On May 6th, overnight, about 273,000,000 LinkedIn results disappeared from Google search results.
We noticed it first, or rather – our technology did.
The below screenshot from our IMPACT analytics platform shows that LinkedIn typically appears on page 1 for “IBM,” but on May 6th — it disappeared completely.
It was the same for individuals. Search results for “Mary Barra” (CEO of GM) showed that although she usually has LinkedIn as the first result after Wikipedia and her corporate bio page, on May 6th it was gone (replaced by a second link to GM’s site).
At the same time that LinkedIn disappeared from search results, the icon for LinkedIn also disappeared from the Knowledge Panel. For IBM it was replaced by a Pinterest icon. See before and after..
The LinkedIn disappearance was across the board.
For example, we track the percentage of Fortune 500 CEOs who have LinkedIn displayed on page 1 of their search results.
Usually it is about 55%. On May 6th, none of the CEOs had LinkedIn displayed.
We did not know what caused LinkedIn to disappear, but we did know that the main site was not being included in Google’s search index.
In other words, LinkedIn had been de-indexed by Google.
A search preceded by “site:” shows all the pages within a domain that are in Google’s index. This search showed that no pages from www.linkedin.com were included in search.
Here is what that search had previously looked like:
The pages still existed, and the site was still live. Someone who went directly to www.linkedin.com would never have noticed this search engine issue.
LinkedIn has country specific domains, which continued to rank in Google. For example, the French site still had all 23 million results appearing in search.
We also noticed that it was an issue specifically with Google.
On Bing, for example, LinkedIn continued to appear in search results.
Initially, we were unsure of the reason for LinkedIn to be dropped from results. It seemed unlikely that Google would choose to entirely remove such a major site. According to Similarweb, LinkedIn is the 26th most visited site in the USA, and over 23% of visitors to the site come via a search engine.
In the past, we have seen the Google algorithm demote sites for various reasons. However, we have never seen Google completely remove a major domain from search before.
It also seemed unlikely that LinkedIn would choose to remove itself from Google search.
Unless it was a mistake.
So as one does, we turned to Twitter.
First to notice this massive shift in search results, Five Blocks tweeted about it, looking for answers. SEO leader Barry Schwartz wrote about it on Search Engine Roundtable, and tagged two of Google’s spokespeople in a tweet.
A short while later, John Muller of Google tweeted a PSA which didn’t mention LinkedIn directly, but seems to have provided the clue to what had happened.
We understood this to be a cryptic message to LinkedIn, telling them that someone accidentally clicked a button in Search Console which removed the site from Google search.
Less than 24 hours after LinkedIn fell out of Google, it returned.
The graph below shows that very quickly, the percentage of Fortune 500 CEOs who had LinkedIn displayed on page 1 returned to about the same point it had been previous to May 6th.
All’s well that ends well.
The LinkedIn disappearance was a perfect opportunity for Five Blocks and our clients to be reminded of some fundamentals of online reputation management.
Here are some of our takeaways:
- It is essential to continue to monitor results every day. Someone, somewhere, accidentally clicking the wrong button can have an enormous, unexpected impact on search results.
- When LinkedIn does not appear in search it can potentially make results noticeably worse. Conversely, having LinkedIn in search can make results much better. That’s why we encourage clients to create and optimize LinkedIn and ensure that their profiles are indexed.
- We cannot rely too heavily on any one domain when working to improve search results. Even prominent and well-established domains can sometimes let us down. The more we use a variety of tactics and tools to enhance reputation the more secure the results will be.
- The Google Knowledge Panel is totally dynamic. Each time a search is made, Google checks the index to know what to display in the Knowledge Panel. As soon as LinkedIn was de-indexed, it disappeared from the Knowledge Panel. And as soon as the site was re-indexed, the icon reappeared. So, just because the Knowledge Panel looks the way it does now, does not mean it will stay that way tomorrow.
You have probably noticed that Google has been steadily elevating the prominence of the “question and answer format” for some time.
Google first introduced the People also ask box in 2015. For a while, it was a low-profile feature, often appearing midway or lower down the first page. As with virtually all new features, Google did extensive testing to ensure they are useful to searchers.
By early 2017, the feature began rising in prominence, including appearing for 10-15% of Fortune 500 companies. In November 2018, the People also ask box started appearing for about 80% of companies and grew from there, today appearing for over 90% of the F500.
It appears that Google has figured out not only the kinds of questions people often ask, but that people appreciate the format as an efficient way to find information quickly.
This new format often presents questions we would have been unlikely to ask ourselves (a search for ‘gaming fingers’ brings up both ‘how do you play the game number fingers’ and ‘can playing video games cause trigger finger’), and returns answers with varying degrees of authority, simply because they seem to match well.
That is…unless getting the right answer is a matter of life and death. Then Google interrupts its regularly scheduled algorithms to present results much differently, cutting straight to the most authoritative answer. Nothing brings this home more clearly than the way Google is handling the COVID-19 crisis.
What is going on? Let’s take a deeper look.
The Paths of Search
When you search online, Google does quick triage:
If there is one unequivocal answer – as in ‘Who is the Prime Minister of Canada’ – that answer will generally be delivered in the form of a Featured Snippet or Knowledge Panel. A question and answer box will appear below this, with related questions, answered by various sources.
If your query is something less clear, like “gaming fingers,” Google’s algorithm is not sure of your intent, so it will generally provide a variety of different results to help disambiguate what you are trying to find; you can pick a specific path or decide you know enough from what you’ve read on the search page.
The People also ask box captures frequent intentions – and Google figures that your intent is likely to be similar to one of them.
But what happens when there’s a lot riding on the answer?
When the answer is crucial, like in an impending weather emergency or a search related to suicide, Google overrides its own algorithm – and provides answers that have the greatest likelihood of keeping you safe.
A Tornado warning in the searcher’s area appears in red at the top of the page when they do a weather-related search. In the case of a suicide-related query, a toll-free hotline number appears in a large font at the top of the page, with the text ‘help is available.”
In these cases, even if Google is not sure what you are asking — and even if there are probably a variety of accurate answers to a question like ‘how to kill yourself’ — Google will err on the side of caution.
In these cases Google sacrifices variety, balance, considerations of the searcher’s own history, and any other elements of its algorithm for the most direct route to possibly saving a life.
The Coronavirus Connection
In our current moment, searches for coronavirus or COVID-19 bring up an expanded knowledge panel — almost a mini- site — featuring symptoms, prevention, treatments, and statistics. The area is branded as a red ‘COVID-19 alert’, and all information is provided only by highly reputable sources.
The People also ask box is farther down the page than normal, and is referred to as “Common questions” instead; answers are sourced from highly authoritative sources – most prominently, the CDC.
Along with this new way of presenting coronavirus answers, Google has a new initiative to help medical organizations optimize their sites, so that their information appears prominently in Google searches, and populates the results for these types of medical questions.
SEO Roundtable has summarized the initiative here, in which Google has provided guidance for health organizations on:
- Mobile optimization
- The importance of good page content and titles
- How to analyze the top coronavirus related user queries
- How to add structured data for FAQ content
While it would obviously be easier for Google to simply give all worthy health organizations more weight in the algorithm (using some sort of ‘authority score’), it seems to prefer that these entities assert their expertise organically.
Even more “interventionist”: As of end March, Google began heavily prioritizing local COVID-19-related results, giving more prominence to local publishers as each country began developing unique approaches to fighting the virus.
It seems that Google is scrupulous about accuracy only when there is a specific unequivocal answer, or when it thinks there might be a case of life and death. Maybe these are the only times Google won’t be called out for using influence in unfair ways?
Google’s opening gambit with health organizations, a half step between the democratic question and answer format and the tightly controlled emergency format, might be an interesting way forward in areas where more authoritative answers may not be immediately crucial, but would be appreciated by users – like nutrition or child-development.
In fact, all companies and professionals, following a best practices optimization strategy (titles, query analysis, FAQ content, etc.) can help establish their expertise on the results page. By assessing what questions people ask, and by answering them in the most relevant and easily understood way, companies may be able to make sure that the answers to the questions people ask are answered by their very own legitimate expertise.
It may not always be an emergency to get the answer right, but users would surely appreciate bumping into more true experts on their search paths.
What searchers think about a CEO’s search results
As an Intelligent Digital Reputation Management company, we are constantly doing research to understand how Google presents brands and executives in search. We also research how searchers utilize search engines to make decisions.
In August, we asked 672 US adults about their likelihood of doing business with Signiphy (a fictitious company) based on a provided set of search results – in visual form (see composite example below ) – for the (fictitious) CEO, Dave Miller.
Respondents were asked to imagine they were already considering doing business with the company and, as a next step, had decided to Google the CEO. Each participant was shown one of several variants of the search results page and was then asked to answer some questions about the page they saw.
Here’s what we found:
1 IMAGE BOX
This is intuitive – but it was borne out clearly in our research: Images on a search page help paint a picture in the searcher’s mind, more easily than any block of text, and affect how the searcher feels about the person they are searching.
We used images that depicted a CEO in a suit, smiling. He was evaluated as friendly and influential, and – notably – drove 80% more people to say they’d like him as a business partner compared to those who did not see the version with images. The majority of people, however, said they needed more information to decide.
Takeaway: Given how salient images are in searchers’ minds, well curated images can be a key factor in cultivating a favorable first impression.
2 VIDEO BOX
Video boxes take the impact of the visuals to another level. Like the image box, the video box in our study boosted impressions of the CEO as influential and friendly.
More than that, though, having a video box with favorable content helps solidify appeal of the subject for the searcher. Nearly 60% of people who saw the version with the video box indicated an increased likelihood to do business with the CEO’s company. Presence of the videos also boosted people’s sense that they had enough information to make a decision about doing business.
Takeaway: Videos that portray the CEO as a thought leader, hosted on known media (such as Forbes)– or even sites that appear to confer authority such as CNBC – are key in shaping searchers’ impressions of the CEO.
3 RATING SITES
Glassdoor is a site that allows current and former employees to rate their employer and its CEO. The site often appears on page one of Google search and is always shown with rating stars. Consequently, Glassdoor results are eye-catching and easier to quickly interpret compared to an all-text result.
The Glassdoor result in our research was shown with a mediocre 2.5 star rating. This led people who saw this result to be less likely to do business with the company, more so, in fact, than those who saw an all-text negative result about a pending lawsuit against the CEO.
People are used to seeing rating stars in searches for products and know to stay away when a product has few stars. That intuition transfers when rating stars are shown for a CEO, making Glassdoor an important result to watch.
Takeaway: If your brand has results with star ratings – Yelp, Glassdoor, or others – it is important to manage these and ensure that the star ratings are high – or alternatively that these results are not appearing prominently. Either way, brands need strategies aimed at managing these types of sites.
4 AUTHORITATIVE SITES
Bloomberg and Wikipedia are consistently among the most frequent sites to appear prominently in search results for CEOs. They typically appear toward the top of the page.
In our research, these two sites were the most frequently mentioned (in open ended replies) as the first places the searcher would click for more information.
Takeaway: Bloomberg and Wikipedia are deemed authoritative and trustworthy. This makes them an important component of a CEO’s digital presence.
5 KNOWLEDGE PANEL
As mentioned above, searchers’ attention will inevitably drift toward the visual, easily processed information on a search page. Even so, it is important to curate the text-only components of one’s search results. (This is especially true when there are fewer attention getting visual elements, where searchers are likely to take the time to read the text-based results.)
Our test pages mentioned both in the Wikipedia result and in the corporate bio that the CEO had been in that position for nine years. That information was duplicated in the knowledge panel blurb, which uses the first sentence from the subject’s Wikipedia page. People deemed that information useful and likely noticed it because it appeared in multiple places.
Takeaway: Curating search results by making sure key information appears in multiple authoritative online sources ensures that searchers will see desired information about the company or individual they are searching.
If someone has negative results appearing in search, does it make sense for them to try to “deceive” Google by using terms associated with the negative result in positive ways? A meditation on models and buses.
Amid the political drama of Brexit, a hung parliament, and upcoming elections, the virtual Boris Johnson also caused a stir, especially among digital media professionals.
A recent Wired article discussed the fascinating, odd phenomenon of Boris Johnson’s “off script” ramblings, rumored to be an attempt to “game Google.”
It began on June 20, 2019, when Johnson gave a most peculiar interview on Britain’s TalkRadio. He was not yet prime minister of the UK but was already the most likely candidate to replace Theresa May. He was asked what he does to relax. His reply, many will remember, was perplexing: “I make things. I make models of buses.” He went into detail about this unexpected hobby.
Almost immediately, pundits began to suspect there was something more cunning behind Johnson’s answer than the average politician’s disconnect from real people. Maybe, they surmised, it was a calculated move to improve his search results.
You see, Boris Johnson was already famous for another bus – one in which he rode around the country before the Brexit vote, encouraging people to vote to leave the European Union. On the side of that bus, in big letters, it said, “We send £350 million to Europe, let’s fund our NHS instead,” which was a contentious statement, and perhaps an outright lie.
First the social media opinion makers, and then everyone from Gizmodo to John Oliver, suggested that perhaps Johnson specifically spoke about model buses, so that anyone searching for “Boris Johnson bus” would read about the models, not the Brexit campaign.
This seemed unlikely at first. But then, in early September, shortly after media reports that the police had been called to a flat Johnson shared with his girlfriend Carrie Symonds, Boris gave a rambling speech while standing in front of a group of police cadets. Was that another attempt to manipulate Google?
And then on September 29, in an interview with the BBC’s Andrew Marr, Johnson insisted he was “a model of restraint.” The phrase was picked up in headlines across the country and internationally. Coincidentally (or not), he said this a day after police opened an investigation into potential criminal misconduct for awarding state money to model Jennifer Arcuri, an alleged mistress.
Oscar Wilde wrote, “To lose one parent may be regarded as a misfortune; to lose both looks like carelessness.” In the case of Johnson, repeating the supposed “trick” three times makes it seem like an attempt to massage his search results and media.
The question is – assuming it is a trick – does it even work? If someone has negative results appearing in search, does it make sense for him or her to try to “deceive” Google by using terms associated with the negative result in positive ways?
If you looked at the search results immediately after Johnson spoke about his model buses, most of the results were about the models, rather than the Brexit bus. So, it was a clear short-term win. But the Brexit bus was already old news, and Boris mumbling about his hobbies was probably just about as exciting for readers.
Furthermore, a couple of months later, the hobby buses have largely disappeared from search results (though “luckily” for Boris, the company that made the buses went bankrupt, so that has dominated the results.)
With the police and the model diversions, Johnson was less successful. Of the average 10 results on page 1 of Google search, one or two were about the positive spin, most were negative, and a couple were articles like this one, discussing whether or not Boris was intentionally trying to manipulate the results (it’s unclear whether these discussions please Johnson.)
But fundamentally, even if the theory is true and he really was using these tactics to manipulate search, it would be the wrong approach to take for a very simple reason: User behavior.
Someone who searches for “Boris Johnson bus” or “Boris Johnson model” is looking specifically for the “dirt.” Even if, hypothetically, there are no negative results on page 1, someone looking for information on Johnson and Jennifer Arcuri will simply click to page 2 or use more focused search key words until they find what they are looking for. With a specific search like that, there is no point trying to manipulate the results (even if such a thing were possible): you cannot fool the searchers themselves.
Perhaps the PR team behind this initiative got a pat on the back from Johnson – at least they managed to remove some of the negatives for a few days. But in fact, they didn’t really make any difference at all to Johnson’s online reputation.
We spend all day thinking about and working on client’s online reputation, and we know that cheap tricks and short-term fixes are not the way to go. Nor does it make sense to focus specifically on the negative terms, because someone who is looking for the negatives will find them.
Rather, we would be more interested in what search results look like for “Boris Johnson” or “UK Prime Minister.” If someone searches with no preconceived ideas, what do they find?
There is such a constant torrent of news about Johnson and Brexit, that it would be impossible to take complete control of his search results right now. Nonetheless, a search for “Boris Johnson” includes his Wikipedia article (which almost always ranks near the top of page 1 for the subject, and which is supposed to be unbiased and fairly written.) His Twitter account and his Facebook profile both rank on page 1. So, he “owns” two of the results on page 1 (and has a Twitter box, which occupies more space that he controls.)
Let’s look at a couple of senior members of Johnson’s cabinet. Jacob Rees-Mogg is Leader of the House of Commons. Page 1 of his Google results include his Twitter and Facebook profiles, his Wikipedia entry, and his profile on www.parliament.uk.
Chancellor of the Exchequer, Sajid Javid, has results from Wikipedia, Twitter, www.parliament.uk, www.gov.uk, and his own website, https://www.sajidjavid.com. These types of “owned” results occupy the top half of the first page of Google, knocking out negative results and allowing those MPs to more closely control their online reputations.
We don’t know for certain whether Johnson was trying to use tricks to manipulate his search results, but we do know that if he were to ask us, we’d tell him this: Sir, you’d be better off focusing your energy on optimizing www.boris-johnson.com (a site not updated since 2016) and/or populating borisjohnson.com, which is not even live. You could also strengthen your owned results for the search term “Boris Johnson,” rather than speaking about model buses or posing with policemen. Your PR team is welcome to contact us. You can go back to the models.
The following is a quick study of the fascinating dynamics behind search page results. Most people don’t think too much about the complex nature of how Google arrives at its results, but that’s exactly what we do here at Five Blocks.
A Wall Street Journal story last month accused Google of sourcing content without giving credit to the source. The article said Google had been scraping music lyrics from genius.com, a popular lyrics site.
Some quick background: When you search for the lyrics of a song, Google often shows the lyrics in what’s called a “rich snippet”.
What are Rich Snippets? you may ask.
Rich snippets are boxes that contain content clipped directly from a third-party web page. Google uses these rich snippets because it is the fastest way to get searchers the answer to their query.
Here is an example:
This is all fine as long as Google credits the source – and ideally provides a link to click.
However, in the case of song lyrics, Google seems to have provided the name of the artist, but not the website the lyrics came from – below is an image from before the WSJ story was published:
As you can imagine, song lyrics appear on lots of sites. So how could Genius.com be so sure that Google was scraping their site?
They ran a test. They embedded patterns of apostrophes in the formatting of song lyrics. If this pattern was replicated in the rich snippet in search results, it would prove that genius.com was the original source. They used two different types of apostrophes throughout the lyrics, as below (as seen in the WSJ.com article referenced above):
Genius checked the rich snippet, and found the apostrophe pattern replicated, proving the content came from their site.
But now the plot thickens — a development we became aware of after recently sharing this story with our partners.
The lyrics site LyricFind offered a rebuttal of the WSJ story, citing their licensing agreement with Google, and claiming that the content came from their site.
In fact, in the days following the publication of the WSJ story, Google began showing LyricFind as the source, and including a link.
But what about Genius.com’s test? Here’s what LyricFind says:
“Some time ago…Genius notified LyricFind that they believed they were seeing Genius lyrics in LyricFind’s database. As a courtesy to Genius, our content team was instructed not to consult Genius as a source. Recently, Genius raised the issue again and provided a few examples. All of those examples were also available on many other lyric sites and services, raising the possibility that our team unknowingly sourced Genius lyrics from another location. [Italics ours. – YG]
As a result, LyricFind offered to remove any lyrics Genius felt had originated from them, even though we did not source them from Genius’ site. Genius declined to respond to that offer. Despite that, our team is currently investigating the content in our database and removing any lyrics that seem to have originated from Genius.”
In any event, it seems that Google, having implemented the change of crediting the source, has admitted that lyric sites, and not only artists, have certain rights — even if those are different than those held by the artist or publisher.
This episode is a reminder of how Google, with its complex automated systems that collect, process, categorize, and display information, needs to be closely monitored and managed.
In the same way that the system seems to have sourced content without proper attribution, that same mechanism often presents incorrect and damaging content about brands and individuals.
With so many stakeholders using Google to help them form opinions on people, companies, and issues, watching every detail related to Google’s top results is critical for your brand.
Another reminder that Intelligent Digital Reputation Management isn’t just “nice to have” — but an essential business strategy.