Google Search is moving heavily towards Artificial Intelligence and deep (machine) learning.
We can see this happening specifically for –
(a) “never before seen querries”… that are basically querries for which Google needs to understand search intent.
(b) to rank pages that have no backlinks to them
There is no other way that Google could rank these pages other than using their deep learning algorithms (that basically spy on your activity on these pages). Did you know that 99.9% pages on the web have zero backlinks to them? If thats the case, how is Google ranking these pages?
That coupled with the statistic that over 25% of daily queries are never seen before queries and its obvious why Google needed something like RankBrain to help with search querries.
However, that’s not everything. Becasue there is no activity (backlinks) on these pages the only way they can rank them in the SERPs if by spying on users using Chrome.
Although they wont admit it… they ARE tracking chrome user behaviour and engagement as you click on search results and visit sites (yes really).
And they need to…
Because without that they don’t have any baseline data to feed into the deep learning algorithm… without which the algorithm is meaningless.
The algorithm needs to see associative patterns emerge from data sets.
And, thats why they’re tracking user engagement and behaviour. They need to feed and grow the datasets so they understand correlations between high qualtiy content pages and user behaviour. The more such data they have, the better the deep leaqrning and prediction algorithms get.
What’s even more clever is that they created as big a data set as they could without letting anyone know they were creating one (Rank Brain has been running for over a year now). Had the news leaked, us SEOs would have skewed and messed around with their data, thereby warping the training of the machine AI.
So there. They now have their base data (which could get invalid were someone to figure out and create user bots that can fool the AI – a near impossible task because they’ve got a year of training already in place and it would be similar to creating a bot that can fool a 4 year old baby into thinking a bottle of pepsi is a bottle of milk… hmm.. not really a great analogy – but I hope you get the point. The AI is growing up fast. The more it sees, the more it learns. As long as real humans keep using Chrome.
So, what’s next?
Well, complete nirvana for Google Search would be when the algorithm can understand the meaning and value of each page on the web, and spit out the best match to a search query it understood fully… without ever having to rely on tracking user behaviour on old or newly created content.
But, luckily for SEOs… they’re not there yet.
They need enough data so they can completely toss out all tracking of user behaviour.
User engagement is just an interim step, until Google Search gets to Nirvana.
Funny, that I just wrote about a machine achieving Nirvana 🙂
Here’s a 60 minute video where I discuss Google ranking factors in 2016.
All thanks to Josh Bachynski for his excellent insights, face reading skills and from whom a lot of this video has been derived. A lot of my experience as an SEO consultant based out of Mumbai and as an SEO expert in India along with what I do at SmarterSEO is seeded from countless online forums and estabilished mentors from whom regularly pick up cues and tips. Ofcourse there’s a lot more to that!
And overview and slides of the video can be found below.
Summary of 2016 Ranking Factors
On Site Factors (60%)
- URL Silo, Pagename, Title, H1, Internal Links, Schema (eg. Product micro data not h atom)
Offsite Factors (30%)
- Backlinks, nofollows, topical mentions, offsite schema, social, brand searches
User Clicks – Rank Brain does this (10%)
- SERP Bounce, Internal navigation (Engagement) , Returns and Conversions.
- They track users through Chrome.
- They were quiet about it initially until they got base data.
- Forbes article – Rank brain got things right 80% of the time compared to engineers 70%
- Still a baby growing up. More data… makes it better!
- RankBrain is a new algorithm, and is Hummingbird plus Artificial Intelligence
- Hummingbird ranks entities by building Knowledge Graphs.
- It reads your site and schema to make an entity database out of it.
- It affected 90% of search queries when it was released in 2013.
- RankBrain tracks searches, clicks and search termination, in addition to doing what Hummingbird already does with sophistication of AI and tracking User Behavior for its Deep Learning
- Example of tracking user behavior: While doing a brand search, the result that gets clicked more, gets boosted.
- It works on … “Know, Do, Go” Queries that are Global or Local (Pigeon update).
- 25% of daily queries are never seen before queries (that is 500 million daily queries are new)
- RankBrain has been running silently for over a year now, gathering data (since early 2015)
Click Through Rate (an All New SEO factor)
- CTR is a Ranking Factor Now.
- You need to get users to click through more with proper Titles & Descriptions.
- Search termination and bounce backs are very important now.
- You can check your CTR in Google Search Console.
- See how Your listing looks in the SERPs and make it stand out for higher CTR.
- Use catchy words that draw the attention of the reader and get them to click!
How To Optimize for better CTR?
- Firstly, users should be searching for your brand.
- OR – Your brand + your service or product
- When they search for your brand, your CTR in the SERPS should be close to 100%
- Anything below 60% is a problem! *Google thinks – why won’t users click your site listing?*
- Then, make sure you satisfy the users search query, when they visit your site (conversion).
- They should definitely not bounce back to Google!
- To improve brand recall you can use a short and catchy name.
- You can do some advertising, so Google sees visits and interaction on your site.
- You need to provide better value that your competitors. Period.
- Make users return and share content and even bookmark your site in their browser.
- *Trick Google* Have a contest! Offer 10% off if a visitor goes back to Google and does a brand search, clicks on your result and comes to your site and can then use the coupon and check out.
Keyword Research (new insertion)
- Use the Google Keyword Planner tool.
- Make sure to initially start research with broad phrases
- Then drill down to long tail keywords and exact match keywords
- And use keywords that are relevant and make sense to use for your content.
- You can filter Keyword Planner results by area and also look at mobile queries
- Make sure to limit your content to ONE primary topic per page.
- Use your primary/secondary keywords in the Title, H1,H2,H3 without keyword stuffing.
- Do not use Exact Match Domains (EMDs) as they are dangerous now.
- Remember – A good (catchy) Title and Description will give you a higher CTR.
- Make sure to update your content regularly, and include published dates.
- Identify your most important pages, and tell Google they are the most important pages by making sure that they…
- Get the most internal links
- Must have links to them from homepage
- Remember that large sites have more Authority, but only if content is good, and…
- Their pages convert…
- The user intent happens…
- Be it – Sign ups, clicks, purchase etc.
Local SEO 2016
- Get a Google Photographer from your local area listed in Google Business Pages to come to your place and do a photo shoot that you then upload.
- *Hint* You can ask him to refer you to other local SEO clients and in return you give him the business of your clients.
- You must have only one highly specific category in your Google Maps Listing.
- Again – the quality of your site is important, so is task completion and search termination.
- You must get good Reviews from BBB, Local Forums, Blogs, G+, Yelp, FB.
- Your site must be Mobile optimized.
- Your Business Name must be clickable! You do this by ensuring that your Business Name includes your business name and your service name.
- Must have REAL social activity on your main social profiles… FB, Twitter, G+
- Your Social Profiles and Pages must be real, organic and active. NO bots.
- Don’t spam your social profiles with selling offers.
- Always post shareworthy content that has value.
- If you choose to do a promotional announcement, do it once every 20 posts or so.
- If you can create potentially viral content, that is a huge plus.
- Engage with your audience. Talk to them.
- Shares of your pages are important ranking signal… you can get high buzz from experts in your niche. (Tip: use the Buzzsumo tool for outreach)
- Remember – clicks from Social pages ARE IMPORTANT and IS a ranking factor (no matter how much Google search team may choose to deny it).
You Want to be POPULAR
- Topical References
- Social Shares
- Brand Searches
- Social Profile Proof
- Ratings and Reviews
To do this you need to…
- Get real aged robust social accounts in your region doing real activity on google search.
- Get visitors to click on your result and visit your site and engage with your content.
- You can incentivize users to do this (sort of blackhat)
- But never buy bot traffic!
- Never use proxy IPs.
- All your Social Profiles must all be active.
- Users must click and terminate search results.
- Links are a bit risky now… but are very much STILL a ranking factor.
- You Should Get Links from….
- High authority sites (real businesses, news sites, charity sites)
- Links from sites where everyone else is NOT there
- Main contextual link
- Anchor must be natural
- Appropriate volume of links
- Do NOT match the SEO spam patterns
- You Should NOT get Links from….
- New blogs
- Manual post sites
- Multiple topic sites / generic sites
- PR, Directories, Articles
- Sidebar or footer links (Dofollow)
- Comments, Social bookmarks, forum profiles
- Spammy links are defined by the SEO community (the algo is trained by them)
- NO more than 10% EMQ, PMQ (even if in URL) anchors
- Do not do them anymore.
- Affiliate links, advertorials and questionable links.
- 12,000 manual reviewers working
- The algorithm bubbles up suspect and spam type links for team.
- Avoid these Links…
Sidebar, footer, directory links, blogroll, sitemap, compilation sources
- These links at best are devalued, and at worst trigger a manual action
- Is a Negative Ranking Factor (a demoter).
- It checks your site regularly… for low quality pages.
- It’s now based on primarily on Usage Metrics & Onsite Factors
- Panda now automatically updates in 2016
- They collected the average baseline clicks and conversions in each industry.
- Remember that Google has multiple quality Algorithms and factors.
- You can see accompanying resources PDF for the full list of factors.
- Users satisfied and converting over 50%
- Professional and clean website look
- A Legitimate business or entity
- Outbound links to authority sites in your niche
- A site that is shared, liked, reviewed, rated etc.
PANDA Does Not Like…
- A boring blog
- A site that has keyword stuffing
- Bad reputation site
- A site that annoys users with ads
- Too many affiliate links
- URL parameters in your navigation links… ?=xyz
- Bad design and bad html
- Is another Negative Ranking Factor and a demoter.
- Works on links mostly (and some on-page factors)
- Works on on-page and off-page over-optimization issues.
- More than 10% of Exact Match Query Links -> Exact Match Query Page on domain, are a problem and will get detected by Penguin.
Example: “Buy red apples” anchors, link to http://yoursite.com/buyredapples.html
- Do not have over-optimized … title, alt, title tags, URL, content, anchors.
- Negative link juice can flow from a internally hit page to your other pages.
- Disavow File – It does not work as intended by Google.
- If it did blackhats would get out easily!
- Can’t say about the new penguin they will release though?
- So, if you get a penalty you really need to delete your toxic links.
- A manual penalty may let you out more easily.
How to Fix A Penguin Hit Site
- Remove on-page keyword stuffing and repetition
- Delete your 80%+ links and then disavow request (if manual penalty)
- Start a new site (but Google may automatically 301 your old site to new site)
- 404 or delete target pages
Manual Actions or Penalties
- Come from Manual Web Spam team
- 6 or more types of manual actions
- Listed in Search Console in “Search Traffic”
- Your site must have Search Console installed on it to see them
- Most important notices to pay attention to are…
Action vs Site – you need to delete all the links and then beg to get out with disavow + reconsideration request.
(Google want you to put spammy links in their disavow file for crowdsourcing their machine learning.)
Action vs Links – Do NOTHING. It’s just a warning to change. They are only taking the links out from your link graph.
Is there Anything like a Duplicate Content Penalty?
- No such thing.
- It just forces google to choose which page to rank.
- And it always chooses legacy + authority
- You don’t want Google to have to choose if your site content is unique or not
- Google trusts sites with duplicate content LESS
How to Fix Duplicate Content Issues..
- Use tools like – pubsubhubbub, xindexer, and tell search console to spider your content first
- Install view-source blockers… the ones that block right click copy paste
- Use Copyscape tool to find copiers (use their batch search)
- File DMCA requests with Google
- Maintain an Accurate Sitemap. This is very important.
- Finally rewrite your duplicate content, and add video and images and did you know factoids etc.
Currently SEO is getting to depend more and more on user behavior tracking.
Google is using links and social signals to find trusted pages to rank and then solidifies the rankings by tracking user behavior and engagement metrics etc.
User tracking is just an interim stage until Google perfects their AI and Deep Learning.
Once they reach the perfect AI stage, Google will then understand and identify the best pages for rankings, automatically without needing any external ranking factors like backlinks and or tracking user behavior.