Why SEO May Soon Become Less Relevant to Your Business

Why SEO May Soon Become Less Relevant to Your Business

Most businesses want to get noticed on the web, but they don’t know the best way to accomplish that. Will amazing content do the trick? What about great design or performance and page load times? Will the right keyword density on your pages make a difference? Or should you focus on inbound links?

Companies struggle with these questions every day, and they have been struggling for the past two decades. Getting noticed online has become an industry in and of itself, with consultants, service providers, brokers, and intermediaries trying to make sense of how search and discovery work. As most businesses have learned, it’s an incredibly complex practice: Google uses more than 200 signals to determine the relevance of a site to a specific search, for example.

If the typical search and discovery protocols seem dated and arcane, it’s because a lot of those protocols are dated and arcane. In this age of deep learning and artificial neural networks, for example, why do publishers still have to worry about whether adding multiple H1 tags in their webpages will adversely affect the discoverability of their sites?

The most important question, however, may be this: Why do changes to indexing algorithms of Google results make or break businesses?

Let’s look at Expedia, for example. At the end of 2019, the company had underwhelming quarterly earnings — and shares dropped by more than 20%. Expedia blamed these declines on changes to Google’s algorithm, which they said put their hotel listings lower in search rankings.

Although a majority of searches start with Google, the internet search giant is like a black box. There is little transparency into how Google ranks pages, prioritizes results, and — above all — how its results often supersede the results of other search providers. This has been the underlying theme of the European Union’s antitrust cases against Google as well as an impending antitrust investigation being pursued by the U.S. Department of Justice and attorneys general from various states.

Although the algorithms governing search results won’t become public knowledge anytime soon, there are some actions companies can take to prepare for emerging trends in search:

1. Natural Language Processing

NLP is a form of AI that helps machines understand human language and communication. With recent advancements in NLP, internet search protocols can begin to understand the content of webpages rather than just treating them as strings and keywords. Companies should assume that new NLP techniques will cause search engines to try to understand content similarly to how an end user would. Therefore, you should be tailoring your content for users rather than bots.

2. Structured Data 

Search engines are increasingly trying to figure out entities in web content (or “things, not strings,” as the saying goes). For example, if the word “Apple” on a webpage is tagged as an “organization,” the search engine would associate it with properties such as location, revenue, management team, etc. If the word “apple” is tagged as a “fruit,” however, it would be associated with attributes like color, nutrition information, etc.

This structure is often defined by standards made by the community initiative Schema.org and represented using standards like JSON-LD or Microdata. Businesses should tag their entities using Schema.org tags to allow search engines to identify them more easily.

3. Voice and Smart Devices

The methods for search optimization have to change because search itself changes constantly. In the future, not all searches will be based on keyword queries typed into browsers. We’re already seeing content search and discovery moving to social media, voice assistants, chatbots, smart speakers, smart cars, AR and VR environments, and so on. Comscore predicts that half of all online searches will be made via voice in 2020, and Gartner predicts that 30% of searches will come through devices that lack screens.

For content to be discovered under these new conditions, it’s less critical that companies produce every possible website link for specific keywords. Instead, these companies should find the answer that is most relevant to users in a given context and then allow those users to further refine their searches by providing more qualifying information.

To achieve this goal, deeper expertise is required in the specific domain of these queries. However, it’s impractical to assume that one company (or one search engine) can achieve that level of expertise specific to every domain area involved.

The solution? Businesses need to align more with content discovery networks, vertical search engines, and other domain authorities (think Alexa skills specific to your domain) to be discovered — rather than rely on the outdated keyword-based search.

This article originally appeared in business2community on April 30, 2020

Want to make better investment decisions? Look to AI

Want to make better investment decisions? Look to AI

While $9.3 billion was invested in AI companies in 2018, many VC investors aren’t leveraging that tech in their businesses. AI models are especially beneficial tool for VC investors because it can comb through massive data sets and sort information to fuel better investment decisions and higher ROI.

In its purest form, AI is a method of curve-fitting. Imagine a random plot of points on a piece of paper and then ask yourself, “Which straight line most closely fits those points?” Instead, we can envision a more complex version of that same diagram. Given a random plot of red and green dots, can we draw a curve that will separate the red dots from the green ones? Can we calculate a mathematical equation for such a curve? If so, then we can predict the color of a new (previously uncolored) dot using that equation.

That’s a lot of mathematical jargon, so let’s apply this theory to a real-life situation: VC investing. In this case, the red dots represent failed companies; the green ones are successful organizations. The goal is to come up with the equation of a curve that separates the two so that you can accurately categorize each new investment (or uncolored dot) as a potential success or failure.

 

Unlike the dots on a piece of paper, however, we aren’t confined to two dimensions. Instead, we must evaluate startup characteristics using hundreds (sometimes thousands) of variables. The problem becomes far more complex, but it’s certainly solvable with the help of modern AI technology.

AI Modeling Takes More Than a Good Face

The challenge is identifying the dimensions necessary to model the problem. Do they even exist?

In venture capital, it’s common to look at founders’ academic pedigrees, their past experiences and successes, demonstrations of their entrepreneurial traits, and their levels of passion. You should then look at market-specific conditions like competition, pricing, product-market fit, traction, and so on.

Some of these dimensions are easier to model than others. For example, it’s mathematically easier to model entrepreneurs’ alma maters and the academic rigors of those schools than it is to track how passionate they are about their products. Similarly, how do you model product-market fit when buying decisions can be based on such nuanced factors as design, aesthetics, speed, and features?

 

Understandably, many VC investors might be reluctant to embrace technology to help evaluate these dimensions to make a decision. It can be difficult to alter your long-trusted processes to fit modern conveniences. However, investors who rely on antiquated processes will not see as much success in their investments as those who embrace AI and other innovative technologies.

Technology Is Here to Help Rather Than Hinder

A lot of tenured VCs will tell you that venture investing is more of an art than a science. This may very well be true given that more than 627,000 new businesses open each year, and 70% of them usually fail within five years. Sadly, no scientific/mathematical solution can prevent failure.

Still, it’s critical for VCs to shortlist potential candidates from the ever-growing pool quickly — and the most efficient way to do this is with technology like AI. Once you’ve done that, you can take a more “artful” approach to decision-making.

Another cliché associated with high-stakes decision-making is to “listen to your gut.” This gut instinct was once thought to be what separated best-in-class decision makers from their peers, but recent discoveries in neuroscience suggest these gut feelings are just our brains giving us a push in one direction based on previous experiences. AI models, which are trained on historical data, could behave similarly to “gut instincts” — and that artificial instinct would get better as it consumes more data.

Finally, most everyone falls into the trap of feeling threatened by technology. After all, it’s common practice for businesses to replace blue-collar workers with automation or retail tellers with self-serve machines. The obvious logical next step would be to replace white-collar workers, right?

This fear is warranted in certain instances, as some jobs are threatened by intelligent machines (think robo-advisors for investments.) It’s not hard to imagine a company like Crowdfundr building AI-based VCs that investors can use to make smart investment decisions while bypassing the venture capital community. The VC community could go from backing disruption through AI to being hoisted by its own petard.

That said, I don’t foresee a near future where technology takes over the jobs of white-collar workers. Instead, I see innovative technologies — like AI — becoming critical tools in decision makers’ arsenals. Without these tools, I imagine it will be hard for VC companies to thrive.

The Future of VC With AI Models at the Helm

We’re far from the days of robo-partners running VC firms, but we’ve reached the point that easy-to-model dimensions can give us a glimpse of future possibilities. AI models can analyze parameters like funding history, investor activity, and sector-wise M&A data from sources such as Crunchbase, AngelList, and PitchBook. We can then use the analyzed data to make better investment predictions.

Similarly, LinkedIn data related to founder and employee background, headcount growth, key hires, and more can be used to generate models around technological superiority and expected growth. Additionally, web traffic data aggregators such as SimilarWeb and SEMrush can be used to model traction in the market, traffic estimates, monthly active users, growth rates, and so on. Finally, we can extract customer sentiment from user comments posted via social media, online discussions, and app store reviews.

This wealth of data can be used as dimensions to model startups. You can then use historical data to train AI models, and eventually, use those models to predict the potential outcome of new investment opportunities.

There are many different ways to use technology like AI to help you make better investment decisions. As long as you embrace technology rather than shy away from it, you’ll gain a competitive edge and see greater ROI.

This article originally appeared in ValueWalk on March 27, 2020

How Mastering Natural Language Processing Can Boost Your Bottom Line

How Mastering Natural Language Processing Can Boost Your Bottom Line

When past generations imagined the future, they pictured humans and machines speaking naturally to one another. While voice assistants like Siri and Alexa still fall short of that vision, they have come a long way from the days when talking to a machine was like pulling teeth. Here’s how mastering natural language processing (NLP) can boost your bottom line.

As we move further into the future, far more forays into the digital world will involve simple speech.

Not to be confused with neuro-linguistic programming (a kind of therapy), NLP is a field of artificial intelligence focused on making machines better at understanding human communication.

Most voice applications today perform simple speech-to-text transcription or text-to-speech synthesis.

However, NLP goes significantly deeper to understand, manipulate, and generate human-like language. In the past, interacting with a computer meant using a mouse and keyboard. That won’t change overnight, but many interactions will start to play out more like seamless human conversations.

To experience the true benefits of NLP, businesses will need to adopt these new technologies proactively.

Consider the annoying exchange at the start of most customer service calls, for example. Instead of having to navigate an endless maze of options by pressing buttons in response to prompts, NLP allows callers to talk to automated customer service agents.

To talk to automated customer service agents is innovative technology. Tech like this makes the experience faster, easier, and more organic, which is why 80% of customer service interactions are expected to be at least partly automated this year.

Conversational computers have countless applications, but the most significant benefit of NLP is less about voice applications and more about its ability to “read” and “understand” written text.

Instead of identifying keywords and taking educated guesses, computers can now understand the substance of text.

Understanding the substance of text includes valuable data sitting invisibly inside countless documents. Some estimates suggest companies that manage to use all the information hidden in their documents could see $430 billion in productivity gains. That could never happen without NLP, though.

Projections suggest the NLP market will grow to $26.4 billion globally by 2024, and that figure isn’t surprising considering what the tech can do. But the excitement should be tempered with urgency. That’s because in five years, not having NLP as part of your business will be like not having a website — which is unthinkable.

Tracking the Course of NLP

Development moves fast in this space. NLP models from just five years ago seem antiquated, and some of the biggest advances happened over the past 12 months.

During that time, the application of deep learning and large scale unsupervised learning techniques has advanced NLP models. Previously models were less than 100 million parameters, and now these have moved to 8.3 billion parameters (think of parameters as loosely synonymous with synaptic connections in the human brain).

The result is performance benchmarks capable of surpassing humans. To put it simply, NLP has become incredibly smart and is getting even more intelligent by the day.

The next challenge involves miniaturizing the technology to fit inside of smaller devices, such as smartphones. Once that happens, NLP will be able to analyze all the speech and text in the world.

Analyzing all the speech in the world includes documents, websites, articles, research papers, and anything else you can imagine.

As NLP consumes the entirety of human knowledge, it will lead to the creation of super-intelligent machines that understand every nuance of human conversation.

Human language includes humor, sarcasm, and context. The NLP machines will be able to draw upon vast amounts of knowledge to respond with incredible precision.

For example, developers at Airbus built a “robo-astronaut” to interact with crew members on an upcoming Space X flight.

Using advanced NLP, the robot can perform routine tasks, hold conversations, understand emotional cues, and display a distinct personality.

A robo-assistant is an assistant as well as a friendly companion — indeed something out of science fiction.

It will be a while before anyone invites robot companions into their offices and homes, but we’re well on our way to that reality. Regardless, NLP is ready to transform the relationship between humans and machines while creating a seamless link between the physical and digital worlds.

With sweeping changes coming, now is the perfect time to get ready.

Elevating Your Bottom Line With NLP

Embracing NLP in your business is a lot easier once you understand the tangible benefits instead of the abstract potential. With that in mind, here are a few ways NLP can elevate your bottom line over the next few years:

  • Optimize customer service. NLP optimizes customer service in two ways. First, it lets companies elevate their service levels by providing answers faster (through the web, chatbots, or voice applications like smart speaker apps), working in multiple languages, and handling higher-level questions. At the same time, NLP allows companies to spend less on human service agents, office spaces, phones, and other costs. With NLP, delivering exceptional service has never been easier or more affordable.
  • Improve regulatory compliance. NLP will transform administration in countless ways because machines can now complete work that previously required human eyes. For instance, this is tremendously helpful with regulatory compliance. Instead of asking compliance officers to pore over oceans of data looking for potential violations, computers can automate the initial review and escalate potential irregularities to compliance officers. Better yet, this approach takes less time and leads to fewer mistakes.
  • Learn from customer data. Customers leave data everywhere — think online reviews, Facebook posts, direct emails, and more. Thanks to NLP, companies can easily process this wealth of information and use it to evaluate customer sentiments. More broadly, NLP lets companies glean more insights from customer data regardless of its source, size, or format.
  • Augment text-intensive tasks. The first pass for a number of text-intensive tasks can be automated using NLP-based applications. This has already been successfully applied to spam detection in emails, but it has numerous applications within enterprises. For example, HR professionals and recruiters could reduce their workloads by intelligently sorting and categorizing heaps of résumés via automation.
  • Search beyond keywords. Most businesses rely on search technologies that are antiquated and based on keyword matching. This produces less-than-optimal results for customers, employees, and partners. Applying NLP to search produces results based on an understanding of the meaning of the query instead of just by matching the keywords.

Embracing NLP in 2020

NLP benefits all businesses, but that doesn’t mean you should rush to implement this technology.

Start by assessing your readiness and identifying what the technology is meant to do. For example, do you want it to improve service, cut costs, reduce churn, or something else?

Next, decide whether you have the resources to develop NLP-enabled applications in-house or need the help of a partner.

Finally, identify the source of the data you’ll use. Like all AI-driven innovations, NLP is only as good as the data it’s using to learn. Pretrained models are available, but you will still need to feed in specific data relevant to your company, customers, or industry.

If you’re still on the fence about NLP, consider it in this context: Language revolutionized how humans interact with each other, and it’s now doing the same thing for machines.

Nothing will be the same, yet almost everything will be better.

This article originally appeared in readwrite on February 12, 2020