This post was originally published on the STAT blog.

Featured snippets, a vehicle for voice search and the answers to our most pressing questions, have doubled on the SERPs — but not in the way we usually mean. This time, instead of appearing on two times the number of SERPS, two snippets are appearing on the same SERP. Hoo!

In all our years of obsessively stalking snippets, this is one of the first documented cases of them doing something a little different. And we are here for it.

While it’s still early days for the double-snippet SERP, we’re giving you everything we’ve got so far. And the bottom line is this: double the snippets mean double the opportunity.

Google’s case for double-snippet SERPs

The first time we heard mention of more than one snippet per SERP was at the end of January in Google’s “reintroduction” to featured snippets.

Not yet launched, details on the feature were a little sparse. We learned that they’re “to help people better locate information” and “may also eventually help in cases where you can get contradictory information when asking about the same thing but in different ways.”

Thankfully, we only had to wait a month before Google released them into the wild and gave us a little more insight into their purpose.

Calling them “multifaceted” featured snippets (a definition we’re not entirely sure we’re down with), Google explained that they’re currently serving “‘multi-intent’ queries, which are queries that have several potential intentions or purposes associated,” and will eventually expand to queries that need more than one piece of information to answer.

With that knowledge in our back pocket, let’s get to the good stuff.

The double snippet rollout is starting off small

Since the US-en market is Google’s favorite testing ground for new features and the largest locale being tracked in STAT, it made sense to focus our research there. We chose to analyze mobile SERPs over desktop because of Google’s (finally released) mobile-first indexing, and also because that’s where Google told us they were starting.

After waiting for enough two-snippet SERPs to show up so we could get our (proper) analysis on, we pulled our data at the end March. Out of the mobile keywords currently tracking in the US-en market in STAT, 122,501 had a featured snippet present, and of those, 1.06 percent had more than one to its name.

With only 1,299 double-snippet SERPs to analyze, we admit that our sample size is smaller than our big data nerd selves would like. That said, it is indicative of how petite this release currently is.

Two snippets appear for noun-heavy queries

Our first order of business was to see what kind of keywords two snippets were appearing for. If we can zero in on what Google might deem “multi-intent,” then we can optimize accordingly.

By weighting our double-snippet keywords by tf-idf, we found that nouns such as “insurance,” “computer,” “job,” and “surgery” were the primary triggers — like in [general liability insurance policy] and [spinal stenosis surgery].

It’s important to note that we don’t see this mirrored in single-snippet SERPs. When we refreshed our snippet research in November 2017, we saw that snippets appeared most often for “how,” followed closely by “does,” “to,” “what,” and “is.” These are all words that typically compose full sentence questions.

Essentially, without those interrogative words, Google is left to guess what the actual question is. Take our [general liability insurance policy]keyword as an example — does the searcher want to know what a general liability insurance policy is or how to get one?

Because of how vague the query is, it’s likely the searcher wants to know everything they can about the topic. And so, instead of having to pick, Google’s finally caught onto the wisdom of the Old El Paso taco girl — why not have both?

Better leapfrogging and double duty domains

Next, we wanted to know where you’d need to rank in order to win one (or both) of the snippets on this new SERP. This is what we typically call “source position.”

On a single-snippet SERP and ignoring any SERP features, Google pulls from the first organic rank 31 percent of the time. On double-snippet SERPs, the top snippet pulls from the first organic rank 24.84 percent of the time, and the bottom pulls from organic ranks 5–10 more often than solo snippets.

What this means is that you can leapfrog more competitors in a double-snippet situation than when just one is in play.

And when we dug into who’s answering all these questions, we discovered that 5.70 percent of our double-snippet SERPs had the same domain in both snippets. This begs the obvious question: is your content ready to do double duty?

Snippet headers provide clarity and keyword ideas

In what feels like the first new addition to the feature in a long time, there’s now a header on top of each snippet, which states the question it’s set out to answer. With reports of headers on solo snippets (and “People also search for” boxes attached to the bottom — will this madness never end?!), this may be a sneak peek at the new norm.

Instead of relying on guesses alone, we can turn to these headers for what a searcher is likely looking for — we’ll trust in Google’s excellent consumer research. Using our [general liability insurance policy] example once more, Google points us to “what is general liabilities insurance” and “what does a business insurance policy cover” as good interpretations.

Because these headers effectively turn ambiguous statements into clear questions, we weren’t surprised to see words like “how” and “what” appear in more than 80 percent of them. This trend falls in line with keywords that typically produce snippets, which we touched on earlier.

So, not only does a second snippet mean double the goodness that you usually get with just one, it also means more insight into intent and another keyword to track and optimize for.

Both snippets prefer paragraph formatting

Next, it was time to give formatting a look-see to determine whether the snippets appearing in twos behave any differently than their solo counterparts. To do that, we gathered every snippet on our double-snippet SERPs and compared them against our November 2017 data, back when pairs weren’t a thing.

While Google’s order of preference is the same for both — paragraphs, lists, and then tables — paragraph formatting was the clear favorite on our two-snippet SERPs.

It follows, then, that the most common pairing of snippets was paragraph-paragraph — this appeared on 85.68 percent of our SERPs. The least common, at 0.31 percent, was the table-table coupling.

We can give two reasons for this behavior. One, if a query can have multiple interpretations, it makes sense that a paragraph answer would provide the necessary space to explain each of them, and two, Google really doesn’t like tables.

We saw double-snippet testing in action

When looking at the total number of snippets we had on hand, we realised that the only way everything added up was if a few SERPs had more than two snippets. And lo! Eleven of our keywords returned anywhere from six to 12 snippets.

For a hot minute we were concerned that Google was planning a full-SERP snippet takeover, but when we searched those keywords a few days later, we discovered that we’d caught testing in action.

Here’s what we saw play out for the keyword [severe lower back pain]:

After testing six variations, Google decided to stick with the first two snippets. Whether this is a matter of top-of-the-SERP results getting the most engagement no matter what, or the phrasing of these questions resonating with searchers the most, is hard for us to tell.

The multiple snippets appearing for [full-time employment] left us scratching our head a bit:

Our best hypothesis is that searchers in Florida, NYS, Minnesota, and Oregon have more questions about full-time employment than other places. But, since we’d performed a nation-wide search, Google seems to have thought better of including location-specific snippets.

Share your double-snippet SERP experiences

It goes without saying — but here we are saying it anyway — that we’ll be keeping an eye on the scope of this release and will report back on any new revelations.

In the meantime, we’re keen to know what you’re seeing. Have you had any double-snippet SERPs yet? Were they in a market outside the US? What keywords were surfacing them? 

Source

Ultimate SEO”

In my last post, I explained how using network visualization tools can help you massively improve your content marketing PR/Outreach strategy —understanding which news outlets have the largest syndication networks empowers your outreach team to prioritize high-syndication publications over lower syndication publications. The result? The content you are pitching enjoys significantly more widespread link pickups.

Today, I’m going to take you a little deeper — we’ll be looking at a few techniques for forming an even better understanding of the publisher syndication networks in your particular niche. I’ve broken this technique into two parts:

• Technique One — Leveraging Buzzsumo influencer data and twitter scraping to find the most influential journalists writing about any topic
• Technique Two — Leveraging the Gdelt Dataset to reveal deep story syndication networks between publishers using in-context links.

Why do this at all?

If you are interested in generating high-value links at scale, these techniques provide an undeniable competitive advantage — they help you to deeply understand how writers and news publications connect and syndicate to each other.

In our opinion at Fractl, data-driven content stories that have strong news hooks, finding writers and publications who would find the content compelling, and pitching them effectively is the single highest ROI SEO activity possible. Done correctly, it is entirely possible to generate dozens, sometimes even hundreds or thousands, of high-authority links with one or a handful of content campaigns.

Let’s dive in.

Using Buzzsumo to understand journalist influencer networks on any topic

First, you want to figure out who your topc influencers are your a topic. A very handy feature of Buzzsumo is its “influencers” tool. You can locate it on the influences tab, then follow these steps:

• Select only “Journalists.” This will limit the result to only the Twitter accounts of those known to be reporters and journalists of major publications. Bloggers and lower authority publishers will be excluded.
• Search using a topical keyword. If it is straightforward, one or two searches should be fine. If it is more complex, create a few related queries, and collate the twitter accounts that appear in all of them. Alternatively, use the Boolean “and/or” in your search to narrow your result. It is critical to be sure your search results are returning journalists that as closely match your target criteria as possible.
• Ideally, you want at least 100 results. More is generally better, so long as you are sure the results represent your target criteria well.
• Once you are happy with your search result, click export to grab a CSV.

The next step is to grab all of the people each of these known journalist influencers follows — the goal is to understand which of these 100 or so influencers impacts the other 100 the most. Additionally, we want to find people outside of this group that many of these 100 follow in common.

To do so, we leveraged Twint, a handy Twitter scraper available on Github to pull all of the people each of these journalist influencers follow. Using our scraped data, we built an edge list, which allowed us to visualize the result in  Gephi.

Here is an interactive version for you to explore, and here is a screenshot of what it looks like:

This graph shows us which nodes (influencers) have the most In-Degree links. In other words: it tells us who, of our media influencers, is most followed. 

These are the top 10 nodes:

• Maia Szalavitz (@maiasz) Neuroscience Journalist, VICE and TIME
• Radley Balko (@radleybalko) Opinion journalist, Washington Post
• Johann Hari (@johannhari101) New York Times best-selling author
• David Kroll (@davidkroll) Freelance healthcare writer, Forbes Heath
• Max Daly (@Narcomania) Global Drugs Editor, VICE
• Dana Milbank (@milbank)Columnist, Washington Post
• Sam Quinones (@samquinones7), Author
• Felice Freyer (@felicejfreyer), Boston Globe Reporter, Mental health and Addiction
• Jeanne Whalen (@jeannewhalen) Business Reporter, Washington Post
• Eric Bolling (@ericbolling) New York Times best-selling author

Who is the most influential?

Using the “Betweenness Centrality” score given by Gephi, we get a rough understanding of which nodes (influencers) in the network act as hubs of information transfer. Those with the highest “Betweenness Centrality” can be thought of as the “connectors” of the network. These are the top 10 influencers:

• Maia Szalavitz (@maiasz) Neuroscience Journalist, VICE and TIME

• David Kroll (@davidkroll) Freelance healthcare writer, Forbes Heath

• Jeanne Whalen (@jeannewhalen) Business Reporter, Washington Post

• Travis Lupick (@tlupick), Journalist, Author

• Johann Hari (@johannhari101) New York Times best-selling author

• Radley Balko (@radleybalko) Opinion journalist, Washington Post

• Sam Quinones (@samquinones7), Author

• Eric Bolling (@ericbolling) New York Times best-selling author

• Dana Milbank (@milbank)Columnist, Washington Post

• Mike Riggs (@mikeriggs) Writer & Editor, Reason Mag 

@maiasz, @davidkroll, and @johannhari101 are standouts. There’s considerable overlap between the winners in “In-Degree” and “Betweenness Centrality” but they are still quite different. 

What else can we learn?

The middle of the visualization holds many of the largest sized nodes. The nodes in this view are sized by “In-Degree.” The large, centrally located nodes are disproportionately followed by other members of the graph and enjoy popularity across the board (from many of the other influential nodes). These are journalists commonly followed by everyone else. Sifting through these centrally located nodes will surface many journalists who behave as influencers of the group initially pulled from BuzzSumo.

So, if you had a campaign about a niche topic, you could consider pitching to an influencer surfaced from this data —according to our the visualization, an article shared in their network would have the most reach and potential ROI

Using Gdelt to find the most influential websites on a topic with in-context link analysis

The first example was a great way to find the best journalists in a niche to pitch to, but top journalists are often the most pitched to overall. Often times, it can be easier to get a pickup from less known writers at major publications. For this reason, understanding which major publishers are most influential, and enjoy the widest syndication on a specific theme, topic, or beat, can be majorly helpful.

By using Gdelt’s massive and fully comprehensive database of digital news stories, along with Google BigQuery and Gephi, it is possible to dig even deeper to yield important strategic information that will help you prioritize your content pitching.

We pulled all of the articles in Gdelt’s database that are known to be about a specific theme within a given timeframe. In this case (as with the previous example) we looked at “behaviour health.” For each article we found in Gdelt’s database that matches our criteria, we also grabbed links found only within the context of the article.

Here is how it is done:

• Connect to Gdelt on Google BigQuery — you can find a tutorial here.
• Pull data from Gdelt. You can use this command: SELECT DocumentIdentifier,V2Themes,Extras,SourceCommonName,DATE FROM [gdelt-bq:gdeltv2.gkg] where (V2Themes like ‘%Your Theme%’).
• Select any theme you find, here — just replace the part between the percentages.

• To extract the links found in each article and build an edge file. This can be done with a relatively simple python script to pull out all of the <PAGE_LINKS> from the results of the query, clean the links to only show their root domain (not the full URL) and put them into an edge file format.

Note: The edge file is made up of Source–>Target pairs. The Source is the article and the Target are the links found within the article. The edge list will look like this:

• Article 1, First link found in the article.
• Article 1, Second link found in the article.
• Article 2, First link found in the article.
• Article 2, Second link found in the article.
• Article 2, Third link found in the article.

From here, the edge file can be used to build a network visualization where the nodes publishers and the edges between them represent the in-context links found from our Gdelt data pull around whatever topic we desired.

This final visualization is a network representation of the publishers who have written stories about addiction, and where those stories link to.

What can we learn from this graph?

This tells us which nodes (Publisher websites) have the most In-Degree links. In other words: who is the most linked. We can see that the most linked-to for this topic are:

• tmz.com
• people.com
• cdc.gov
• cnn.com
• go.com
• nih.gov
• ap.org
• latimes.com
• jamanetwork.com
• nytimes.com

Which publisher is most influential? 

Using the “Betweenness Centrality” score given by Gephi, we get a rough understanding of which nodes (publishers) in the network act as hubs of information transfer. The nodes with the highest “Betweenness Centrality” can be thought of as the “connectors” of the network. Getting pickups from these high-betweenness centrality nodes gives a much greater likelihood of syndication for that specific topic/theme. 

• Dailymail.co.uk
• Nytimes.com
• People.com
• CNN.com
• Latimes.com
• washingtonpost.com
• usatoday.com
• cvslocal.com
• huffingtonpost.com
• sfgate.com

What else can we learn?

Similar to the first example, the higher the betweenness centrality numbers, number of In-degree links, and the more centrally located in the graph, the more “important” that node can generally be said to be. Using this as a guide, the most important pitching targets can be easily identified. 

Understanding some of the edge clusters gives additional insights into other potential opportunities. Including a few clusters specific to different regional or state local news, and a few foreign language publication clusters.

Wrapping up

I’ve outlined two different techniques we use at Fractl to understand the influence networks around specific topical areas, both in terms of publications and the writers at those publications. The visualization techniques described are not obvious guides, but instead, are tools for combing through large amounts of data and finding hidden information. Use these techniques to unearth new opportunities and prioritize as you get ready to find the best places to pitch the content you’ve worked so hard to create.

Do you have any similar ideas or tactics to ensure you’re pitching the best writers and publishers with your content? Comment below!

Source

Ultimate SEO”

With the new year in full swing and an already busy first quarter, our 2019 predictions for SEO in the new year are hopping onto the scene a little late — but fashionably so, we hope. From an explosion of SERP features to increased monetization to the key drivers of search this year, our SEO experts have consulted their crystal balls (read: access to mountains of data and in-depth analyses) and made their predictions. Read on for an exhaustive list of fourteen things to watch out for in search from our very own Dr. Pete, Britney Muller, Rob Bucci, Russ Jones, and Miriam Ellis!

1. Answers will drive search

People Also Ask boxes exploded in 2018, and featured snippets have expanded into both multifaceted and multi-snippet versions. Google wants to answer questions, it wants to answer them across as many devices as possible, and it will reward sites with succinct, well-structured answers. Focus on answers that naturally leave visitors wanting more and establish your brand and credibility. [Dr. Peter J. Meyers]

Further reading:

2. Voice search will continue to be utterly useless for optimization

Optimizing for voice search will still be no more than optimizing for featured snippets, and conversions from voice will remain a dark box. [Russ Jones]

Further reading:

3. Mobile is table stakes

This is barely a prediction. If your 2019 plan is to finally figure out mobile, you’re already too late. Almost all Google features are designed with mobile-first in mind, and the mobile-first index has expanded rapidly in the past few months. Get your mobile house (not to be confused with your mobile home) in order as soon as you can. [Dr. Peter J. Meyers]

Further reading:

4. Further SERP feature intrusions in organic search

Expect Google to find more and more ways to replace organic with solutions that keep users on Google’s property. This includes interactive SERP features that replace, slowly but surely, many website offerings in the same way that live scores, weather, and flights have. [Russ Jones]

Further reading:

5. Video will dominate niches

Featured Videos, Video Carousels, and Suggested Clips (where Google targets specific content in a video) are taking over the how-to spaces. As Google tests search appliances with screens, including Home Hub, expect video to dominate instructional and DIY niches. [Dr. Peter J. Meyers]

Further reading:

6. SERPs will become more interactive

We’ve seen the start of interactive SERPs with People Also Ask Boxes. Depending on which question you expand, two to three new questions will generate below that directly pertain to your expanded question. This real-time engagement keeps people on the SERP longer and helps Google better understand what a user is seeking. [Britney Muller]

Further reading:

7. Local SEO: Google will continue getting up in your business — literally

Google will continue asking more and more intimate questions about your business to your customers. Does this business have gender-neutral bathrooms? Is this business accessible? What is the atmosphere like? How clean is it? What kind of lighting do they have? And so on. If Google can acquire accurate, real-world information about your business (your percentage of repeat customers via geocaching, price via transaction history, etc.) they can rely less heavily on website signals and provide more accurate results to searchers. [Britney Muller]

Further reading:

8. Business proximity-to-searcher will remain a top local ranking factor

In Moz’s recent State of Local SEO report, the majority of respondents agreed that Google’s focus on the proximity of a searcher to local businesses frequently emphasizes distance over quality in the local SERPs. I predict that we’ll continue to see this heavily weighting the results in 2019. On the one hand, hyper-localized results can be positive, as they allow a diversity of businesses to shine for a given search. On the other hand, with the exception of urgent situations, most people would prefer to see best options rather than just closest ones. [Miriam Ellis]

Further reading:

9. Local SEO: Google is going to increase monetization

Look to see more of the local and maps space monetized uniquely by Google both through Adwords and potentially new lead-gen models. This space will become more and more competitive. [Russ Jones]

Further reading:

10. Monetization tests for voice

Google and Amazon have been moving towards voice-supported displays in hopes of better monetizing voice. It will be interesting to see their efforts to get displays in homes and how they integrate the display advertising. Bold prediction: Amazon will provide sleep-mode display ads similar to how Kindle currently displays them today. [Britney Muller]

11. Marketers will place a greater focus on the SERPs

I expect we’ll see a greater focus on the analysis of SERPs as Google does more to give people answers without them having to leave the search results. We’re seeing more and more vertical search engines like Google Jobs, Google Flights, Google Hotels, Google Shopping. We’re also seeing more in-depth content make it onto the SERP than ever in the form of featured snippets, People Also Ask boxes, and more. With these new developments, marketers are increasingly going to want to report on their general brand visibility within the SERPs, not just their website ranking. It’s going to be more important than ever for people to be measuring all the elements within a SERP, not just their own ranking. [Rob Bucci]

Further reading:

12. Targeting topics will be more productive than targeting queries

2019 is going to be another year in which we see the emphasis on individual search queries start to decline, as people focus more on clusters of queries around topics. People Also Ask queries have made the importance of topics much more obvious to the SEO industry. With PAAs, Google is clearly illustrating that they think about searcher experience in terms of a searcher’s satisfaction across an entire topic, not just a specific search query. With this in mind, we can expect SEOs to more and more want to see their search queries clustered into topics so they can measure their visibility and the competitive landscape across these clusters. [Rob Bucci]

Further reading:

13. Linked unstructured citations will receive increasing focus

I recently conducted a small study in which there was a 75% correlation between organic and local pack rank. Linked unstructured citations (the mention of partial or complete business information + a link on any type of relevant website) are a means of improving organic rankings which underpin local rankings. They can also serve as a non-Google dependent means of driving traffic and leads. Anything you’re not having to pay Google for will become increasingly precious. Structured citations on key local business listing platforms will remain table stakes, but competitive local businesses will need to focus on unstructured data to move the needle. [Miriam Ellis]

Further reading:

14. Reviews will remain a competitive difference-maker

A Google rep recently stated that about one-third of local searches are made with the intent of reading reviews. This is huge. Local businesses that acquire and maintain a good and interactive reputation on the web will have a critical advantage over brands that ignore reviews as fundamental to customer service. Competitive local businesses will earn, monitor, respond to, and analyze the sentiment of their review corpus. [Miriam Ellis]

Further reading:

We’ve heard from Mozzers, and now we want to hear from you. What have you seen so far in 2019 that’s got your SEO Spidey senses tingling? What trends are you capitalizing on and planning for? Let us know in the comments below (and brag to friends and colleagues when your prediction comes true in the next 6–10 months). 😉

Source

Ultimate SEO”