Google Serves Up a Fail for Waffle House

I’m not from the South and have never been to a Waffle House. So, without ever having given it too much thought, I could only imagine that eating there is probably fairly similar to any other fast food experience. Today, though, I had occasion to search for Waffle House on Google. Based on their search results, with all due respect to customers of Waffle House, I have learned that Waffle House seems to be a magnet for the lowest class of society. Highlights from their search page include:

Waffle House Google Search Results Marked Up

As an SEO researcher, I am particularly intrigued by the in-depth articles Google’s algorithm selected for inclusion here. In-depth articles often appear in search results for a known entity, such as a person, company, brand, or concept. They appear as a group of three articles and are typically long form journalistic coverage of the searched term. Often, their coverage has a negative slant. For example, one of Apple’s in-depth articles is “iPhone Killer: The Secret History of the Apple Watch.” For Uber, we find “The Inside Story of Uber’s Radical Rebranding.”

But these in-depth articles aren’t about Waffle House at all. They are stories about people, some crazy people, whose lives happened to intersect with Waffle House. Is that really what people searching for Waffle House are interested in finding? More than anything, they create a highly unfavorable impression of Waffle House, even though these stories have little to do with the brand itself.

Yes, these articles all mention Waffle House and yes, they are all long form journalistic coverage. In that sense, Google’s algorithm got it right. But I think nature abhors a vacuum and Google abhors one even more. In the absence of any other in-depth article-worthy coverage of Waffle House itself – positive or negative – Google’s algorithm scraped the bottom of the barrel and came up with these.

Ultimately this is one big fail for Google’s algorithm that leads to an even bigger fail for the Waffle House brand. Google is probably the best place to go when you want to size up an individual, company or brand. You are quickly exposed to a variety of sources and types of information: corporate website, social media, news, YouTube, Wikipedia etc.  But when Google gets it wrong – as they do in this Waffle House example – the cost can be very high for the brand, with stakeholders getting a negative and undeserved impression of the brand.

Lessons from Medical Informatics

Medical Informatics and SEO

Introduction

For the past 2½ years, I’ve worked as the CTO at Five Blocks while also studying epidemiology and biostatistics at Ben Gurion University, and applying that knowledge in the Clalit Health Services Research Institute in Tel Aviv. Being involved in multiple fields at the same time has given me a great opportunity to integrate different areas of knowledge and apply lessons and experiences across disciplines. While reputation management and medical research might seem completely unrelated at first, I have found many interconnections in the work that I do; I believe the cross-pollination has been fruitful.

Different Work Cultures

Medical research and internet startups have extremely different cultures. Medical research is very conservative – most researchers use software that is very old. Few engage in explicit methodological innovation; the methods are mostly taken as a given and applied to new research problems.

In the world of online business, by contrast, things change fast. Nobody wants to be using last years device, operating system, software, or SEO tactics. There is a recognition that if you are working the same way you did last year, you are falling behind more innovative competitors. This leads to agile business processes that are constantly being refined and made more lean and efficient.

But while medical research lags in efficiency, it has much to teach the reputation management and marketing communities about scientific rigor. The culture of peer reviewed research and formal modeling of variables is far ahead of what passes for “analysis” in the online business world. Most of the “evidence” used for internal business analysis would be considered hearsay by the readership of a medical journal.

Common Challenges

One area that I have found to be common between both fields is the importance of proper data management. When I learned about computer science in college, the main focus was on programming, either procedural or object oriented. Since then, the scale of data available to even the smallest business has skyrocketed. The problem of how to maintain order in all that deluge of data has become very significant. Whether looking at the problem of a brand’s reputation across the internet, or at a medical research question that impacts a large population, the data collection process is much more substantial than one might think. Furthermore, the management of that data, once it has been collected, becomes a task unto itself. It is impossible in real-world conditions to keep data rigidly structured, and thus coping with heterogeneous data becomes another major challenge. As we use different technologies at Five Blocks from those used at Clalit, I’ve had the opportunity to see how different toolkits are used to address these issues.

Another important commonality is the need for visualization. It might be obvious that busy executives, especially those who are not strong on math, would prefer charts over raw numerical data. But I think even career statisticians and math nerds also share the same need. In our information-rich environment, the ability to summarize reams of data points into an intuitive chart is extremely valuable. Ultimately it reduces the cognitive load required to perceive the meaning implied by the data.

Choosing the right mode of data analysis for the job is another important job that applies both in medical research and in reputation management. Medical researchers speak about randomized controlled trials. Web marketers are familiar with A/B testing. But these are essentially the same study design. Business people know about automated data mining for sales optimization, while epidemiologists are more comfortable with manually built regression models. Knowing how to use the full range of analytical tools available gives us many more choices and ensures that we can select the right one for the task at hand.

Ultimately, I think that the science and art of collecting, storing, analyzing and presenting data is really a field unto itself. For this reason, we can learn a great deal from how it is practiced in different fields of application. Often, within a particular industry there is an echo chamber effect, where the same methods and tools are used by all practitioners. By exploring how other industries address similar challenges, we can enrich our perspective and diversify our toolbox.