Matt Rogerson, Director of Public Policy, Guardian Media Groupwritten evidence (FOJ0114)

 

Factors influencing news rankings in search

 

I write in relation to some points made to the Committee’s ongoing investigation into the ‘Future of Journalism’ regarding the various factors that influence the ranking of news in search.  I write, in particular, to provide what I hope will be useful background to evidence provided by DMG Media which sought to claim[1] that subjective factors were at play in relation to the ranking of Daily Mail content in search.  DMG Media noted in evidence that,  

 

MailOnline’s three month rolling average share of visibility in Google search for the term 'Brexit', at the very height of the Brexit debate, was 1.29pc. The Guardian's was 20.83pc - 16 times as high - and the BBC's 11.49pc.”

 

At a broad level, Guardian Media Group (GMG) would agree that algorithm changes implemented by large online platforms can have an impact on the visibility of news brands.  In the case of Facebook, this can include explicit moves[2] to reduce the volume of news within the news feed.

 

In the case of Google, we know that major updates have impacted on the visibility of Guardian News & Media (GNM) journalism within Google search.  While it is sometimes suggested that these changes result from some kind of inherent bias regarding the political outlook of a publisher or the content itself, we believe objective analysis shows that ranking in search engine results pages (SERPs) is overwhelmingly determined by specific, technical factors relating to websites - their speed, quality of links and other related factors, which we describe below.

 

We believe it is important that these technical factors are taken into account when considering the prominence of news among search and other platforms. We believe these factors should be made more explicit and transparent by platforms, in order to rebuild confidence among regulators, policy-makers and publishers themselves. But we have seen no evidence to suggest that there are "politically-biased algorithms" which affect SERPs.

 

By comparing traffic to global GNM properties with a definitive list of core search updates[3] in the last two years, we can identify two specific dips in traffic.

 

        Mobile Speed Update July 9, 2018

     The SEO expert site moz.com stated of this change: ‘Google claimed that this only affected the slowest mobile sites, and there was no evidence of major mobile rankings shifts.’[4]

     This update had a major and lasting negative impact on Google referrals to the Guardian.

        March 2019 Core Update March 12, 2019

     The website moz.com ‘confirmed a "core" update, stating it was the third major core update since they began using that label. No specific details were given about the nature of the update.’

     As a result, we have no understanding of why we were adversely affected.

 

Competition authorities across the world are right to recognise that where algorithms are subject to significant change by platforms, the reasons and consequences of those changes, should be outlined in advance to publishers, where possible enabling publishers to provide feedback on proposed changes.  In relation to specific cases highlighted to the Committee, GMG believes that there may be objective reasons why one news publisher may rank more or less highly than another in relation to a specific search term.

 

Factors affecting ranking within search engine results pages (SERPs)

 

In an article for the Guardian in September 2018, Stanford academics Jeff Hancock, Danaë Metaxa-Kakavouli and Joon Park, noted that “When Google’s search engine first launched in 1997, it started as a well-documented academic project. Its now-famous algorithm, PageRank, was totally transparent for anyone to know. This had some unintended side-effects, since spammers and other self-serving actors took advantage of their knowledge to try to trick Google into prioritizing their content. After years of the ensuing arms race, with Google making secret changes to the algorithm and spammers trying to game it, the result is that Google’s search algorithm is a behemoth whose inner workings are a near-complete mystery outside the company.”[5]

 

As a result of balancing the need for transparency of factors affecting search ranking, and the need to preserve a degree of opacity as to how search rankings are determined, it is not clear precisely how any single factor affects ranking in SERPs.  There are, however, a range of factors that we understand do have an impact on where publisher content ranks within SERPs.  The table below, for example, sets out 10 factors which SEO website optinmonster.com suggests impact on search ranking[6]:

 

A Secure and Accessible Website

Page Speed (Including Mobile Page Speed)

Mobile Friendliness

Domain Age, URL, and Authority

Optimized Content

Technical SEO

User Experience (RankBrain)

Links

Social Signals

 

Real Business Information

 

 

 

In relation to a number of the key factors listed in the table above, it is possible to use publicly accessible Google tools to measure the performance of websites on a technical level.  Again, these are technical criteria, which Google publicly state will affect the ranking of sites in search, not subjective or unclear criteria.  Google would argue strongly that these are objective technical criteria, the end goal being to serve their users the highest quality websites that provide the most relevant and accurate information.

 

Accessing theguardian.com and dailymail.co.uk websites via an incognito browser, it is possible to test the home page, specific editorial sections and individual articles on both websites in order to understand how those pages compare according to the publicly stated criteria on which Google ranks pages in search.

 

With screenshots of these tests included as appendix 1 to this note, the tests are clear that, in relation to:

 

        the respective home pages of The Guardian and the Daily Mail, theguardian.com performs significantly better than dailymail.co.uk across five out of five key factors by which Google developer tools rate websites;

        the Brexit-related editorial subsections of The Guardian and the Daily Mail, the relevant subsection of theguardian.com performs significantly better across five out of five key factors by which Google developer tools rate websites; and

        individual articles by The Guardian[7] and the Daily Mail,[8] the individual Guardian article performs significantly better across five out of five key factors by which Google developer tools rate websites.

 

Google’s PageSpeed insights tool gives the dailymail.co.uk[9] website an overall score of 11 out of 100, with theguardian.com[10] on 29 out of 100, and BBC News[11] on 39 out of 100.[12]

 

image14.pngimage1.pngimage9.png

 

Google’s rating tools that analyse the performance of dailymail.co.uk are publicly available to staff at the Daily Mail via any Chrome browser[13].  The metrics that Google uses to rate the performance of a website are within the control of the Daily Mail, and could be remedied by the Daily Mail if this was deemed an area of improvement in which they chose to invest.  The fact that the Daily Mail site rates poorly on key Google audit metrics, compared to other news media publishers operating in the UK market, is likely to have an impact on its appearance in SERPs.

 

Site speed and user satisfaction

 

Using Google tools[14] to compare the site speeds of a number of major international news websites, it appears that the dailymail.co.uk website suffers from significant site speed load delays. This is likely to have an impact on how Google measures user satisfaction of search results.

 

image12.png

 

One metric believed to be used by Google as a proxy for broad quality of user experience is the average speed of page loading times. Websites that are poorly constructed in technical terms, littered with low quality ads, and intrusive tracking codes, will tend to be significantly slower than more user friendly websites.

 

But search experts also believe Google measures quality of engagement and user satisfaction with search results at a story level, by taking into account users who bounce back to a search results page.[15]

 

The satisfaction metric means two things.  First, that if readers don't click on a link to a news story in SERP, then the publisher's story will be penalised as a less relevant result.  Second, if readers do click on a search result, but then bounce out quickly, the publisher's story will be penalised on the assumption that it's an unsatisfying result (either because it's not relevant despite appearing to be so) or the user experience is poor (see the appendix on site performance). This benefits publishers who present their stories responsibly, but it also benefits publishers who have made a strategic choice to invest in driving longer user engagements on the page, as the Guardian has.

 

Prominence of original journalism in SERPs

 

As an investor in high quality original journalism, GMG believes that trusted, original journalism is something that it is in the interests of the public to be able to find through a prominent position in SERP.  The use of fair, objective and transparent metrics as the basis for search ranking is key to fostering a pro-competitive online news environment, enabling consumers to access high quality journalism within the online search environment. 

 

Fair, objective and transparent criteria are vital to ensure that the dominant positions of news organisations in an offline world are not simply used as an assumed basis for prominence in the online world.  Similarly, it is essential that the prominence of journalism is not influenced by factors such as payments to search engine providers for prominence.

 

Google has recently restated its commitment to promote “original reporting, an endeavor which requires significant time, effort and resources by the publisher,” making “ranking updates and published changes to our search rater guidelines to help us better recognize original reporting, surface it more prominently in Search and ensure it stays there longer. This means readers interested in the latest news can find the story that started it all, and publishers can benefit from having their original reporting more widely seen.[16]  This can mean, for example, that where news websites republish agency copy or publish copy produced without a named byline, this may not feature as prominently in search results as genuinely original reporting from a journalist with a recognised track record in a specific subject field.

 

The focus by search engines on rewarding investment in original report and journalistic excellence is crucial to creating a competitive, pro-consumer online environment.  As a consequence of creating these incentives within SERP, it may mean that articles produced by the biggest incumbent news providers may not always be ranked first in search, even if their website produces journalism in other verticals that is widely read by the public. 

 

The promotion of original fact-based journalism by a native digital publisher over and above journalism produced by an incumbent, should be seen as a sign that competition is working, not that there is some kind of subjective bias within the SERP algorithm against the incumbent.  It would be a poor outcome for consumers in any sector, if, simply by dint of commercial incumbency, SERPs were calibrated to promote products that were of a lower standard or poorer quality.

 

Investment in key verticals

 

As noted above, in addition to investment in a world class website, world class audience teams and in-house audience analytics tools,[17] GMG has led its growth strategy with investment in world class journalism on issues that matter to an engaged global audience.  Through investment in deeper coverage across many areas of politics, the implications and negotiations around Brexit, the climate crisis and coronavirus, GNM has built a growing audience and strong presence within search rankings.  Such prominence is not due to the subjective bias of a search engine, but to the time, effort, and resources that have been invested into amazing journalists, who produce world class journalism every day.

 

The decision on which verticals to invest in is down to each individual news publisher, depending on their business strategy.  We know for example, that in terms of harnessing programmatic audience-based targeting from advertisers, there is currently less appetite for advertisers to place marketing around harder news content[18].  This may incentivise a publisher to invest more resources into sport, celebrity, lifestyle and viral content that is more likely to attract advertiser spend.  It is a reality that while articles about coronavirus are some of the most important and most well-read articles on our website, the increasing use of keyword blocking technologies against related words is having a massive impact on the ability to monetise advertising across our site.[19]

 

From a GMG perspective, while advertising remains a key revenue stream to support a sustainable commercial business model, investment in key public interest news verticals serves the broader purpose of building engagement, loyalty and a willingness to pay amongst a growing readership.  We absolutely respect the business decisions of other news organisations to prioritise greater investment in ‘softer’ news verticals.  We would expect, and observe, that as a result of this investment and specialisation, articles from those news organisations in those softer news verticals appear more prominently in Google search.

 

A number of examples of keyword searches demonstrates this point (see appendix 2 for screenshots of relevant searches).

 

Searching for Kim Kardashian in Google search, over a third of search results are for the Daily Mail and Metro, both of which are owned by DMG Media.  The Guardian does not rank anywhere on the first page of SERPS.

 

Chart

 

Searching for “TV gossip” results in The Sun appearing first with the Daily Mail in second place.   The Guardian does not rank anywhere on the first page of SERPs.

 

Chart

 

A search for “Royal News” delivers an overwhelming search victory for the Daily Express, occupying six out of the first ten search results.  The Guardian does not rank anywhere on the first page of SERPs.

 

Chart

Investment in social media

 

Different publishers take different approaches to distribution of their journalism, depending on the strategic focus of the business at any given time.  As well as noting the ongoing investment that GMG makes in The Guardian’s website and other digital products, the key focus for GMG is building deeper relationships[20] with readers of The Guardian.  This includes focus on a strategy to produce less journalism,[21] but driving a higher page view or engagement time with each article.  Other publishers may take a different approach, seeking to drive a higher number of unique engagements, with a greater volume of journalism.

 

The Daily Mail is a hugely popular online news brand, consistently the highest ranking UK news publisher on Facebook.  Data published by Newswhip[22] for the month of August 2020 shows that the Daily Mail is the fifth biggest publisher on Facebook globally, with over 63 million engagements.  The data also shows that the Daily Mail published more articles than any other publisher in the top 25 publishers on Facebook, publishing forty-eight times more articles (55,432 articles versus 1,155 articles) than were published by the no.1 news publisher on Facebook, the conservative leaning dailywire.com, and foxnews.com among others. 

 

A distribution strategy to ensure prominence on social media is very different to a distribution strategy that seeks to drive prominence and engagement through SERPs.  These are choices that individual publishers make when determining how their distribution strategy relates to their wider business strategy.  There is no one size fits all approach to the distribution of journalism online, no one model that can deliver equivalent success on all search and social platforms.

 

Academic research on search rankings

 

In the face of accusations that Google search rankings are biased or manipulated, a growing body of academic research has been commissioned to examine these claims.

 

For example, academics at the Stanford Media Lab published research in November 2019 which examined the coverage of candidates in the 2018 US election over a six month period.  The study reviewed the first page of Google search results for every candidate running for federal office in the 2018 U.S. election over a six-month period. After a systematic audit of about 4 million URLs scraped from the search engine, they found that sources from either end of the political spectrum are not being excluded from results. For the most part, the researchers found that the news sources most commonly held a relatively centrist point of view.[23]  Commenting on the research, lead academic Jeff Hancock noted that, Our data suggest that Googles search algorithm is not biased along political lines, but instead emphasizes authoritative sources.[24]

 

In June 2019, The Economist built a predictive model to examine whether there is bias in search algorithms based on the political position of publications.  The Economist experiment “a browser with no history, in a politically centrist part of Kansas, we searched for 31 terms for each day in 2018, yielding 175,000 links. Next, we built a model to predict each sites share of the links Google produces for each keyword, based on the premise that search results should reflect accuracy and audience size, as Google claims. We started with each outlets popularity on social media and, using data from Meltwater, a media-tracking firm, how often they covered each topic. We also used accuracy ratings from fact-checking websites, tallies of Pulitzer prizes and results from a poll by YouGov about Americanstrust in 37 sources.

 

The research noted that if Google’s algorithm did hold a subjective bias for a particular point of view, the study should find that liberal andleft-wing sites would appear more often than our model predicted, and right-wing ones less. We saw no such trend. Overall, centre-left sites like the New York Times got the most linksbut only about as many as our model suggested. Fox News beat its modest expectations. Because most far-right outlets had bad trust scores, they got few search results. But so did Daily Kos, a far-left site.[25]

 

image13.png

 

Again, the Economist research suggests that alongside the satisfaction of key technical requirements outlined at the top of this document, the Google algorithm does not hold a subjective bias against certain publishers or points of view.  Rather, it prioritises journalism published by trusted, authoritative news sources, which is a positive outcome for consumers seeking out reliable news and information.

 

Similarly, in ‘Auditing Partisan Audience Bias within Google Search’, academics from the Northeastern University, Boston, USA, performed an “audit of partisan audience bias and personalization within Google Search”[26]. The research examined the search results of 187 participants who both completed a survey and installed a browser extension to enable the collection to SERP from their computers. The research found that results positioned toward the bottom of Google SERP were more left-leaning than results positioned toward the top, and that the direction and magnitude of overall lean varied by search query, component type (e.g. "answer boxes"), and other factors. Utilizing rank-weighted metrics that we adapted from prior work, we also found that Google's rankings shifted the average lean of SERP to the right of their unweighted average.

 

 

In Burst of the Filter Bubble?, academics from the Department of Communications and Media Research, LMU, Munich, Germany, sought to test the effect of both implicit and explicit personalization on the content and source diversity of Google News.”  While the report found minor implicit personalization on content diversity, it did not find evidence to support a filter bubble hypothesis.  The report did find that Google News overrepresented certain news outlets and under-represented other, highly frequented, news outlets. Given the over-represented outletsconservative nature, this bias can be troubling, especially in terms of viewpoint diversity.[27]

 

 

29 September 2020

 


Appendix 1 - Comparing site performance using Google Lighthouse tools[28]

 

Tests were run on 28th September between 1.30 - 2pm using a Chrome browser running in incognito mode.

 

theguardian.com                                                                      www.dailymail.co.uk

 

image2.pngimage6.png

 

https://www.theguardian.com/politics/eu-referendum

 

image10.png

 

https://www.dailymail.co.uk/news/brexit/index.html

 

image7.png

 

 

 

 

 

 

 

 

https://www.theguardian.com/politics/2020/sep/28/confidence-in-uks-global-role-plunges-after-brexit-poll-finds

 

image11.png

 

https://www.dailymail.co.uk/news/article-8779337/Boost-Brexit-talks-sides-make-key-concessions-deadline-looms.html

 

image3.png

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 


Appendix 2 - Google search results on specific issues

 

Tests were run on 28th September between 2 - 2.30pm using a Chrome browser running in incognito mode.

 

Google search results for “Kim Kardashian

 

image17.png

 

Google search results for “tv gossip

 

image15.png

 

 

 

 

 

 

 

Google search results for “Royal News”

image16.png

14

 


[1]              https://committees.parliament.uk/writtenevidence/9118/pdf/

[2]              https://www.facebook.com/zuck/posts/continuing-our-focus-for-2018-to-make-sure-the-time-we-all-spend-on-facebook-is-/10104445245963251/

[3]              https://moz.com/google-algorithm-change

[4]              https://moz.com/google-algorithm-change

[5]              https://www.theguardian.com/commentisfree/2018/sep/06/google-search-results-rigged-news-donald-trump

[6]              https://optinmonster.com/seo-ranking-factors/

[7]              https://www.theguardian.com/politics/2020/sep/28/confidence-in-uks-global-role-plunges-after-brexit-poll-finds

[8]              https://www.dailymail.co.uk/news/article-8779337/Boost-Brexit-talks-sides-make-key-concessions-deadline-looms.html