Friday, June 24, 2016

Facebook announces four new mobile ad formats

Facebook the most mobile engagement of any platform, seeing more than 1 billion daily mobile users.

With that in mind, Facebook made four announcements at Cannes this week:

1. Launching the Creative Hub

With a simple interface and a guide to Facebook and Instagram ad formats, Creative Hub is designed to make it easy for users to sample different tools and features, and work together and experiment.

For instance, there’s a collaborative area for marketers to preview, evaluate and showcase their creative. There are also options to create and preview mocks on mobile, as well as create preview URLs to share with stakeholders.

Built with the guidance of several agencies such as Ogilvy & Mather, McCann and Droga5, Creative Hub is currently testing and should be available to Facebook advertisers in the next few months.

2. Upgrading Canvas

We’re sensing a pattern with Facebook, which initially announced its Canvas ads, immersive mobile experiences that load 10 times faster than typical mobile sites, in Cannes last year.

The product was launched globally in February and since then, people in more than 180 countries have spent about 52.5 million minutes – otherwise known as a century – viewing Canvas.

New updates will make it easier for marketers to design, create, share and learn from these ads. Canvas will have a new feed unit designed to increase engagement, while marketers will have more detailed metrics, such as clicks-per-component and dwell time (the average is about 31 seconds).

The option to create Canvases for organic page posts has already rolled out.

3. Adding Audience Insights API

Audience Insights API will give advertisers better insights into the audience they’re serving, using aggregated and anonymous demographics, psychographics, topic data and reports from Facebook IQ. Currently in beta, the feature is testing with brands like Mondelez and Anheuser-Busch InBev, and will be widely available early next year.

Mondelez used Audience Insights for Cadbury’s “Taste Like Joy Feels” campaign, analyzing people’s feelings toward chocolate at various times throughout the day. Brand recall was improved by 40 percent, according to Cadbury.

 4. Improving slideshow ads

Another popular Facebook ad format is the slideshow, which allows businesses to create videos from static images. However, they load significantly faster than traditional videos, on account of using five times less data.

New features include the ability to create slideshow ads from mobile devices, audio and text overlay, and integration with Facebook’s Pages and Shutterstock photo libraries.

That focus on video isn’t to say photos aren’t doing well by Facebook. Instagram announced yesterday that its user base has doubled over the past two years.

The platform now has more than 500 million monthly active users around the world, 300 million of whom use the app on a daily basis.

How to speak ‘Search Engine’

The challenge of how to ‘speak’ search engine and tell it how to surface our content is what Search Engine Optimisation is all about. But are we doing it as well as we could?

Christian J. Ward, partnerships lead at Yext, gave a webinar in partnership with Brighton SEO on ‘How to Speak Search Engine’, in which he looked at the current state of search and the problems inherent in how we produce the content that we expect search engines to find.

Search has changed dramatically since Google first began indexing the web in 1998, both in scale and in nature. Google alone executes more than two trillion searches every year – a scale that we can barely comprehend. Search, said Ward, is not just a process for a brand; it’s becoming the number one way that we interact with information generally.

But the way that we search has changed, too. At a recent CMA Digital Breakfast, digital journalist Adam Tinworth remarked that Google is becoming “much more of an answer engine” than a search engine – searches are increasingly phrased in the form of a question, and innovations like the Knowledge Graph and Featured Snippets aim to answer searchers’ questions without them needing to leave Google.

one great answer

We all want Google’s ‘answer engine’ to surface our content in response to searcher queries. One way to help ensure this happens is to write content that will satisfy questions that users might have when coming to our websites.

But even once we have, how can we direct Google and other search engines to the content that will provide the best answer?

Feeding baby Google

To illustrate a problem inherent with the way that we approach content online, Ward used an image which has to be the best depiction of ‘peak content’ that I’ve seen so far.

A presentation slide featuring a photo of an unhappy looking baby being fed with a spoon. The baby is wearing a bib with the word "Googoo" made up of letters from Google's old logo. To the left is a list of content types: Blogs, Ad copy, Featured articles, Webpages, Product write-ups, Menus, blurbs, Services, Lists. Underneath this the text reads "Unstructured..." and then "YUCK!"

These days, brands and websites are churning out more content than ever before in an effort to keep up with each other: blogs, ad copy, sponsored content, product write-ups, ordinary webpages and lots more.

“We’re trying to feed Google – the baby – great content information that, to some degree, it doesn’t want,” said Ward.

At least, not in a form that it can’t easily interpret.

“We pump out so much content that it is very difficult for Google to analyse it and to know what we’re talking about. And it’s partially because it’s unstructured content.”

As an example of how confusing this can be in practice, Ward looked at the search term “tombstone”, which has a whole array of possible meanings: Tombstone is the name of a popular 90s Western; it’s the name of a town in Arizona (for which the film was also named); a word meaning ‘headstone’ or ‘gravestone’; a brand of pizza; a Marvel comic book, and more. Which of these is going to be most relevant to the searcher?

A Google search results page for the keyword "tombstone". A drop-down list below the search bar shows the suggested searches "tombstone - film" and "tombstone - city in Arizona" as well as "tombstone cast" and "tombstone pizza". The search results mostly relate to the film Tombstone, and also include a Twitter user named TheLivingTombstone.

Of course, part of the game here is trying to guess what the searcher intends when they search for the word “tombstone”. But in our content, as well, we have to make it clear which “tombstone” we’re referring to, so that Google can more easily hone in on the right content and serve it to the user.

If you have a webpage about tombstones, and Google can’t tell whether it’s about headstones or pizzas, it won’t be able to show it to a user who is searching for one or the other.

Search engines want to provide their users with more rich data in search results: useful information like event dates, reviews, menus and other details that can answer their query at a glance, or at least help them decide which result will be the most relevant.

Ward quoted Sundar Pichai, the CEO of Google, who in his keynote speech at Google I/O, said,

“It’s not just enough to give [users] links. We really need to help them get things done in the real world.”

Ward believes that Google is working towards an eventual solution which means users will never have to open an app or website.

While this sounds like a very distant future (after all, there are bound to be some circumstances in which users are searching in order to find a website or app, not just an answer from Google), there’s no denying that Google has taken a huge step in this direction in recent years.

Putting definition around the cow

So what can content creators do to move with this trend, and set their websites apart from everything else in the vast sea of online content?

Warner showed a black-and-white photograph, which has been used by Ellen Langer in her work on mindfulness, and asked webinar attendees to volunteer what they thought it was a picture of.

cow illusion

Suggestions came back: a turtle, a skull, the Hindenburg. But when a few guiding lines were added to the image, the subject became clear: it is in fact a picture of a cow.

cow illusion 2

“Now that you see it, it’s impossible to unsee it,” said Ward. “There’s a lot of relationship around that, where just a little bit of definition can burn a pathway. And search works a lot like that.”

In other words, content creators need to put that bit of ‘definition’ around their content in a way that tells search engines what it represents, and what type of content it is. There’s a way to do this using search engine ‘language’, and it’s called structured data.

Structured data has been around for a few years now, and is known as a way to help search engines assess and understand content in order to better place it on the SERP. Yet in spite of this, a shockingly low proportion of website owners actually make use of it.

The Schema.org logo, consisting of white sans-serif text reading "schema.org" on a dark red background, with a slight shadow around the letters.

Take Schema.org, a markup language that is the result of a collaboration between Google, Bing, Yahoo! and Yandex to create a structured data vocabulary that can be understood by all search engines.

A study by Searchmetrics in 2014 found that 36.6% of Google search results incorporated Schema rich snippets, yet only 0.3% of websites actually made use of Schema markup at all.

The study also found that pages which used Schema ranked on average 4 places higher in search than pages which didn’t, although Searchmetrics was keen to emphasise that this might not be entirely down to structured data.

But search results which use Schema are widely agreed to result in higher click-through rate, as they include more useful, relevant and attractive information like pictures, reviews, opening hours, pricing information and more.

So since this study was conducted two years ago, has the number of pages marked up with Schema increased significantly?

Ward did some quick calculations. The Schema.org website proudly proclaims that “Over 10 million sites use Schema.org to markup their pages and email messages.”

A slide from Christian Ward's webinar with white text on a dark background. The title is "Really? 10 Million?" and the text reads, "We passed one billion websites in September of 2014, and it's closer to 1.08 billion today. 10,000,000 divided by 1,080,000,000 = 0.926%. Less than 1%. Nice work, everyone!"

While this figure might sound impressive, it becomes less so when you realise that we passed one billion websites in September 2014, and the number today is closer to 1.08 billion. 10 million as a percentage of 1.08 billion equals… 0.926%. That’s an increase of only 0.626% since Searchmetrics’ study, and still less than 1% of the total websites out there.

“It’s staggering,” said Ward of the low number, “when you think of the ramifications of how much better search does when we can explain it.”

It’s not easy speaking search engine

So then why do so few website owners and content creators use Schema markup on their sites? “There’s a good reason for this,” Ward said. “We all know this is hard work. I don’t think it’s that we mean to be lazy, I just think that ultimately this is very hard to do.”

Until quite recently, for example, all Schema markup code had to be added in-line around the individual elements of the page.

Every element, from addresses and opening hours to reviews, needs to be defined individually with Schema, resulting in a lot of coding legwork and no small amount of headaches when it came to fitting it in with all the other code already on the page.

restaurant schema example

Just like any other language, learning to ‘speak’ search engine is going to require a lot of investment of time and effort. But, Ward maintains, it is definitely worth our while.

“This effort is a way to truly distinguish the work that you do and the work that our community does on behalf of our customers and clients. It just takes a lot of time.”

He pointed to the example of a search for the latitude and longitude of the Empire State Building, the answer to which is displayed in Google’s knowledge graph at the top of the search results page.

The website which provides this information uses Schema.org markup to point Google to where the relevant content is on its page, resulting in the “great user experience” of “one solid answer.”

lat-long schema example

And best practices around structured data are constantly evolving, making it easier for website owners to incorporate it into their code. Google used to only support Schema markup if it was written inline, insisting that the markup needed to be “visible to human users” as well as search engines.

But it has since reviewed this stance and expanded its support for a type of notation called JSON-LD, which allows structured data to be added to the header and footer of a page instead of inline.

Google’s introduction to structured data on Google Developers now states outright that JSON-LD is the recommended markup format for structured data.

“Schema, its use, and the taxonomies – they’re evolving constantly,” said Ward. “We have to get more involved in this process, as a community. We need to be working with Google, and with Yandex, and Yahoo! and Bing.

“Let’s start banding together to try and get some structure out there.”

A screencap of a Siri voice search asking "How big is the Serengeti?" Siri's answer pertains to the breed of cat, answering "Medium", rather than to the region in Africa.

If you need just one more reason to start incorporating structured data into your website markup, it should be the rise of voice search.

Ward cited statistics from Mary Meeker’s recently-released Internet Trends report which show that the volume of Google voice search queries is now 7x what it was in 2010, with 65% of smartphone owners using voice assistants like Siri, Cortana and Google Now.

Users are getting used to being able to ask their voice assistants increasingly specific questions and get a single, definitive answer; but to make this possible, website owners need to be adding the structural markup around their information that will tell the assistant where to look.

“In the end, I want to be able to ask Alexa to email me the logo of the local 7-11, or, ‘Can you tell me if this place is closed or open right now? Do they have any specials right now? What’s the number one item on their menu?’” said Ward.

“All of that data has to be incredibly well-structured in order for us to get the result we’re looking for.”

Five most interesting search marketing news stories of the week

Welcome to our weekly round-up of all the latest news and research from around the world of search marketing and beyond.

This week we have as many updates from Google as you can stand and some hilarious/depressing social shenanigans.

Let Google take care of you, there there, this won’t hurt a bit

Dr Google is on the ward and ready to roll out a few new procedures. According to a blog post this week the FIND A TERM OTHER THAN ‘SEARCH GIANT will provide searchers with immediate medical advice in its Knowledge Graph answer box when searchers type in particular symptom queries, which apparently accounts for 1% of all search terms.

Symptoms Search

Now I was fully prepared to suggest this was a terrible idea, but as anyone who suffers from health anxiety can attest, the internet is a terrifying place when it comes to self-diagnosis.

Every search for even the most minor of ailment tends to lead to a cancer diagnosis. However Google recognises this fact and is working with trusted doctors and high quality medical information to bring you these results.

Rather than this guy…

dr nick billboard

Google Analytics will now warn you about Hackers

According to its Webmaster Central Blog and following on from the launch of Safe Browsing, a service that warns users of malware or phishing attacks, Google will expand its set of alerts in Google Analytics by adding notifications about sites hacked for spam in violation of its Webmaster Guidelines.

“In the unlikely event of your site being compromised by a third party, the alert will flag the affected domain right within the Google Analytics UI and will point you to resources to help you resolve the issue.”

compromised_sites_suspected

Google also revealed that it has seen a 180% increase in sites hacked for spam compared to the previous year, however direct contact with website owners increases the likelihood of a fix by 75%.

Google won’t however warn you about watching the movie Hackers, there are plenty of other online resources for that, most of which will tell you it’s dated horribly.

And it really has.

Search Console adds new ‘rich results’ filter

Following on from the addition of a rich cards section to its Search Console service, Google has now also added a ‘rich result’ filter to its Search Analytics.

Just navigate to Search Traffic>Search Analytics, then click on Search Appearance filter to select the ‘rich result’ option.

google-filter-rich-results

This will tell you how well your rich snippets and cards are doing in terms of impressions, clicks, CTR and position.

Thanks to Danny Sullivan over at SEJ for the info and screen grab.

Facebook is testing a new way to make damn sure your friends see your posts

As I reported earlier in the week, Facebook has been trialling out a new way of notifying your friends directly about your status updates.

facebook notify friends

Although only rolled out to a handful of people in the UK, Canada and Spain, this is an interesting experiment, that means you can nudge up to 10 of your friends and say, “Hey look at me, why don’t you pay attention to me anymore? Is it because of spammy behaviour such as this?”

We’ll see how long it lasts.

Just because they’re sharing, it doesn’t mean they’re reading

And finally that hilarious/depressing research I warned you about. As reported by us this week, research has revealed that only 41% of people actually read the links they share on social.

Basically 6 out of 10 people just click retweet upon seeing a headline (and less than only 140 characters of context) and nothing more.

Despicable behaviour I’m sure you’ll agree.

Thankfully we here at SEW have incredibly attentive readers who will conscientiously read every single word we write before sharing it, and who are also incredibly attractive, brave, generous and I think I can wrap this up here, nobody will notice if I just fill the rest of my word count with some ‘hipster ipsum.’

Messenger bag vice whatever biodiesel affogato pug 3 wolf moon beard bushwick celiac art party flannel. Flexitarian jean shorts offal, celiac tofu chicharrones retro chia fingerstache gastropub asymmetrical dreamcatcher yr pop-up kogi. Craft beer seitan salvia, typewriter organic photo booth.

 

Thursday, June 23, 2016

How to keep the ‘person’ in ‘personalization’ without being a creep

Some brands don’t target well enough, while others go way too far, creeping people out. With personalized marketing, striking a delicate balance is the key. 

Personalization is an important skill for any marketer to master, but it’s also quite a difficult one. There are just so many different ways it can go wrong.

If your ads don’t have any personalized components, people perceive them as not being relevant enough, making them that much more likely to use an ad blocker (provided they’ve heard of ad blockers).

But on the flip side, it is possible to go overboard with personalization. If people perceive your brand to be like Big Brother, they’ll be just as turned off.

As with many things, the answer is somewhere in the middle. If you’re looking to personalize your ads, the most important thing to remember is that the root of that word is “person.”

The pitfalls of getting too computerized

Jan Jensen, chief marketing officer (CMO) of Cxense, points out that as marketing gets more complicated, the sophistication of automation technology makes it easier for advertisers to get away from the person.

“In this day and age, where we have more moving pieces, it’s very complex. Being able to know the challenges, pain points, grievances and profiles of your audience is 20 times more important than it was five years ago,” says Jensen. “Depending on what people do, the data we have, their interest and intent around the content they consume and the profile we have on them, we can predict what they want next.”

A businessman is consulting a crystal ball to foretell the future.

As a result, things can get a bit too automated. Marketers often create a journey for customers, automate it, and move onto the next thing. But in the process, they’re assuming too much. There should be a balance.

For instance, artificial intelligence (AI) lacks the human emotions to realize it’s doing something insensitive. A robot would serve someone tons of ads for diet products, whereas a person could see how doing that constantly could hurt the customer’s feelings.

Jensen believes this extends to all the brands who have been using chatbots as customer service tools.

“I think chatbots have their place somewhere extremely straightforward, but human beings aren’t very straightforward,” says Jensen. “I think they need to be much more aware of who you are; it needs to know more than that I’m a male and my name is Jan.”

… as well as too creepy

On the flip side, tech companies like Amazon and Facebook know far more about us than our names and gender. They may even know us better than some of our loved ones; does your mom know what you Google? (Ew, she does? Gross.)

But you have to be careful about showing your cards. If you let people know just how much you know about them, more than being simply freaked out, they can hurt your ROI. Richard Sharp, chief technology officer at Yieldify, points to a concept called psychological reactance that will be familiar to anyone who’s raised – or been – a teenager.

pixar-slamdoor

“If people feel their behavioral freedom is being restricted or manipulated, they will explicitly react against that in order to restore their freedom,” explains Sharp. “A lot of research shows that when people perceive creepiness online, it results in this feeling of, ‘You’re trying to manipulate me and force me to take this cause of action,’ which causes people to react against the brand, which decreases purchase intent by 5 percent.”

The way around that, according to Sharp, is to be very upfront about the value proposition. If an email from Bloomingdale’s is highly personalized, people may be a bit taken aback. But if the email has information about a sale at the nearest Bloomingdale’s location, they may perceive it differently.

“If you trigger a really well-designed campaign at the point when someone is about to leave the site or stop browsing, it catches people’s eyes and draws them back to the site,” adds Sharp. “It converts really well without interrupting the customer journey or annoying people.”

In order to keep the person and personalization, Sharp recommends user-centric research.

“A lot of marketers like looking at numbers and click-through rates and conversions and cost-per-acquisitions, which is a view that takes the human out,” he says. “Get some qualitative data to back up this quantitative data so the human side doesn’t get lost behind a wall of numbers.”

Google enters the Artificial Intelligence race with Magenta

The words Artificial Intelligence can bring to mind far-fetched, sci-fi ideas and a society where robots have replaced humans. Well, this idea may not be too far off given Google’s recent innovations.

Google recently released Magenta, a computer based system that has the ability to create pieces of music.

Even though its first melody sounds like a generic song pre-programed to a keyboard, the project is considered a success because the system taught itself. The system composed the 60-second melody with little human intervention.

Google engineers only provided Magenta with four notes to begin the process. They also added drums to add a bit of flair to the song.

Magenta is a project from the Google Brain team that questions the traditional view on computers. In the past, computers were generally seen only as an electronic device used for storing and processing data. But now, Magenta questions all of that.

With Magenta, the Google Brain Team is asking ‘can computer can learn to create compelling music all by itself?’

According to a blog post from Google, Magenta comes with two goals. First, it stands as a research project to advance the state of the art machine technology for music and art generation.

Second, the Google engineers hope that with Magenta they will create a community of artists, musicians, developers, and machine researchers.

While Google is the first to admit the program is in its infancy, the company is hopeful of its future potential. The company writes:

“We don’t know what artists and musicians will do with these new tools, but we’re excited to find out. Look at the history of creative tools. Daguerre and later Eastman didn’t imagine what Annie Liebovitz or Richard Avedon would accomplish in photography. Surely Rickenbacker and Gibson didn’t have Jimi Hendrix or St. Vincent in mind. We believe that the models that have worked so well in speech recognition, translation and image annotation will seed an exciting new crop of tools for art and music creation.”

This isn’t Google’s first rodeo with artificial intelligence, but the company sees Magenta as a stepping stone into the world of natural language processing. This move comes after Google noticed that more and more searches are being done by voice. Consequently, users expect their machines to understand the context of their commands.

Aritifical intelligence stat IDC

Sundar Pichai, CEO of Alphabet’s core Google division, explains “We think of this as building each user their own individual Google. Google does a lot of things, but if you peel away everything and you distill it, this is the heart of what we do. It’s what we are about.”

But Google isn’t the only tech company in the AI game. Earlier this year, Microsoft machine learning technology drew a Rembrandt painting through a 3D scanning device that gathered information from 300 plus paintings. The result was an original, unique self-portrait of the Dutch artist.

IBM has been working since 2005 to develop a supercomputer named Watson and Google’s Android has created an ‘open-ecosystem’ that will let users incorporate different technologies on one domain.

Even Space Exploration Technologies Corporation (SPACE-X) believes that these developments in artificial technology will create computers so sophisticated that humans will eventually  need to implant “neural laces” in their brain to keep up.

But the true fight to be the best in artificial technology is between Google and Facebook.

Recently, Google released to the technology world that their program AlphaGo was able to beat the ancient game ‘Go’, which has long been considered the most challenging game for any artificial intelligence to learn.

Within days, Facebook mentioned that they were close to achieving the same success which demonstrated their seriousness in joining the AI race.

The social network also introduced its Deep Text understanding technology recently. The innovative software can understand the textual context of thousands of posts per second with almost human accuracy.

Deep Text can span over 20 languages, and Facebook plans to use this technology to improve user experience in different ways. It wants to help identify the best quality comments in a public post, improve transaction performance, and increase its Messenger app’s user-friendliness.

Both companies have completely different methods to achieve success in the AI world. While Facebook is generally concerned with improving its users’ experience, Google is hoping to integrate AI into all aspects of their services.

So where does this leave us?

AI stat IDC

According to the International Data Corporation (IDC), by 2020, the market for machine learning applications will reach a whopping $40 billion and about 60% of those programs will run on the platforms of four different companies: Amazon, Google, IBM, and Microsoft.

Both Facebook and Google are choosing to attract advertisers to their platforms by offering compelling techniques that will give marketers a better return on their ad spending. This action will undoubtedly make technology more accessible and improve its adaptability.

For Google, it hopes to change the artificial intelligence game. The company knows it won’t happen immediately, but its goal is to make their intelligence software applications widely accessible.

The IDC estimates that, by 2018, at least 50% of developers will include AI features, so it is clear that Magenta is just the springboard for bringing artificial intelligence mainstream.

As to how this news affects the SEO world – well, Penguin still hasn’t reared its head. Could it be that the update’s AI isn’t done learning yet? If so, how smart will it be?