SMX 2011: Social Signals & Search

A particular current obsession of mine and a topic on which I have about three gigs in the next month, so really hoping for some interesting thinking on this subject.
Panellists are; Bas van den Beld of State of Search, Cedric Chambaz of Microsoft; Marcus Taylor of SEOptimise and Jim Yu of BrightEdge.

Bas van den Beld

Google are clearly seeking to define user intent which is of course a difficult thing to do. Quite often user intent and be inferred by UX data, e.g. rich snippets vs ordinary results will give richer data feedback.

Don’t think that Google doesn’t “get” social – perhaps not in the network sense but later PageRank iterations incorporate “social” elements (link popularity).

2 Ways social is integrated

1. Social media content integrated in search, Hotpot – recommendation engine, Google +1

bas says don’t make the mistake of thinking +1 is analogous to “like” it’s a qualitative research tool; a CTR/popularity research method.

Sites like Quora etc are indexed.

2. Google profiles are ranking higher and are being pushed (obviously because this as rel=xfn data hub is the key to data growth in social circle and +1)

G profiles now surface highly, and in a test we (Bas, Joost and I did) I can actually see Joost’ private Google profile data as a 2nd degree connection. I.e. I’m connected to Bas, who is directly connected to Joost, and via that rel=contact link on G profile inclusion the extended reveal occurs.

In addition – Bas recently has seen personal signals attributed to searching whilst logged in on a State of Search account. Clearly using IP – relative contact details etc Google have assumed with a degree of certainty that “generic” account  is “Bas”.

Social Building Blocks

  • Identity and profile
  • friends & contacts
  • Activities
  • location

Marcus Taylor

A case study on the importance of FB “likes” and search/seo impact to ranking and indexation.

Correlation between pages that rank well and pages that have been FB shared or liked is high. Correlation studies are great but Marcus wanted to look at causation.

Looking at messaging it is very ambiguous and conflicting therefore experimentation is required.

1. February – two non-indexed domains. 0 backlinks no ability to ping. Both domains has “likes” added progressively (10 a day). Within several hours of starting the “likes” the domains were indexed.

Test was expanded to 100 non-indexed URLS, and varying numbers of likes and shares per URL commenced. Tracking raw log files to observe first visit from Googlebot. Result = 0 visits from Googlebot, 0= pages indexed. Didn’t work icon_sad-2025999

Has something changed?

Second experiment was post-Panda and post +1 launch (any connection?)

Is it the traffic that FB drives rather than the “like” action that contributes to observed indexation/rank effect?

Jim Yu

Will be talking at SEO in social, multiple KW data in multiple industries again with correlation studies.

%age of listings in top 10 high correlation with high social activity.

Important to look at relative playing feed (level of social per vertical).

What are the elements?

Like and share buttons on deep product pages is an obvious action, but thinking about temporal products e.g. anti-virus, this is extremely relevant and seems to correlate to high P1 placements.

Shareable tools – same rule applies; viral content assets too.

Final point – create parallel structures of social objects and interlink. E.g. Amazon Fashion FB page, Twitter account, etc and self-referential link matrix to site section.

We’re having a technology fail so I’m publishing now – check back for final speaker plus Q&A

Cedric Chambaz

Microsoft are all about relevance. Link relevance is of course only one external criteria.

Nowadays we have had 100,000% growth in web since emergence of Google.

25% clicks lead to “back” 42% of sessions need refinement and almost 50% of searches last 30 minutes. Still the experience is generally frustrating.

Cedric shows example of a query on Bing “cheap running shoes” – concept of cheap is relative. Plus Cedrics’ idea of a “running shoe” will definitely differ from that of Usain Bolt. Search by it’s very user-inputted nature is a personal request.

What is more personal than your phone and address book?

Cedric would like to draw a distinction between searching social… and social search. Searching social = Go to bing.com/social and you can see social results (Twitter). In addition they integrate the firehose API into Bing maps.

Social search = serving results unique to you (NS>er… isn’t that anti-social? i.e. personalised search – Je ne comprend pas).

Bing/MSN logged in to FB – gives me FB like data informing my results. Valid for shopping, restaurants etc. In addition they have RichAd inserts which incorporate social data which seems to be interdependent of login. In addition you can have rich media e.g. we’re looking at a video below the ad – which is then shareable.

Everything you do in the social media spheres will appear in the search engine at some point. Interestingly Cedric says crisis data will be retained longer (presumably because popularity and demand feedback from volume and CTR triggers a data retention signal.)

Q&A

Q will/do social signals help/affect established sites more than fresh sites.

A – Marcus – yes but of course established sites have existing links.

Leave a Reply

Your email address will not be published. Required fields are marked *