Google and the Tip of the AI Iceberg

I was surprised by the amount of surprise in the technology community when Google recently announced that its RankBrain artificial intelligence (AI) is being used to help understand natural language queries and serve results.  This should come as a surprise to no one.  Google has always been an AI company.

This statement from Larry Page is about as clear as it gets (emphasis mine).

Around 2002 I attended a small party for Google—before its IPO, when it only focused on search. I struck up a conversation with Larry Page, Google’s brilliant cofounder… “Larry, I still don’t get it. There are so many search companies. Web search, for free? Where does that get you?”… But Page’s reply has always stuck with me: “Oh, we’re really making an AI.”

For most people, the term “AI” tends to conjure up ideas of an all-knowing computer program that understands humans completely, and returns answers in a context that is immediately understandable to humans, like the Star Trek computer.  At SXSW in 2013, Google’s Head of Search, Amit Singhal said, “The destiny of search is to become that ‘Star Trek’ computer and that’s what we are building.” In fact, Singhal recently demoed a Star Trek-like lapel pin that interacts via voice with Google Now.

Even if you don’t take public statements and product demos very seriously, you can simply look at where Google has been investing in creating IP.  Here’s a count of research papers published by Google.  You’ll notice that Artificial Intelligence & Machine learning have 143% more published papers than areas like Information Retrieval & the Web, which is at the core of traditional approaches to search.

google research papers by topic

However, despite this massive amount of AI and machine learning (ML) work being done at Google, it’s just in the last year that RankBrain is being used to help field queries.

Google have been working actively on AI & ML in relation to search for a long time now (in Internet years).  In 2008, Anand Rajamaran had a discussion with Peter Norvig (former Director of Search Quality at Google and author of Artificial Intelligence: a Modern Approach) about ML (emphasis mine).

“The big surprise is that Google still uses the manually-crafted formula for its search results. They haven’t cut over to the machine learned model yet. Peter suggests two reasons for this. The first is hubris: the human experts who created the algorithm believe they can do better than a machine-learned model. The second reason is more interesting. Google’s search team worries that machine-learned models may be susceptible to catastrophic errors on searches that look very different from the training data. They believe the manually crafted model is less susceptible to such catastrophic errors on unforeseen query types.”

And this has been corroborated by other sources.  A former Google Search Quality engineer had this to say on Quora in 2011 (emphasis mine):

google search quality and machine learning

I want to call out this sentence in particular (emphasis mine): “In a machine learning system, it’s hard to explain and ascertain why a particular search result ranks more highly than another result for a given query. The explainability of a certain decision can be fairly elusive.

This is a result of how machine learning models work.  ML models optimize for accurate predictions, and don’t care much about why they are accurate.  From HBR:

“One important difference from traditional statistics is that you’re not focused on causality in machine learning.”

This is very important.  ML models are just interested in the best result, not an understandable explanation of how each ingredient in the recipe contributes to making that recipe so delicious.

ML models decide what variables to use, and sometime build their own variables called “features” in order to make better predictions.

“Think of “feature extraction” as the process of figuring out what variables the model will use. Sometimes this can simply mean dumping all the raw data straight in, but many machine learning techniques can build new variables — called “features” — which can aggregate important signals that are spread out over many variables in the raw data. In this case the signal would be too diluted to have an effect without feature extraction.”

This leads to the “black box” problem that Edmond Lau pointed out.  ML models can build their own synthetic metrics, and do not explain causally why a particular combination of metrics leads to a better result.

From Amit Singhal’s perspective, this leads to team that lacks “direct control” of its own algorithm, which makes it harder to intentionally shape its direction based on inputs that make sense to humans.

Fundamentally, this is because machines do not “think” in the way that humans do.  The problem with the label “Artificial Intelligence” and batting around reference to the Star Trek computer is that we anthropomorphize these computer systems and want them to be like human minds.  We created them, so in a way they’re just mimicking us, right?

Wrong, of course.  Machines execute a series of operations until the program tells them to stop.  Machines don’t “think” the way we do, but that won’t stop them from doing things that astound us. What an AI needs is massive data to learn from, and massive computing power to crunch it all, and Google has both.

Andrew Ng, who has taught AI at Stanford, built AI at Google, and then moved to Baidu to continue developing AI. He recently said:

“When machines have so much muscle behind them that we no longer understand how they came up with a novel move or conclusion, we will see more and more what look like sparks of brilliance emanating from machines.”

Right now, Google’s AI isn’t really being given that chance.  It’s essentially being asked to work clean-up duty on all the completely new / novel and hard-to-understand queries that Google sees in a given day.  From Bloomberg:

“The system helps Mountain View, California-based Google deal with the 15 percent of queries a day it gets which its systems have never seen before.”

This isn’t glamorous work, but it doesn’t need to be.  The AI doesn’t care.  What it does is learn.  A “learning machine” creates a positive feedback loop (as it learns, it discovers way to accelerate learning). From Demis Hassabis, CEO of deep learning company DeepMind (acquired by Google in 2014 for $400MM)

“I also think the only path to developing really powerful AI would be to use this unstructured information. It’s also called unsupervised learning— you just give it data and it learns by itself what to do with it, what the structure is, what the insights are. We are only interested in that kind of AI.”

RankBrain is just the tip of the AI iceberg. As we’ve seen, Google has thought of itself as an AI company from the beginning, but, they’ve been cautious in their use of AI. And we know that their ambitions are toward a much more powerful and self-directed AI. The fact that RankBrain has advanced sufficiently to be included in Google’s ranking algorithm is a big step, but it’s just the beginning.

Google is in a uniquely powerful position in the AI field. They have massive data and massive computing power. Put the two together and very interesting things can happen. It may seem like a big leap to get from typing queries into a search box to speaking to an omniscient Star Trek computer that understands every word we say and the context around it. But we’re much closer than you think.

Mobile Apps vs the Mobile Web for B2B

TL; DR – Mobile apps are used for entertainment purposes.  If you’re a business, you should make the mobile version of your website your #1 priority.  Your mobile app can wait.

Mobile apps are getting a lot of press and buzz, and as a B2B marketer, you see lots of large corporations grumbling and making noises something to the effect of “we’ve got to get a mobile app for our business”, “gotta get an app built this year”, etc.  Why?

The Perceived Market Opportunity

It comes down to the perceived market opportunity.  Lots of people are using mobile apps, and spending lots of time in them:

number of mobile apps used and time spent per visitor

Image via Statista 

Those numbers seem really impressive.  App users use, on average, more than 20 apps per month and spend 21 – 37 hours in those apps on a monthly basis.

Other mobile app stats of interest:

  • 102B (billion) mobile apps have been downloaded by the end of 2013
  • That’s forecast to rise to 225B by 2016
  • 93% of those apps were free
  • 7% were paid

 mobile app downloads

Image via Statista

Again, those are impressive-looking numbers.  What’s missing is any discussion whatsoever of what the app using versus non-app using populations look like.

The Market Reality

Businesses should be aware of what’s happening when the media chattering classes are investing in a technology hype cycle:

garnter technology hype cycle

Image via Gartner

They are trying to get you excited, so they can get you Clicking so they can make MONEY.  The media will continue to be a fluffer for whatever topic gets clicks.  You will see them quoting big, top-line numbers like “103 billion apps downloaded in 2013.”

That seems like a lot, right? We’ll let’s take a closer look.

Consider data from comScore’s most recent Mobile App Report that shows that 2/3 of mobile users don’t download any apps on a monthly basis:

comscore app usage

Image via QZ

Digging in even further, you find that the top 7% of users account for 50% of total app downloads!

So it seems that apps are broadly not that interesting to 2/3 of the population, with a tiny fraction of users accounting for most of the app-related activity.

For the 1/3 of the population that does download apps, the usage patterns are pretty interesting.

Flurry analysis of 26,176 apps over 3 years shows that 50% of apps will lose 50% of their peak users within 3 months of launch.

Mobile App Uninstall Rates

Image via Flurry

To put an even finer point on it: 60% of downloaded apps are used 10 times or less:

app retention

Image via Localytics

And finally, you have analysis from Gartner that predicts that less than 1% of apps will be considered a financial success by the companies that develop them through 2018.

Anecdotally, app developers seem to be figuring this out as well:

So far, the data tell us that 2/3 of people aren’t very interested in apps, and most apps won’t be financially successful.  But what about those people who do use apps?

Mobile Usage Patterns

When it comes to how average mobile users are spending their time online, again mobile apps win hands-down:

  • 86% of total time spent using the Internet on a mobile device is done via apps
  • 14% is spent in a mobile browser

apps vs mobile browser

Image via Flurry

And it gets really interesting when you look at how that time is spent:

  • 37% of time spent in apps is for Gaming
  • 33% of time is spent in Social Media apps
  • 9% is spent in Entertainment

That’s means a whopping 79% of time spent in mobile apps is for entertainment.  If it is true, then, that in about 4 of 5 cases, apps are being used for entertainment, what does that indicate relative your high-level goal to “get an app built”?

The Business Case

The starting point for a mobile app should be a very well-defined set of user needs which call for a clean, mobile solution, where the mobile website experience offers too many options. Here, the business case for a mobile app can be pretty straightforward.

If you have a ton of loyal, repeat mobile customers who only need to accomplish 3 or 4 things, and these needs are recurring, you can package that into an app and get great adoption and usage.  Loyalty + limited use cases + recurring need = mobile app fit.  For B2B users, this can include access to reporting or analytics, or surfacing data to field sales teams.

But, if your visitors have more varied needs while they’re mobile, you should consider how your website renders on the mobile web.

Google recently did a survey of 1,500 consumers who use their mobile phones while shopping.  Google’s survey confirms that 65% of consumers use the mobile web to do their retail research – whether researching products; finding where products are sold; making price comparisons or finding promotional offers – rather than 35% who use mobile apps.

Google’s survey points to a strong user preference for the mobile web, but I always prefer studying analytics data of what users actually did, rather than what they say they will do.  Skava, a mobile commerce platform, recently released data that points to an even greater user preference for the mobile web after 2014’s Black Friday (emphasis mine):

“When it comes to holiday shopping, consumers have clearly voted for the convenience of the mobile Web over retailers’ applications.

In an analysis of 46 million page views from mobile devices on Skava’s clients’ mobile Web sites and apps, including Gap, Staples, Macy’s and Toys ‘R’ Us, the company found that the mobile Web accounted for 97 percent of retail sales from mobile devices on Black Friday while apps accounted for just 3 percent.”

That’s about as strong an indication of user preference as you’ll get.

Investing in Apps vs the Mobile Web

I’m a big believer in aligning marketing efforts in a way that maximizes business outcomes.  Building a mobile app for your business is not a “win” unless it results in a greater number of conversions [or insert your favorite metric here] than would have been the case had the same resources been devoted to another marketing channel.

So I tend to look at mobile apps from a utility perspective.

But there’s another way to look at mobile apps versus the mobile web.  There is deep, nerdy philosophical divide about open versus closed data architectures, and what that means for the future of the Internet. It goes something like this:

  • Open data architectures
    • Personified by:
      • The open-source software community (Linux, Apache, Hadoop, etc)
    • Characterized by:
      • Use of open-source software to create products and services findable and available on the open web
    • Controlled by:
      • Decentralized control, where any community member can change and modify code
  • Closed data architectures
    • Personified by:
      • Companies building closed ecosystems (Facebook, Apple)
    • Characterized by:
      • Software used to create products and services that are accessible only from within their own ecosystem
    • Controlled by:
      • Centralized control, where companies dictate changes to the platform

Tim O’Reilly had a debate with John Battelle a few years ago, with O’Reilly arguing on the side of open data architectures:

(I’m) surprised that you’re not more worried about the consequences of the shift from the “front end” of browser-centric computing to the “back end” of apps, closed networks and proprietary connections between massive data servers and specialized clients.

Why? Because the first defaults to open and the second defaults to closed.

But from a social, innovative, and macroeconomic perspective, open is almost always better. As Michael Wolff points out in his companion piece, it’s no surprise that media moguls (old and new) are pushing to regain control. What’s a surprise to many is that they’re getting it.”

One interesting area where this divide is playing out is in mobile apps versus the mobile web.  Apps are siloed, the web is open.

Summing Up

So, to review:

  • 60% users don’t download apps frequently
  • 80% of app usage is for entertainment purposes
  • 90%+ deliberately turn to the mobile version of your website rather than an app while making purchase decisions

With that perspective, just how sure are you that you need an app?

Consider the benefit to your business if you took those same resources, and improved any of the major metrics around your mobile website.

What would the business impact of investing that same amount of money in conversion rate optimization?

What would be the business impact of investing that same amount of money in improving your website’s performance?

The Aberdeen Group study showed that a one second delay in page load time equals:

  • 11% fewer page views
  • 16% decrease in customer satisfaction
  • 7% loss in conversions

Apps are for people who already know you and like you enough to download your app.  The mobile web is for everyone.  And that includes people who have no idea that your business exists (yet), but will use the web to find you, and become your customers.

Your Mobile Website Experience is Frustrating Your Users

Mobile users don’t like to be frustrated by user experiences and websites that aren’t optimized for mobile. But a “mobile-friendly” website is table stakes (Google is even incorporating mobile-friendliness as a ranking signal on April 21).  Mobile-friendly will only get you so far.

After creating a responsive version of their website, many organizations will sit back in satisfaction, considering their work done.  They shouldn’t, and neither should you.

You should prioritize web performance (aka page speed) ahead of any other factor of your mobile website experience. Statistically speaking, here’s why:

  • Slow pages are the number one issue mobile users complain about.
  • 75 percent of users will leave a site that takes longer than 5 seconds to load.
  • The average download time for a full web page is 11 seconds on an iPhone 5s over 4G LTE.
  • 85 percent of mobile users expect pages to load as fast or faster than they load on a desktop.
  • 38 percent of users have cursed at, screamed at, or thrown their phones when pages are slow to load.

mobile behavior

These are not pretty statistics. But investing six figures in a responsive web design – only to see your bounce rate skyrocket and your conversions nosedive – isn’t pretty, either.

Here is the impact of a one-second delay on your web engagement metrics:

  • 7 percent decrease in conversions
  • 11 percent decrease in page views
  • 8 percent increase in bounce rate
  • 16 percent decrease in customer satisfaction

But even with these very compelling data points to make the case for improving mobile website performance, most mobile websites continue to fall short.

Many organizations have a very skin-deep view of their mobile website experience. They labor under the delusion that users care most about how the site looks, and so long as the site is a responsive design, users will be happy.

Buy your mobile experience is about much more than just how your website looks on a mobile device. It’s also about how it functions, and a critically important aspect of your mobile web experience is speed.

Simply put, when your web pages download quickly, your users are happy.

happy baby

When your web pages are slow, users get frustrated, and start to think bad things about your brand.

angry baby

Let me give you an example.

Researchers performed a study where users were divided into two groups, and asked to complete a series of tasks on the mobile version of the Tesco website.  Users in both groups saw exactly the same web pages, and were asked to complete exactly the same tasks.

The only difference between the test cases was that one group interacted with a version of the site that had been slowed down 500 milliseconds. The test was blind, so users did not know they were seeing a “slow” version of the site. The differences that this half-second delay made were profound.

Exit interviews were performed with all users, and the words associated with the Tesco website shifted from mainly “easy-to-use,” to a range of negative associations. Users who had to endure the extra half-second delay used three times as many negative adjectives, including “slow,” “tacky,” “inelegant,” “clunky,” “boring” and “hard to navigate.”

          Normal conditions                                         500ms delay

mobile word cloud

This echoes research from Jakob Nielsen, who found

“Slowness (or speed) makes such an impact that it can become one of the brand values customers associate with a site.”

65 percent of users say their opinion of a brand was affected by their online experience. Do you want “slow,” “inelegant,” and “clunky” associated with your brand?

Increasingly, it’s a mobile-first world. But, you must have an eye on the entire user experience, rather than just how your site looks on a mobile device. Your users are discerning, and how quickly your pages load on a mobile device has a big impact on how your users view your brand, and how much they will be willing to interact and spend with you. In 2015, expect to see the brands that get mobile web performance right steal market share from organizations that neglect it.

Building a Business Case for Online Communities

B2B companies frequently look at online communities with mixed emotions.  On the one hand, they see the obvious success of sites like the Amex OPENForum, and they aspire to reap similar benefits.  On the other hand, there are the standard fears that many companies have:

  • Fear that community will devolve into a quagmire of complaints from upset customers
  • Fear that they will not be in control of the community discussion
  • Fear that they will invest substantial resources in a digital product whose ROI is difficult to calculate

These are all valid concerns, but they frequently recede into the background as more is learned about how communities develop, and how to think about the value they create.

Building a Business Case

The ROI question is truly a challenging one for most B2B organizations. The point of community, of course, is to provide value to the community members, not to push product. But even setting that expectation can set some B2B organizations on edge.  They end up asking some version of the question: “If I can’t measure a lift in products sold, how can I prove value?”

The over-arching message should be about building a strong sense of trust and relationship within the community, which increases customer loyalty and improves brand image, advocacy, and preference.

And there’s research to prove this (emphasis mine):

  1. “Research… reveals that customers who socially interact with other customers, via participation in brand communities, often exhibit an intense loyalty to the sponsoring brands.”
  2. “Positive influences of participation on recommendation behavior, brand image of the community sponsor, and intention to continue community membership can be confirmed.”
  3. “Researchers believe that in the future online communities will be the fundamental basis of companies’ marketing strategy. Especially, understanding consumers’ brand community behavior is considered a key success factor in doing business in the digital age. Virtual consumer communities (communities centered on a specific brand or consumption activity) can be of great value to marketers. Firstly, a company can use virtual brand communities for data mining and learn more about its customers, products, and even competitors. From active communities it is also easy to find potential candidates for pre-release product testing. Secondly, the most active members of these kinds of communities can do some of the work traditionally done by the company marketers by providing information seekers with purchase incentives and reliable information. Brand loyalty is closely linked to virtual interaction and to formation of strong social ties between individuals in a community.”
  4. “In the quest for building long-term successful brands, many marketers have become increasingly interested in how to create and foster successful communities of brand users. The appeal of such an approach to relationship marketing lies in the recognition that members of brand communities tend to exhibit favorable brand-related behaviors and intentions.
  5. “The results from the structural model suggested that perceived social support and consumer-brand relationship were important drivers of relationship mediators (i.e., sense of online brand community), which led to relational outcomes (i.e., brand commitment, brand preference, brand advocacy, and behavioral loyalty).”

The next challenge is to take these softer notions of loyalty, preference, and advocacy, and translate them into something measurable:

Web Engagement metrics:

  • Total visits by channel
    • Referring sites
    • Direct
    • Search
    • Social
  • Total pageviews
  • Time on site
  • Bounce rate
  • New visitors
  • Repeat visitors
  • Frequency & Recency

Community Satisfaction (qualitative):

  • Community responses to online surveys
    • Qualaroo, Foresee, FluidSurveys, etc
  • Bi-annual “Community Health Index” survey
    • Email survey to all active community participants asking for feedback

Community Engagement (quantitative):

  • Registrations
    • Ratio of new visitors to registrations (conversion rate)
    • Number of registrations over time
    • Registered users vs active users
  • Contributions
    • Total number of threads
    • Total unique contributors
    • Number of posts per active member
    • Posts per day
  • Topic interaction
    • Most popular subject areas
    • Thread depth (posts per thread)
  • Moderation effectiveness
    • Speed of replies to discussions
    • Number of volunteers

Community Development

Online communities are organic entities.  They have to grow and develop over time, and they often grow in surprising directions.  That said, community development is not a passive endeavor.  Communities require proactive management and participation in order to maximize positive business outcomes, and reward positive behaviors.

Despite the surprising directions that communities can grow in, the motivations for community participation are often quite uniform:

  1. The opportunity to participate in like-minded discussion
  2. Being connected to something larger than themselves
  3. To give and receive help
  4. To receive affirmations of the importance of their contributions
  5. The opportunity to elevate themselves as “experts” within the community

In order to position a new community to attract, retain, and grow its member base, the following considerations must be taken into account:

1. Seed the Community

  • Send invitations to participate to people who you think would be a good fit
  • These can be thought leaders, business partners, and loyal customers
  • Choose a core set of SMEs who will be accessible to community members
  • Exposure to, and the ability to network with, known experts is a core part of a community’s value proposition

2. Develop Your Moderators

  • Moderators must be open, candid, and “real people” not a “corporate mouthpiece”
  • Moderators must be responsive to activity within the community
  • Moderators must recognize and react to the needs of the community
  • Members of sufficient standing may be offered a position as Moderator over time

3. Expect 2-Way Communication

  • Think of members as “advisors” to your company
  • Be open to receiving feedback and advice from community members
  • Show community members how their feedback is influencing your company, your priorities, and your products

4. Encourage Profile Development & Content Creation

  • Allow members to accrue “points” based on volume and quality of interactions
    • Posts
    • Responses
    • Likes” of comments
    • Etc
  • Allow users to build out profiles that represent themselves
    • Personal interests
    • Professional interests and qualifications
    • Detailed descriptions of their businesses
  • Score users based on how “good” their profile is
    • E.g. LinkedIn “Profile Completeness” meter
  • Encourage content creation
    • Elevate guest blog content of outstanding quality to the main blog
    • Proactively promote the best content outside of the community
      • Primary social media accounts for the community should share this content via social media for added exposure

Setting Expectations

Finally, you have to set expectations around how quickly the community will grow.  Every B2B organization I’ve ever discussed community would prefer that the community be populated, vibrant, and self-sustaining as soon as it’s launched, but that’s obviously not a realistic expectation.  You must actively invest in your community, and those investments can take a significant amount of time to pay off.

For example, everyone in B2B knows that Amex OPENForum is a powerhouse with massive traffic, attracting very notable contributors.  What’s easy to forget is that OPENForum had very humble beginnings.  In 2007, it was originally started as a place to house video assets.  Early versions of the site were heavy on entertaining videos from Mario Batali and The Blue Man Group.

But the early site also had a live chat functionality, where employees of Amex could participate in discussions with actual community members, to discover what they needed.  In Lean Startup practice, this is called Customer Discovery.  Amex didn’t know what their customers’ problems were, so they asked them.  Here is an actual group chat transcript from Feb 2007:

Lee from Amex: Hello. Thank you everyone for coming to this on-demand experience of an OPEN event. A new OPEN Forum site is going to relaunch very soon with all new community features as well as propiertary business content. Please check back to on 6/18 and for the next Iconic Event taking place at the LA Film Festival on 6/26 with a new slate of brand builders.

stacy: Hi Alshatu Mamudu, I am in India right now. You can establish small business in India as you can save on the manpower and infrastructure cost. mail to : [email protected] if you wanna know more details.

page Bouchereau: What about Tesla?

toks: sadjo how are you? do you have any business ideas you can hint me?

sadjo: has anyone ever seen the blue man group before?

kararaina metekingi: hi there l am curious about finding out how to start any business not sure what l wanna do but just curious

AIshatu Mamudu: I really need a guide on how to begin a small bussiness of my own.


Natalia Mbamba: hey how do i join this it is very intresting.and i would like to get the guide i need in opening a business of my own

Ashura: Am intrested in this course and i would like to get educated about opening a small business of my own so please guide

Wow!  Amex had to pivot substantially from their “video platform” positioning in 2007 toward what became the OPENForum in mid-2008.  The process of identifying the true customer need involved LOTS of hands-on discussions with their end users.  And this process of finding their true purpose and building the platform to address customer needs took time:

Amex OPENForum growth

Image via ContentLab

After its launch in 2007, the OPENForum only accumulated 425,000 pageviews in all of 2008!  If we assume a modest 3 pageview per visit (significantly lower than the 6.4 PVPV they get now), that translates into fewer that 12,000 visits per month!  For a massive organization like American Express with massive budgets at their disposal, that’s a modest start.  It’s also a really potent reminder that no amount of resources can cheat you ahead in the process of establishing product-market fit.  It’s an exploratory process, and it often takes longer than you would hope it does, even if you’re a gigantic corporation.

Another example that I like to use is Wikipedia. True, it is not a “community” website in the same sense that a B2B forum is a community website.  Wikipedia’s potential user-base is everyone in the world, writing about almost anything in the world, in contrast to a B2B site whose target audience and focus is much more narrow.  But it’s still instructive to look at how long the ramp-up time was for Wiki:

Wikipedia growth

Image via Wikipedia

Some highlights:

  • Getting from zero 100,000 articles took 3 years
  • Getting from 100,000 to 200,000 articles took 1 additional year
  • Getting from zero to 500,000 articles took over 5 years
  • Getting from 500,000 to 1,000,000 took 1 additional year

This is a powerful reminder that patience is a virtue in community development. Done well, the rewards for community development are substantial.  But they don’t happen overnight.

Understand the long-term view of how community can help you develop, understand, and communicate more effectively with your customer base.  Know that the first turn of the flywheel is going to be slow, expensive, and require a huge amount of effort.  Go into the project with your eyes wide open.  Done well, the rewards for community development can be massive.

Technical SEO in a Semantic Search World

This is the deck I presented at a recent session “Advanced SEO” for the Digital Summit conference in Denver, called “Technical SEO in a Semantic Search World”.

This presentation was actually a really fun opportunity for me to talk about something that’s been on my mind – the changing role of SEO in the world of Semantic Search. I’ve been doing SEO professionally since 2003. In the past 3 years, I’ve seen more big, substantive changes to the field than I saw at any time in the 9 years that preceded that. We as SEOs, and as an industry, need to think about how our roles evolve and develop so that we can continue to add value to our organizations and clients, rather than selling them a Chinese menu of outdated optimization tactics.

On slide 2, I talked a bit about the state of the web at large. And the state of the web is that it’s a mess. The web is a heaving mass of unstructured data, that grows exponentially, is completely unsupervised, and constantly has new technologies stirred into the stew.

On slide 3 my point was that technical SEO represents our best toolset for elevating our sites out of the mess that is the web. But today, technical SEO is table stakes. In the game of organic search, technical SEO is simply the minimum amount you can ante to be in the game. 10 years ago, doing technical SEO meant you were guaranteed some solid first page rankings in return for your effort. Today, basic technical SEO will get you ranked somewhere in the top 100 for a few of your keywords. It’s that much more competitive.

And so, on slide 4, I talked about all of the things that we as SEOs have to keep our eyes on. SearchMetrics rankings correlations show all the factors that might be correlated with a single URL ranking. MOZ study data show the difference between page-level vs domain-level ranking factors, and whether they’re keyword dependent, or keyword independent. Add to that the Penguin and Panda algorithms, and keyword data disappearing into (not provided), and you get a sense for how busy the average workday of an SEO is.

I think that we as an industry are falling into the trap of paying attention to the tactical elements of SEO that we are familiar with, and not thinking about the forest from which all these trees are growing.

On slide 5 I discussed the new world of semantic search. Specifically, if you were to ask most any SEO today “What is semantic search?” you would be likely to hear something along the lines of:

  • Semantic search is about understanding the meaning of queries, and delivering results based on an understanding of the context in which the query was made
  • Semantic search is about “things not strings
  • Things in semantic search are “entities”
  • Entities are organized according to an ontology
  • Entities can have multiple properties based on their type
  • Entities can have relationships to other entities
  • Semantic search has its underpinnings in artificial intelligence (AI), deep learning, and natural language processing (NLP)

However, despite knowing all these things academically, they often haven’t filtered through to the extent that we have identified specific tactical strategies to address semantic search with technical SEO.  We may be fine talking about semantic search in the abstract, but we often default back to obsessing over character count in our title tags, because that is a more “knowable” field where we feel efficacious in our action.

All of that has to change.  So we know that semantic search changes things for our profession, but how?  Specifically, what does this change mean for us in our day-to-day jobs as we think about how to align our strategy and tactical delivery to the world of semantic search?

And honestly, I don’t think that semantic search requires a massive retrofit for us as SEOs.  Most of the stuff that you consider fundamental, technical SEO is still important, it just might be important for different reasons.  I.e., having descriptive, compelling title tags is still a best practice, because it allows you to put an effective marketing message in front of a human visitor right when they’re likely to be responsive to it.  In practical terms, a visitor may not see your title tag exactly as you wrote it. Google may be rewriting your title tags based on what it thinks is best for the visitor, or changing the visual formatting of the title.

So the fundamentals are still important, but they’re evolving.  You don’t have to throw out everything you know to adjust your SEO strategy to the semantic search model.  And there’s nothing more fundamental than the 3 pillars of on-page SEO:

  1. Information architecture
    • Depth of site architecture is one of the most important and overlooked aspects of site design
    • I provide steps so you can do your own analysis of your site’s IA
  2. Code
    • Organizations need to start thinking about how they’re going to integrate into developer workflows now
    • I provide examples for how you can insert correct syntax onto your pages to give your developers examples to code to
  3. Content
    • With semantic search, your content has never been more critical to your site’s ability to produce good ROI over the long term
    • Panda is just the most visible (and frightening) manifestation of this new understanding of the “quality” of your content
    • I provide steps for how perform an audit for thin content at scale.


Make Your Own Gl1tch Art

You know that really cool thing that the Internet does where, out of nowhere, you find something that you didn’t previously know existed, and you now become absolutely insane about?  Yeah, that thing.  It just happened to me, again, and it is making your own glitch art.

Glitch art is the process of taking an existing image and modifying the code used to render the image to make it… different.

Take one of my favorite recent art installations, Florjentijn Hofman’s giant rubber duck that floated into Hong Kong harbor.  It is rad-looking.


But what if you could modify it to make it even cooler?  You can!  German designer Georg Fischer (aka snorpey) is a glitch artist who has done a wonderful service to me personally by making a do-it-yourself glitch art creator. You just take an image, and then play with it until it is messed up in the exact way that you find aesthetically pleasing:

Ducky - glitch1

Ducky - glitch2

Ducky - glitch 3

Hooray, Internet!!!  Kiss an hour of your day goodbye!!  You’re welcome!!

Get Your Selfie On

SelfieCity is a data visualization tool analyzing 3200 selfies taken by people in 5 cities across the globe, then grouped by the city they were taken in (New York, Berlin, Moscow, Sao Paolo, Bankok).  The analyzed lots of different aspects, like age, facial expressions, glasses / no glasses, head tilt, etc:

SelfieCity dashboard
They started the project with 120,000 randomly selected photos and used Mechanical Turk workers to help classify them.  Stats of interest:

1. Only ~4% of photos taken are selfies
a. This is way lower than the ~20% that I would have guessed

2. Significantly more women take selfies
a. 61% of selfies in NYC, 82% of selfies in Moscow

3. Young people take selfies (duh)
a. Oldest average age of selfie takers was 27.6 in NYC

4. Women strike extreme poses in selfies (double-duh)
a. Average angle of head-tilt was 50% greater for women

The takeaway:  IF YOU ARE A GROWN MAN, STOP TAKING SELFIES.  Science has spoken.

As laughable, naval-gazing and ridiculous as I find selfies, their rise to prominence is fascinating to me.  The LA Times called 2013 “the year of the selfie“, after the Oxford English Dictionary chose “selfie” as word of the year. The New Yorker wrote 900 words on “The Return of the Selfie“, and there are even graduate theses being written about them.

The best thoughts I’ve read on the topic were from Stephen Marche in Esquire, where he wrote:

For all of human history, and even into human prehistory, making an image was an act that required not just conscious thought but immense effort. That’s why a picture was something removed from ordinary life — images possessed sacred properties, special auras and powers. The camera made the image much simpler to make of course, but it didn’t remove the consciousness of the act. Taking photographs required expensive machinery and skill… The leap in the ease of taking and disseminating images from the year 2000 to the present is as great as the leap from drawing in caves to the year 2000. And yet we still think of photographs as if they require effort, as if they were conscious works of creation. That’s no longer true. Photographs have become like talking. The rarity of imagery once made it a separate part of life. Now it’s just life. It is just part of the day.

I’m fascinated by whether selfies are trend that will peter out, or become so commodified that they’re no longer cool (“Nice selfie, MOM!”).  Or, maybe they’ll continue on their current trend, and 2 years from now, the only photos that can be taken by cameras are selfies.  If the frame doesn’t contain at least one person making duck-face, the camera won’t function.




What the Disaster Means for Your Website

Putting politics aside, the launch of the website has been a train wreck. I’ve been building and marketing websites for a dozen years, and I have never seen a launch go so spectacularly, publicly wrong. This is partly due to the highly visible and contentious nature of the Affordable Care Act. But more interestingly, this marks the first time in my memory that there was an implicit assumption that a big, complex program and a website were essentially the same thing.

If:    = Affordable Care Act

Then: is broken = Affordable Care Act is broken

This logic is fundamentally flawed, but that doesn’t stop it from being the dominant perception in the marketplace.  In fact, if you look at the relative popularity of these terms on social media, “” gets 2.6 times as many tweets per day as the name of the program itself, “affordable care act”, and Google Trends shows the terms neck-and-neck in popularity:

Topsy Analytics for and affordable care act

Google Trends data for and affordable care act

What this marks is a shift in the mindset of the American public – your website is not a part of your business, it is your business.  If your website doesn’t represent your company well, or even worse, doesn’t function smoothly, the bad smell doesn’t just hover over your website, but over your business as a whole.

But that knowledge comes with a huge upside.  As a website owner, you are in a uniquely powerful position – no one in the world knows your customers better than you do.  You will have some or all of the following ingredients at your disposal:

  • Customer knowledge
    • Direct customer experience and feedback
    • Your sales and support staff who talk with customers every day
  • Keyword data
    • Years of search keyword data specific to your audience finding your website
    • Social media keyword data
    • Keyword data from publicly available tools
  • Content performance
    • Engagement and conversion metrics of content you’ve published over the years
    • Analysis of competitors’ content marketing and social media campaigns
  • Conversion path knowledge
    • Knowledge of the specific steps involved in your customers’ journey
    • Knowledge of the common stumbling blocks your customers encounter
    • The ability to proactively present solutions to problems that customers may experience

These ingredients should be combined and recombined with user personas to continually refine the kinds of content and services that you offer to address changing audience needs.  If a piece of content does not directly satisfying one of the primary needs of your core audience, you should question whether it deserves to be prioritized.

Create content that solves user needs as its first priority, and then rigorously QA and analyze conversion bottlenecks to ensure that customers aren’t encountering barriers along the conversion path.

Granted, if your website fails at providing a good user experience, and the efficient delivery of its primary value to your customers, you’re not likely to have your failure broadcast on The Daily Show.

People will just quietly click the Back button, and find your competitors.

Daily Show tweet about issues

This is a re-post of my Forbes article.

What Google’s Changes Mean for Your Content Marketing Efforts

91% of B2B marketers use content marketing as a tactic, spending $118 billion in 2013 on content marketing, social media, and video.

If you don’t understand some key aspects of Google’s new search algorithm, you may be flushing your content marketing dollars down the drain.  True, talking about search algorithms tends to make eyes glaze over.  But, if you’re like the millions of other businesses that have identified content marketing as a key channel for educating prospective customers and getting them into the sales funnel, you need to know how the game has changed, and what that means for your business.

In the past 2 months, Google has made some significant changes.  First, they completely replaced their core web search algorithm.  Then, they hid the keywords that visitors use to find websites in organic search.

1.        Google Hummingbird

In August 2013, Google completed the change-over to their new search algorithm, Hummingbird.  A complete algorithm change is a BIG event for Google.

If the algorithm is the “recipe” that Google uses for ranking and retrieving results, then individual ranking factors within the recipe (such as using keywords in your title tags) are “ingredients” in the recipe, and Google has stated that there are over 200 ingredients.

All the early evidence indicates that Hummingbird still uses the vast majority of the same ingredients, in the same way.  So, doing good on-page and off-page SEO is just as important as always – more than ever, it’s table stakes for effective web marketing.

The big change with Hummingbird is in how it understands natural-language queries and processes them.  Hummingbird allows Google to understand the intent of queries in a much more intelligent way.  Now, Google may return results that may not contain the exact keywords you used in your query, in that exact order, but the results will match the intent of what you asked for.  For example, you may have Googled “the best French Cajun food in Baton Rouge” and get a website returned that only talks about Acadian cuisine, and doesn’t use the words “French” or “Cajun” anywhere on the page.  It matches your intent, but not the exact words.

Google is now much better at understanding “entities” or “things” and not just keywords or “strings”.

2.       Google Hides Your Organic Keyword Data

Next, in September, Google announced they’re encrypting all searches performed on Google.  That doesn’t mean anything to the average user of Google, but it means a foundational shift for companies that care about the organic search traffic that their site gets.  Anyone with access to web analytics for any site they work on will notice that a huge percentage of organic search traffic is now being lumped into the black box of “(not provided)”.  What this means is that most companies that get 80% or more of their organic search traffic from Google will no longer be able to see what terms visitors typed in to find their sites.

There are ways to work around this new limitation, and your resident SEO should be able to speak to them.  But those are tactical fixes, and what you should be looking for is a strategic solution.

1+2 = 4

So, let’s do a quick summary of what we know:

  • Hummingbird allows Google to understand the intent of queries much better
  • Google applies this refined query knowledge to entities (“things, not strings”) which allows them to consider a larger set of documents as potential results
  • Google no longer lets you see exactly what keywords visitors used to find your site

If Google is now very good at understanding the intent of a searcher’s query, but no longer lets me, as the site owner, see exactly what that query was, where does that leave us?

We are now in a place where the burden is on you, as an organization, to really understand your customers, identify each facet of the business problems that they face, and provide solutions to those problems, or at least being able to describe how your product fits into the landscape.

Absent query data is actually a big problem for lots of organizations.  Query data allowed us, at a tactical level, to see exactly how users were finding our sites, and then use standard web engagement and conversion metrics to attempt to rationalize how well our content matched that user’s needs.  If we didn’t have the right content, we could create new content, or re-write existing content to do give us context-appropriate coverage for that keyword.  It was a very tactical fix.

That approach resulted in SEOs feeding keyword “opportunities” back up the funnel to content writers to plug obvious gaps.  Often, the result was creating a shallow piece of content that contained the keywords, but didn’t address the end users’ needs in a substantive way.

What is required, now more than ever, is a top-down approach to understanding customer needs and developing comprehensive content sets that service those needs.  User personas need to be a driving force behind your content creation.  Are your editorial team and SEO team able to articulate all of the major points along the customer journey for each of your main user personas?  Do you have user personas that you update?  Do you regularly interview your sales staff to integrate feedback from the front lines into your personas?

More than ever, Hummingbird underscores the important of writing content that addresses real customer needs, not just specific keywords.  Conceptually, your content needs to be substantive and solve problems, not just fill “keyword opportunities”.

1 2 3