There are many sites that curate content and links on the Web, including many blogs and a number of social sites that do it through submissions by their members, who can also vote upon those submissions. The inventors of PostRank came up with an algorithmic approach to rank articles and blog posts and other content on the Web, and present it to people based upon those rankings. I’ve found a patent application at the USPTO that provides some insights and details on how their approach worked.
Google acquired PostRank last June, as was announced on the PostRank blog on June 3, 2011. Given Google’s increasing move towards looking at more social signals for the potential ranking of content shared by others, it’s worth wondering how this social engagement data technology might be used by Google, and what the PostRank team might be bringing to the effort. PostRank Co-founder Ilya Grigorik, who now appears to be a web performance analyst with Google, noted in the post announcing the acquisition:
We know that making sense of social engagement data is important for online businesses, which is why we have worked hard to monitor where and when content generates meaningful interactions across the web. Indeed, conversations online are an important signal for advertisers, publishers, developers, and consumers but today’s tools only skim the surface of what we think is possible.
Google added social metrics reports to Google Analytics this past March, and that might be the area that PostRank team members were working upon after joining the search engine. ReadWriteWeb discussed that possibility while telling us about what it was exactly that PostRank did:
Here’s what PostRank does: you plug in any RSS feed to the system and it scores each post in that feed by the relative number of comments, inbound links, mentions on Twitter, saves on Delicious and other social media metrics. Then you can subscribe to a filtered feed of just the 10% most-discussed items in any feed. It’s magic, it’s gold and it’s all too often unappreciated.
Given the amount of information (and misinformation) that appears on the Web every day, there’s a great amount of value in an approach like that, applying filters when you don’t have time to do that filtering on your own.
It would be a little disappointing if the only thing to come out of the PostRank acquisition would be some analytics reports in Google Analytics. Is it possible that members of the team might be engaged in other areas of Google as well?
Are they engaged in working out how social sharing might be used to influence search results with the technology acquired from Grouptivity?
Might they be working on a way to track the impact of rumored Google Comments for third-party sites?
It’s possible that they might be engaged in making Google search more social in a number of ways.
Social filtering for news and for information that specific searchers find interesting could be a significant part of the more social web envisioned in Google’s Search Plus Your World social and personalized search results, as well as the user rank and credential scoring that appears to have been originally developed for Google’s Confucius Q&A sites.
Social search results have the most benefit when they respond to recency sensitive queries, like information about a natural disaster or some newsworthy event, or provide an opinion or review from someone we know on products or services or events, or give access to expertise that someone might have in a particular area.
The PostRank patent tells us that it fulfills a number of needs:
- Efficiently allowing users to effectively navigate through significant amounts of web content
- Enabling intelligent filtering and customization of web content reflecting a users’ unique tastes and interests
- Providing a system that is relatively unobtrusive, passive, and undemanding of the user
The patent application is:
Method, System and Computer Program for Managing Delivery of Online Content
Invented by Ilya Grigorik and Kevin Thomason
US Patent Application 20090164408
Published June 25, 2009
Filed: December 21, 2007
Abstract
A method for delivering online content is provided including the steps of
(a) providing access to online content including a plurality of data objects;
(b) obtaining information regarding the relevance of and/or likelihood of interest in the data objects by searching for online social engagement with the data objects by one or more users, so as to define social engagement data; and
(c) ranking and/or filtering the data objects for relevance of and/or likelihood of interest based on the social engagement data. A system and computer program for online content delivery is also provided.
Social engagement metrics are the responses to online content, and how the public refers to that content. It assumes that if the content is likely to be of interest to a lot of users that it will more likely be of interest to a particular user. If it’s ignored by many, it might also be ignored by an individual who might be more engaged with other content.
A social score might be generated from both active feedback and passive feedback. Active feedback can include recommendations from people about content and things like a thumbs up or thumbs down or numeric rating. Passive feedback includes things like how much time might be spent viewing something, if it’s bookmarked, and so on.
As noted in the snippet from ReadWriteWeb above, PostRank is intended to work with a system that might allow filtering of content from RSS feeds, which is filtered by social engagement metrics. While Google could possibly incorporate a system like this into Google Reader, a PostRank extension for Google Reader was retired in January.
The patent filing provides a very detailed look at how PostRank worked, including an example algorithm that shows how data from trackbacks, Technorati, Bloglines, comments, and bookmarks at delicious and Digg could be used together. It’s detailed enough so that if you wanted an example road map to help you build upon and improve something like PostRank, it might be worth spending some serious time with.
Extensions of PostRank
Where things get interesting in the patent application is the discussion of how PostRank can be used and can be extended.
For Example, it would help analyze and understand social activity on Large E-Commerce Retailer sites, such as Amazon.com. While there are often many thousands of items on such a site, explicit social data such as “product reviews, buyer statistics, product recommendations, vendor ratings, and overall sales rankings” can help visitors decide upon a book to buy, as well as lists generated by users of the site of recommended products to buy. Implicit information might also be useful in making that kind of decision, such as wishlists from users, and content external to Amazon itself, such as recommendations of products sold on Amazon upon blogs.
The social engagement metrics from PostRank could be applied to those blogs, and to wishlists to get a sense of the desirability of different items. Sentiment detection methods could be used on reviews and commentary on those items and reviews as well.
Those different types of data could be included in a calculation of the overall desirability of items at an ecommerce site. Imagine the addition of a dropdown box on an Amazon search that includes “desirability” as a ranking signal when displaying results.
One of the benefits described to people using mobile devices is that a “sync to mobile” feature of an RSS feed might automatically be set to only send the “best” feeds to a mobile device. A user would be able to set other levels as well, such as “good,” or “great.” This could help reduce bandwidth and carrier charges on phones.
In addition to filtering content based upon social metrics scores, an extended system could allow for filtering by keywords and by categories, so that you could see things that might interest you the most and in categories that you find important.
This system could also cluster content, and detect topics within content from feeds for you, so that you wouldn’t necessarily have to set the categories you were interested in before looking at that content.
Recommendations and personalization could also be included to recommend or directly display content based upon a specific user’s current preferences and habits, and what they’ve been reading lately.
The patent application also discusses how analytics could be incorporated into such a system to understand social engagement for different sites and content, and even for advertisements that might be included in feeds.
One that definitely stood out for me was a short section on the detection of influencers/authority. Just who are the top influencers and authorities in different topics, and what is the authority of specific articles?
Social Engagement Data Takeaways
The patent filing is pretty long, but it’s worth going through if you want to get an idea of why Google might have been attracted enough to the technology and the team behind it to acquire the company and hire its employees. PostRank as originally envisioned as a way to filter blog posts and articles in a feed reader no longer exist, but ideas behind PostRank involving measuring and using social engagement metrics can be used in many other ways.
The description of how those social metrics could be used as a way to filter items in an ecommerce store based upon desirability is a pretty interesting example.
We don’t know for certain how Google is doing things like determining which results to show in a search on Google Plus itself, or why Google might show some additional pages in Search Plus Your World, but there’s likely some element of social engagement metrics involved, and those likely go beyond just counting numbers of Circles or pluses or comments.
If you perform SEO and you’re not working to understand the different ways that social engagement metrics might influence the visibility of content, you may just be missing out.
Added May 1, 2012 – PostRank had been continuing to provide a social engagement analytics service on the pages of its site, though the service is scheduled to close today. Himanshu Sharma wrote about the three main items it tracked an measured in a post published this morning titled 6 things you can learn from PostRank about Social Engagement Metrics. Himanshu details what those engagement measures covers, and provides some interesting takeaways on the service. Definitely worth a read.
Added May 4, 2012 – The Google Analytics Blog made an announcement yesterday titled Expanding Google Analytics Social Reports: Tracking Links To Your Site Content. The writer of the post was Ilya Grigorik, so it looks like the PostRank team has been actively working upon bringing more of their expertise on social engagement analytics to Google. It’s not unusual for someone who has worked on a particular addition or update at Google to be the one who writes the announcement about it on one of Google’s blogs, and there’s a decent chance that happened here.
Ilya Grigorik tells us that we’ll see information about trackbacks in Google Analytics now, as well as other social signals. A snippet from the post:
Have you ever wondered which other pages on the web link to your own? Wouldn’t it be nice to know which sites are talking about your content, and in which context? Well, a problem no more: now you can see all the backlink URL’s, post titles, and more right within the new Social reports.
This is cool…but not a great feature…google must make it a 50-50 between them….ie not too much nor too low….thats it
Makes total sense Bill, and even without researching the patents (thanks for doing so – i dont know where you find the time!) we are all getting the sense of more and more social factors influencing results. If your site/content/products/services are good enough, then people should be talking about you and you should be ranking for your keywords. Its a simple concept (that’s also overwhelmingly complicated).
If Google ranks based on Social Metrics like comments, delicious, digg, twitter etc then amateuer blogs have to shutdown. Only top blogs get natural diggs,delicious, twitter links etc unless its manipulated.
@Mahesh – That was my first thought as well and that will probably be true in a small set of niches (the large popular ones). However there are a lot of niches where there are no real debate and sharing (e.g. industrial plumbing) will take place and therefore this social elements will have not of little effect. At least that is what I think 🙂
@Mahesh Mohan,
Quote”If Google ranks based on Social Metrics like comments, delicious, digg, twitter etc then amateuer blogs have to shutdown.”
No they get comments.
I agree with Mahesh, this really doesn’t level the playing field. Also I wonder which metrics will be important, with so many social media sites now available will content now need to be syndicated to them all?
At SXSW 2012 in the “Dear Google & Bing: Help Me Rank Better!” session Google and Bing both made the point that social will gain importance in ranking. With the information from the post above I can see just how this is possible. Of course it will be one of thousands of signals in the ranking algorithm. Google is also leveling the paying field to give new and smaller sites a chance with recent updates as discussed in the session, so I don’t think the amateur bloggers have to worry.
From what I know about Google and experienced first hand I believe the PostRank contribution will more heavily impact those who are signed into Google and have choose to share social data with Google by Gmail or Google+.
I dont understand how Google can rank post rank. Lets say a company shrewd me over and I post what they did to me on my facebook. Does this mean I got shrewd and now I am helping them screw everyone else? Does this mean any publicity is good publicity?
Google has everything in place to make it very, very simple. Google+ gives them the insight they need, but their main problem now is continuing growth beyond the tech industry.
It’ll be interesting to see what they do with PostRank and how it all weighs in with their index.
@Monika
What I mean is… Twitter, Facebook and other top social networks are all dominated by posts from TechCrunch, Mashable, SEJ, CNN etc. So they will probably get a further boost.
@Mahesh Mohan
you mean old bulls.
If you do the same as old bulls you have no chance, if google is measuring social networks or not.
You have to find your hunting ground =>your unique feature or your emotional unique feature.
I think it will be interesting to see how results are presented when one type of site (perhaps low end populist pages) that gets a lot of Twitter comments but might be less relevant might regularly out rank a more relevant site but is less likely to win a popularity contest with teenage girls on Facebook so gets pushed down the search results. I think this can only work with popular media events, celebrity gossip, current affairs, sports events like the Olympics etc. but the bulk of serious search results could be harmed by this.
Hi Aromona in respnse to your statement “Google is also leveling the paying field to give new and smaller sites a chance with recent updates as discussed in the session, so I don’t think the amateur bloggers have to worry.” I don’t believe Google has levelled the player field at all as with the penguin up date massive brands have have the biggest benefit!
Hey Bill, great write up. I didn’t realize the impact that the acquisition PostRank is having on Google. Himanshu’s post was eye opening (BTW the link to Himanshu’s post is broken). I’m really surprised to see that social signals don’t have as large of an impact on rankings as they should. It’ll be exciting to see how much social signals will impact rankings in the future.
G’Day! Seobythesea,
Speaking of which, Direct reaction entrepreneurs know the significance of headlines. They devote a great deal of time brainstorming headlines that “simply click.” They pay even extra afterwards to exam which headlines do the job best to push everyone into responding to the offer you.
Cheerio
I too would be disappointed if the PostRank acquisition didn’t translate into something more substantial than social reports in Google Analytics. Soon after the purchase I wrote about how PostRank fit into Google’s desire to measure the engagement graph.
The PostRank system was interesting but really, rather crude. I imagine that the resources that Google could give the team would allow them to take it ‘to the next level’. I think it’s also valuable to see the type of things that are contained in the Google+ Activity API and the Activity Strea.ms framework.
Visitors to a blog are like customers, those who comment are happy customers. Savvy blogs and e-commerce businesses have been developing relationships with their visitors for years drawing them back for return visits. It’s called branding and engagement and it’s a good thing if you’re in marketing.
I see too many built for SEO blogs that have all the technical elements in place (good titles, good architecture and mostly original content – so called “amateur blogs” almost never own their own media) that do nothing to differentiate themselves from the pack. They built it and usually spent considerable time doing so now expect to be rewarded but it doesn’t work that way. Just because you have the best food in town doesn’t mean you’re going to get customers in your restaurant.
Figure out how to get people in the door and then engage and you can compete with the big boys…or else. The Field of Dreams analogy no longer applies.
In certain niches there certainly will be very limited engagement. Perhaps people are less likely to comment on a plumber’s blog. But fear the guy who figures out how to do it well, people should strive to be that guy. I’d say in area where engagement opportunities are limited, surveys and questionnaires are a good starting point.
@Alan it is true that some big site have had a boost but there are a number of small sites that have received a boost as well. Sites with clean SEO and good users interaction are seeing benefits.
Hi Mahesh,
I’m not sure that I agree. I’ve seen blogs that get shared via social networks, listed in social sites like hacker news and elsewhere that aren’t business or professional sites. There are also a lot of different reasons why people blog, and many are often less interested in how they rank in search results than that they can share with others who might be interested in what they write or blog about.
Hi Shony,
It’s hard to tell how much influence social engagement metrics might have on search results in the future. We are seeing from Google’s PostRank acquisition that they are interested in learning better how to track and measure those kinds of signals, but have no idea at this point how they might use them.
Hi Eric (evolvor)
Thanks. I sort of consider researching patents and papers part of doing some due diligence and research that’s less likely to set me off in wrong directions than the research I might do elsewhere. 🙂
Good points about social networks and signals. Links seem so much less complicated, and yet we all have been learning how to be social for most of our lives. How tough can that be? Pretty tough, actually. 🙂
Very interesting post and an interesting way of segmenting peoples comments, reviews and consumption patterns.
When talking SEO the post of Ilya Grigorik about the posibility of viewing all backlink URL’s, post titles, and more in the Social reports is a great plus. 🙂
This was needed very soon as social sites fetched some of the best stories from around the world. The bigger brands are doing well on the social front and taking care of every social signal counted by Google. Lesser known blogs would find it harder to compete with them. This patent would do good for the social sites as more people especially marketers would be using them. The end result of getting relevant results on the search engines is the thing to watch out for. Currently, Google is doing too much in making constant updates which is making the rankings fluctuate so often.