Knowledge Graph Makes Google Smarter
Google has been talking about a semantic refresh of their search engine for months, and it finally arrived as the Knowledge Graph on 5/16/12.
The Knowledge Graph is Google’s latest step into artificial intelligence, and their large scale attempt to understand not just keywords, but entities that those keywords represent. It’s not their first step toward entity or semantic search, as we explained in a previous POV. For years Google has been able to provide answers for simple questions related to facts, such as “when did John Lennon die?” or “when is Bastille Day?”
The Knowledge Graph; however, takes this a step further and provides information on the search results pages based on structured data and what searchers are most commonly looking for.
Eventually the Knowledge Graph will allow Google to answer more complex queries such as “how deep is Lake Tahoe?” or “how many women won the Nobel Prize?” which could transform search as we know it.
How it Works
Certain queries about entities will not just provide text links, local information, images and video, but will summarize the information in the right rail of search results.
For example, the query [charlie chaplin] provides information about the legendary filmmaker in the right rail:
The information provided on the right rail is based on a multitude of public sources, such as Freebase, Wikipedia, and the CIA World Factbook and is a result of what searchers are actually looking for. Google verified this when Google’s head of Search Quality, Amit Singhal, mentioned in a blog post that the information shown for the query [tom cruise] answers 37% of the next queries about him.
With this update, ambiguous queries become less ambiguous because Google can more accurately predict a searcher’s intent and will suggest a number of topics based on related entities. For example, a query for the Chaplin film [modern times] also shows information related to the Bob Dylan album of the same name:
Additionally, there has also been a change to the UI when you ask Google questions. Now instant answers are bolder and bigger. For example, when you ask “when did John Lennon die?” it used to return its “best guess” as December 8, 1980. Post Knowledge Graph it has gained more confidence and will return the date.
It is important to note that the Knowledge Graph doesn’t always get it right. For example, for the query [susan sarandon spouse] (trick question—she’s not married), it identified Chris Sarandon as her spouse. It can’t currently distinguish between husbands and ex-husbands. If Google wants to achieve true knowledge, it should be able to understand simple distinctions like these. As a remedy for this issue, there are inputs to report a problem in each field, which should help Google clean up any data errors over time.
These changes have also been rolled out to mobile and tablet interfaces, but have a slightly different experience in that the Knowledge Graph listings are contained within the SERP. In the listings we’ve spotted on smartphones and tablets, the Knowledge Graph listing appears no lower than third on the page.
With the Knowledge Graph update, Google will better understand what searchers are looking for and provide that information quickly, sometimes without having to click through to another page.
For the searcher, this means faster and more relevant search results, which will give searchers another incentive to use Google on a regular basis.
For marketers it’s another evolution into how Google works, which is bound to have some implications for how SEO and PPC work. What those implications are in the long term are not yet known, but there are some clear takeaways:
1) The Knowledge Graph does not signal the end of the practice of optimizing web pages, videos, and images in favor of short answers on the search results page. If anything, SEO becomes more important with this update, as for certain queries the algorithmic results may become less visible for broader queries as a result of the images and bullets in the Knowledge Graph.
Writing compelling titles and descriptions, implementing schema, having an optimized Google+ page and optimizing images and videos are all ways to get searchers to notice listings and click through to web sites. These tactics can be helpful when optimizing for broad informational queries.
2) The above optimization recommendation is even more critical for SEO on smartphones and tablets. This is because the Knowledge Graph listing appears not just on the right side, but within the first three organic listings. Since we know from Google that there’s a 90% drop off from first to fourth position in mobile results, any shift in the first three results could potentially have greater implications for the mobile results than the desktop results.
3) The bigger question to SEOs is whether optimizing for queries that can be answered by a simple yes or no is a sustainable marketing strategy. The challenge in keyword targeting is to determine which keywords are most relevant and have the best chance of bringing qualified traffic to a site. Many informational queries that are affected by this update are not the kinds of queries that would convert at a high rate or bring highly qualified leads to a site. This old problem of keyword targeting becomes more complex with the Knowledge Graph, but there are still plenty of opportunities for marketers to drive qualified traffic to their sites from Google.
4) Search engines obtain the information they display from public sources, and a company’s willingness to provide that structured data to Google does bring it additional visibility in search results.
5) There are some marketers who are speculating that this change will lead to less traffic from Google for certain queries, as searchers can find information within Google without having to go to another site. However, we remain skeptical, as historically Google has provided instant answers and OneBox results that allow searchers to receive information directly on the search results page at least since 2006.
The query [weather] potentially takes away search traffic from weather.com and other sites that offer weather forecasts, yet Weather.com managed to quadruple their organic search traffic in the last three years according to SEMRush. Time will tell the true impact on traffic, and we expect to see eye and click tracking studies in the future to better measure this impact.
6) Most Resolution Media clients will probably not notice large swings in traffic or revenue due to this change. SEO at Resolution has long incorporated semantic maps into the SEO process in order to create content that’s more relevant to a searcher’s query. The Knowledge Graph is just Google’s attempt to use entities to improve relevance. Because understanding the intent of a query and optimizing content to increase relevance to that query is at the heart of what we do at Resolution, more relevant search results will never damage our clients’ search engine optimization efforts.
1) Similar to point #3 of the SEO takeaways, currently the Knowledge Graph is only impacting a small number of informational queries (i.e. “tom cruise”, “charlie chaplin”, “taj mahal” for PPC. Generally, advertisers do not bid on these types of search terms and the impact on PPC programs hasn’t been significant.
2) Google has been moving towards integrating more consumer information into their ads over the past few years, such as: displaying products in the results (Product Extensions) or by giving searches an indication of what people think about products (Rating Extensions). Based on this evolution, we would predict that Google will continue to make their search ads more informational, which aligns with the goal of the Knowledge Graph.
As search ads become more information heavy, the weight of impressions can potentially take more precedence because consumers might make their decisions based on simply viewing ads as opposed to clicking through to the website.
3) In order for marketers to avoid a decrease in traffic to their site, they will need to ensure their on-page experiences and ad copy messaging are more compelling than what Google offers with the Knowledge Graph update.
The Knowledge Graph is Google’s attempt to make their search engine understand entities beyond the keywords that represent them. For searchers, this should mean faster and more relevant results. For marketers focusing on delivering relevant, high-quality optimized content this change should only bring more visibility to their sites. Optimizing content to bring more visibility to SEO listings (e.g. image and video optimization, schema implementation, compelling keyword-rich titles and descriptions) and for PPC, developing compelling on-page experiences and information rich search ad copy, will be even more important now.