Facebook is becoming a leader in AI and in advancements for self-taught AI. Usually requiring pre-programming from data sets, Facebook made it possible for AI to learn from images on the internet without curated data sets or annotation.
Weeks after developing the computer vision system that can accurately return image results with minimal supervision, Facebook is back in the lab working on their latest venture, Learning from Videos.
Before we get into that, here’s a peek at what’s to come:
- Facebook’s AI is getting ready to understand video content
- Google giveth and Google taketh away featured snippets
- SEO myth-busting
Facebook Is Watching Your Video Content
Earlier this month, on March 12, Facebook announced a new project — Learning from Videos.
This undertaking will allow Facebook to automatically learn audio, textual, and video representations from public videos; and it will benefit Facebook’s AI core, improving existing AI-based features and providing new experiences to users.
Since 2020, Facebook has been working on improving the video experience with Instagram Reels. Reels has been running with the Generalized Data Transformation (GDT) system to recommend content to the user that is similar but not a near-identical version of what they recently-watched.
In 2019, 81% of the 1,800 marketers surveyed by Buffer said they post video content on Facebook. This beat out 63% on YouTube and 58% on Instagram.
So, how will Learning from Video affect videos published on Facebook?
Geoffrey Zweig, director at Facebook AI, chatted with VentureBeat and had this to say about the future of AI and video.
“In the future, when applied in production, we believe they could do things like caption talks, speeches, and instructional videos; understand product mentions in videos; search and classification of archives of recordings. We are just starting to scratch the surface of self-supervised learning. There’s lots to do to build upon the models that we use, and we want to do so with speed and at scale for broad applicability.”
That’s right; Facebook will add captions to your content (which you should be doing manually ATM) and understand product mentions.
Understanding the video content and catching product mentions in the video will benefit you. Video content is going to translate better via search results either from your account or from a product mention in other account’s video content.
- Video search is rolling out with new capabilities and greater accuracy.
- Facebook’s AI can recommend similar video content while avoiding duplications.
- Video content has been a rising trend for the past several, and social media platforms are finally prioritizing it.
Featured Snippets Where’d You Go
Moz caught a dramatic drop in featured snippets on February 19. 2021.
On February 19, SERPs with snippets dropped 3% and maintained this level until March 12, when snippets returned, levelled out, and slightly increased on March 15.
In typical Google fashion, we have no solid answer on why this happened, but we know where these drops occurred.
The above graph is using February 18, February 19, and March 12 as the data points. As you can see, one-word keyword snippets nearly went extinct on the 19th, dropping to about 4% of such keywords featuring a snippet.
Since the recovery on March 12, the trends have levelled out, and snippet security feels like it is back, but who knows when they could disappear again.
What about the pages that had owned the snippet section before they dropped off?
Moz dove in to figure this one out, and they found 60% of keywords retained the featured snippet they held pre-drop and displayed the same featured URL on Mar 22nd.
Before we all start debunking that stat because SERP rankings are in constant flux, Moz ran another comparison for the 22 days pre-drop (Jan 27 vs Feb 18). This comparison found that 71% of keywords maintained featured snippets, and highlighted the same URL over the 22 days.
But this drop created an 11% discrepancy from pre to post-dip in featured snippets.
Is this a big deal? Not really; due to the constant flux of page one and algorithm changes, an 11% difference isn’t enough to cause a stir in the SEO world.
Here’s what we can take away from this: one-word keywords are at risk.
Seeing this decrease in snippets focused on one-word search terms could be the push we’ve needed for long-tail keywords and a friendly reminder to diversify targeted keyword lengths.
Maybe there was a glitch in the system, or maybe this drop was the result of an unrelated update. Whatever the cause, we know that snippets have bounced back, and the battle is back on for claiming that top spot 💪.
- Google may be trialling the removal of single word keyword featured snippets — or not.
- Diversification of targeted keyword lengths is an excellent choice to secure your place in the race for multiple snippets.
Don’t Believe Everything You SEO Know.
SEO tactics and strategies typically arise out of trial and error stemming from the little bit we know about how search engines work. This is why many tactics are simply myths.
Incorrectly attributing a rise in organic traffic to a change made on-site, or outdated advice as algorithms change all lead to false information. Sometimes the tactic works in particular scenarios not applicable to all industries or verticals, and this is how SEO myths often come to be.
Fact vs fiction—Foundation is about to expose some SEO myths.
Myth #1: PPC Advertising Helps Rankings
Supposedly, pay-per-click advertising on Google will benefit your organic SERP rank.
Google has different algorithms for ranking organic search results and placing PPC ads… so no, PPC ads don’t help rankings.
Myth #2: Longer Content Is Better
Does longer content really rank better than concise, well-crafted content?
This tactic has been shared often in SEO forums, but we haven’t seen anyone back it up with valid evidence. Google’s John Mueller recently tweeted about this ~myth~.
Having the same word-count as a top-ranking article isn’t going to make your pages rank first, just like having a bunch of USB chargers isn’t going to get you to the moon. But, I’m still tempted to buy some of those USB chargers…https://t.co/TIuJHwHufn
— 🍌 John 🍌 (@JohnMu) February 8, 2020
Myth #3: Backlinks Are A Rank Or Break Factor
The importance of backlinks is a debated topic. We know they effectively benefit domain ratings and site authority, but are relying on backlinks as a tactic the hill to die on?
Mr. Muller kept it short and sweet in a tweet about the importance of backlinks. They’re still important… just not that important.
Links are definitely not the most important SEO factor.
— 🍌 John 🍌 (@JohnMu) July 31, 2020
Myth #4: Keywords Need To Be In URLs
Google has been an advocate for promoting user experience and has long moved away from rewarding keyword-stuffed posts and meta descriptions. So how this myth came to be and why it is believed in the first place is a myth to me 😅. Google does not reward keywords in URLs when it comes to ranking; URLs are for users.
Myth #5: Website Migration Success Relies On Redirects Alone
When migrating a website, you better be redirecting old URLs — but that is only half of it. Not even half of it — it’s only a fraction of what goes into a proper website migration. There are many factors at play, from layout and UX, to domain and to new site content, that will affect SEO performance.
Search Engine Journal has gone in and dug into Google’s public statements on SEO, to bust twenty myths, so we can focus on what matters. Don’t be fooled by forums; trail a tactic through SEO A/B testing to see if it works for you before committing to a site wide optimization project.
- Until we know the ins and outs of how search engines work, we will never fully understand how to master SEO.
- The world of SEO is based on trial and error, and sometimes a popular tactic is just a well-known hoax.
- Don’t believe everything your favourite SEO thought-leaders say, do a bit of background checking.
OTHER NEWS OF THE WEEK:
🛍️ Leading eCommerce and content marketing platform, Yotpo, closed $230 M in their latest Series F funding round bringing their total valuation to $1.4 B.
🪙 In a Series C funding round led by Coatue, Ribbit, and Stripe, crypto infrastructure provider Fireblock raised $133 M. Fireblock is approaching a $1 B valuation and a rare achievement of unicorn status.
📊 Dataminr has raised $475 M in Series F funding, bringing the real-time insights data company’s valuation to $4.1 B.
BRAIN FOOD OF THE WEEK:
Facebook Reality Labs (FRL) is creating an AR device-based experience that will blend the natural world and augmented reality. The goal is to do so in a way that doesn’t make the user choose between interacting with a device and the world around them 🤓🤯.
“Rather than constantly diverting your attention back to a device, the interface should simply come in and out of focus when you need it, and it should be able to regulate its behavior based on your very, very lightweight feedback to the system about the utility of its suggestions to you so that the entire system improves over time.”
– FRL Research Science Manager Tanya Jonker.
The team at FRL had to start at ground zero to create a product that can achieve this. Initially, they were working solely on always-available AR glasses; however, they ran into limitations very quickly. They realized there needed to be more pieces to the puzzle, so they began working on a wristband to accompany the glasses, which led to improving user modelling and AI’s ability to provide personalized user intent.
There is no release date, but the product is in the prototyping phase now. The future of AR integrated life experience is near!
TWITTER THREAD OF THE WEEK:
WHAT WE’RE WIRED INTO THIS WEEK 🎧:
Originally sent out, by me Cali B, on Thursday, March 25, 2021. Stay up to date with all of our latest findings by subscribing to our newsletter today. Signing up also gives you early access to Ross’ Tuesday essay full of exclusive industry insights.