means it's becoming more important to keep all of these different content locations linked up."

Bridget Randolph - Consultant, Distilled

Keywords to intent in 60 seconds

Back in the rose-tinted days of the early 2000s, Google was a simpler place. In general, we'd type a search query, being sure to include a number of keywords that related to what we were looking for. As you'll see in our natural language section, searching in this manner is not how we'd actually ask for things from another person. As the likes of implicit queries, compound queries and natural language come into play, we have to spend more time ensuring we understand the intent behind a search query, rather than just the keywords used.

This shift away from keywords has stemmed from a number of algorithmic and technological advances. For example, the Hummingbird algorithm is designed to understand the relationships and make-up of queries and the words contained within, making conversational search more powerful. Also, the rise of smartphones have helped determined context (location, mode of transport, etc.), which would have previously needed keywords to make clear.

To optimise your website to keep up with the shift from "strings to things", it's important to focus less on search volume and instead put yourself in a position to be the right answer for the right person at the right time. Google's EAT (expertise, trustworthiness and quality) guidelines are hugely important here, as content will need to move away from highly targeted long-tail landing pages to broader and higher-quality content.


It's important to understand the various factors, including keywords, that signal the user's intent.


"When you create content, assume that Google will keep asking "why?" Intent is all about the why."

Craig Bradford - VP London, Distilled

Cross-device browsing in 60 seconds

We're well past the point of mobile users overtaking the number of desktop users, and with the likes of tablet PCs, and possibly smartwatches, it's no longer realistic to equate a browser with a person. People use multiple devices, and will likely visit your site on a number of these.

When looking at cross-device browsing, it's most important to consider the people behind the numbers, and how they are experiencing and sharing your product. While you might find a great hotel website on your desktop, when you share it with a friend who opens it on a poorly-designed mobile site, their opinions and subsequent actions will be very different.

User testing is an excellent way of getting behind the numbers, but it's also important to improve your analytics understanding too. If you're site heavily involves users logging in, then User ID will help you track how and when they are visiting your site. Beyond that, you simply need to keep track of how your site handles different user flows, such as by ensuring landing pages are optimised for particularly important devices.

Cross-device browsing is already an important piece of the SEO puzzle, and will only continue to become more important as devices become more numerous and fragmented.


Consumers take a multi-device path to purchase.


"This is one of the most pressing Searchscape predictions, as to a certain extent it's already here. Cross device browsing's importance will continue to grow, as mobile isn't going anywhere, and cross-device tracking will be needed to keep tabs on it."

Tim Allen - Consultant, Distilled

Personal index searches in 60 seconds

The personal index refers to a user's own content, including things such as emails, photos, apps, appointments and more, which can appear alongside public content in search results.

With the release of Google Now and Spotlight, users are becoming accustomed to finding answers on one screen, rather than looking around multiple sources on their phone for all this data. Plus, Google Now on Tap and Proactive Siri, means users are expecting to be given information without even searching, and the personal index can be tapped into to allow these apps to do exactly that. For example, Google Now can send push notifications on when to leave for a meeting, taking into account travel time.

The personal index is simply another facet of search engines becoming intelligent personal assistants. As marketers, there are a few things we can be doing to help our content become a part of this personal index. For example, structured markup for email, while still requiring whitelisting from Google, can allow emails to appear in a user's search query.


An example of a personal index search appearing in a Google search result. Source: Google.


"We are starting to see the personal index (or personal corpus) become a more important part of search; as a first step we should be thinking about opportunities to integrate client data into searchers' personal indexes such that we can serve this new type of search query."

Tom Anthony - Head of R&D, Distilled

New devices in 60 seconds

On the face of it, new devices might seem like a vague subject, but as we've learned from the introduction of the smartphone and tablet computer, it would be foolish to dismiss the swathe of new ways of accessing the web as fads. The questions we must try to answer are "which of these devices will become popular enough to incorporate into a marketing strategy?" and "how do we better understand the users of these devices and their intent?"

While the amount of new technology, and existing products with new ways of connecting to the internet and to each other (often referred to as the Internet of Things), is ever expanding, there are a few that are particularly exciting. These include the smartwatch, which Apple have added to their product line-up; and smart home technology (like Nest and Hive).

Taking smartwatches as an example, it quickly becomes clear that the challenges of reaching customers with new devices are multiple. The obvious one is the different interface. The super-small screens require to-the-point content. It also adds another fragment to cross-device browsing, and as smartwatches often act as accessories to smartphones, the time available to grab attention is very low.

New and interconnected devices have the potential to be game-changing, not only from a marketing point-of-view, but also at the heart of how we interact with the world. As smart home technology becomes more readily available to consumers, and technologies like augmented reality come to fruition, we'll see changes that go well beyond the tactical challenges of optimising for smartwatches.


The smartwatch is just one of the many new devices marketers may need to contend with. Image source: Apple


"The rise of new devices is analogous to the rise of smartphones. It's primarily about context and user intent. A lot of contextual information can be gathered from these devices, and used to serve users relevant search results."

Bridget Randolph - Consultant, Distilled

Ad blocking in 60 seconds

Ad blocking is one of the topics in Searchscape that feels as though it has been around for a long time; and to a certain extent that is true - Adblock Plus has been around since 2006. However, there has been a recent uptake in the use of ad blockers, particularly on mobile. iOS 9, the latest Apple mobile update, has allowed developers to create ad blocking capabilities on the iPhone; something which had not previously been possible.

Ad blockers, at their simplest, are pieces of software that prevent browsers from loading PPC ads, email ads, display ads, video/YouTube ads and many common tracking setups. Users of ad blockers are often looking to avoid intrusive content, improve load times and eliminate the complication of deciphering ad content from non-ad content.

Will your brand be affected by ad blocking? It depends on how heavily you use ads. While you shouldn’t reactively abandon using adverts, you’ll want to consider how much organic marketing you do alongside, as well as the CPC for your ads.

The future, in terms of ad blocking, looks like a compromise. Most ad blockers have whitelists, where they will allow certain advertisers to continue showing adverts that meet certain criteria. Going down this route will allow you to continue advertising, while also (in all likelihood) serving ads that your audience won’t find ‘spammy’.


The growing global cost of ad blocking. Data source.


"What's critical to understand [is that ad blocking] is not just about the percentage of people that have ad blockers but rather the percentage of quality ads that are blocked. Those are two very different things."

Jared Belsky - President, 360i

Back to Searchscape

Enjoying Searchscape?
Get more Distilled advice and tips in your inbox:

Header image background: Taken Under the "Wing" of the Small Magellanic Cloud, courtesy of NASA.