SearchLove London | Dave Sottimano, 'Using Data to Win Arguments'

Data driven SEO

David Sottimano

Searchlove 2014

Can a post rank solely by having keywords

in the URL?

Yep.

What does meta NOINDEX do?

Removes a page from the index..

But it can lower Googlebot crawl rate too.

Are meta keywords actually useful?

Don’t be silly.

Data driven SEO Using data to win arguments

David Sottimano

Searchlove 2014

Do this.

Because. {Insert Matt Cutts video link}

Caveat, caveat, caveat….

Meaningful, conclusive data is hard to come by.

Algorithms can be specific to queries.

http://searchengineland.com/google-pay-day-loan-algorithm-google-search-algorithm-update-to-target-spammy-queries-162941

Data we need is out of reach.

Actual click

through rates?

Actual bounces

back to search

results?

Our “good” isn’t Google’s “good”

Clues are scarce, and often vague.

Source: http://insidesearch.blogspot.com.es/2012/04/search-quality-highlights-50-changes.html

Would you trust the information presented

in this article?

http://googlewebmastercentral.blogspot.co.uk/2011/05/more-guidance-on-building-high-quality.html

Presence of author

Presence of author information

Presence of author image

Presence of logo

Presence of contact information

Presence of social proof

This is why we need a data driven approach.

Because “best practice” isn’t a good

enough answer.

Throwing stuff against the wall doesn’t

make us any wiser!

Be curious!

Question everything!

More input, less valuable output

Sometimes, simple is best.

How’s this idea guys?

It’s pretty shit.

*not actually what they said

How I completely failed* to win arguments before.

*pretty much all the time

This could have been avoided.

If I had done this…

Keyword If you move off page 1 Money you will lose

Keyword 1

-3,000 visits

-$10,000

Keyword 2

-2,000 visits

-$7,500

-5,000 visits per month

-$17,500 per month

My first time.

“We’re going International, what do we do with hreflang?”

Get the right people to the right pages in search &

Don’t screw up rankings / traffic

Hreflang, canonical or both?

Okay, test it.

> 2 Analytics

WMT

Rank tracking

Logs

Testing configuration

Did you know Distilled had an Australian office?

Think about all the variants you want to test first

Ask for testing methodology feedback.

Wait.

How will I know if it worked or not?

1) Rankings

2) Organic traffic

3) The right pages display in the right countries

Custom reports

Fancy shmancy segmentation

mmm custom dashboards

Share it with clients to follow along.

Set it and move on. Remind yourself!

So, what happened with the hreflang project?

No conclusive ranking improvements Display issues completely corrected

A few tips.

Scenario1: I forgot to track the data.

Historical search results

http://www.semrush.com/info/gmail+download+all+attachments+(source)?domain=davidsottimano.com&position=4&ts=1413494980

Historical screenshots

http://www.screenshots.com/ https://archive.org/web/

Historical rankings (specific keywords)

http://www.spyfu.com/Ranking

Scenario 2: How do I find examples around the web?

Brilliant source code search, by Nerdydata.com

http://nerdydata.com/

Peek by Linkrisk. Search by SEO metrics.

http://linkrisk.com/peek/

Scenario 3: I can’t open the entire CSV in

Excel.

No, I don’t know how to code.

No problemo.

Use one of these.

http://delimitware.com/

*windows 7 >

http://recsveditor.sourceforge.net/c

sv02.htm *independent

Scenario 4: I need to gather data from

webpages.

I don’t know how to code.

Scraping is fun, really fun.

https://import.io/ http://scrapinghub.com/scrapy-cloud

The (highly experimental) future

Search is becoming too complex.

Why are we trying to analyse vast amounts of

machine data?

Why not fight fire with fire?

I had goals…

Reverse engineer why Distilled blog posts do well in search.

And predict how successful new blog posts would

be (organic traffic)

I foolishly expected...

and failed.

URL Majestic Status URL Majestic CitationFlow URL Majestic TrustFlow URL Majestic Ext Back Links URL Majestic Ref Domains URL Mozscape Domain Authority URL Mozscape Page Authority URL Mozscape External Equity Links URL Mozscape MozRank URL Mozscape MozTrust URL Mozscape Subdomain External Links URL Mozscape RootDomain External Links URL Mozscape Juice Passing Links URL Mozscape Subdomains Linking URL Mozscape Root Domains Linking URL Mozscape Links URL Mozscape Subdomain Subdomains Linking URL Mozscape Root Domain Root Domains Linking URL Mozscape Subdomain MozRank URL Mozscape RootDomain MozRank URL Mozscape Subdomain MozTrust URL Mozscape Root Domain MozTrust URL Mozscape External MozRank URL Mozscape Subdomain External Domain Linking Juice URL Mozscape Root Domain External Domain Juice

Reading Time Sentiment Sentiment Score Dale-Chall Score Flesch Kincaid Grade Level Flesch Kincaid Reading Ease Score Flesch Kincaid Reading Ease Gunning Fog Score Smog Index Images Images with Alt Images without Alt Videos External Link Count Internal Link Count Total Link Count Author Author URL Robots File Allowed Robots Meta Robots HTTP Header Canonical HTTP Header Canonical Head Date published Year published Alchemy Sentiment score Alchemy top concept Alchemy top keywords

HTTP Status Redirected Original HTTP Status Code Original HTTP Status Content Type Content Length URL Google Indexed Hash HTML Length Text Length Text to HTML Ratio Title Title Length Description Description Length Word Count Sentence Count Header Count Paragraph Count Last cached date # likes # shares # tweets # retweets # g+ Theme (custom) Type (custom) Alchemy entity Sessions Bounce rate

I used organic sessions as my objective field, to

classify what was good/bad.

Mean

Good

Bad

0

~16,000

~110

< 20%

90% >

80 70 60 50 40 30

Not so

interesting

So, longer posts = profit?

I fed garbage in, and got garbage out.

Tip! Don’t use metrics that are well correlated with rankings.

There’s so much opportunity here.

So what can you do about it?

Get better at defining “great content”.

If it gets links, shares, converts, we usually

class it as “good”.

But what made it “good” ?

Tutorial

Technical > contains code

Controversial

Breaking news

Funny

Serious

Off topic

Controversial

List post > top 5,10, checklist

Tool review

Try it. A free version is available.

http://goo.gl/NKtXOl

Two little things I want you to remember.

Build a better practice by binning best practice

Prove it.

Data or it didn’t happen

Thanks

@dsottimano