SEO Reporting with (not provided)
- (not provided) keyword traffic affects all analytics platforms.
- Google SEO keywords used to hold a referral with q parameter holding keyword e.g. q=keyword
- Now redirects referral to take out q parameter, but retains other information
- Distinct from (not set) or (direct) traffic, which has no referral information at all e.g. iOS browsers, https secure search
- What Data Is Left?
- User/Page Based Reporting
- Current Tools
What Data Is Left?
- Google Webmaster Tools
- SEO Rank Checkers
- Google Referral Information
- Historic Analytics Data
- AdWords Keyword Reports
- SEO User Behaviour
Google Webmaster Tools
- Keyword - Sometimes misses important keywords
- Impressions - rarely displays all SEO impressions
- Clicks - rarely displays all
- Avg. Position - looks reliable compared with other ranking methods
Top Pages - Same as above but on URL level
Can download via Python tool, but keywords per page only available manually
Impression/Click metrics only useable for trend splits
SEO Rank Checkers
SEOmoz, AdvancedWebRanking, AuthorityLabs etc.
- Can provide what keywords rank for which URL
- Prone to personalisation, location inaccuracies
- Against Google ToS
Can be used to find which URLs are ranking for which keyword, but rankings may be inaccurate.
Google Referral Information
Take the Google referral URL, find what other information is available aside from query.
- Keyword Ranking still available for now (cd=)
- What type of search result click (sitelink, video, knowledge graph etc.)
- Landing Page URL
Can be used to narrow down what type of result hit a page, to compare with Rank Checkers
Data from 2012 verifies GWT Average rankings well (85% correlation with median position)
Historic Analytics Data
- Take metrics from keywords from when you had data
- Use Forecasting, Seasonal cycles to provide what trends keywords take
Paid Search still has all metrics available.
Run exploratory AdWords campaigns to get benchmark data, or use existing campaigns.
- Get reliable Search Impressions
- Per Keyword Conversion Rates
- Use GWT link with AdWords to find Paid/SEO relationships
SEO User and Page Based Reporting
SEO User Based Reporting
- Keywords are powerful as they indicate the intent of a user hitting a website.
- But can this information be found elsewhere?
- Cohort Analysis - get demographic data from a user, infer what lifetime value of that user is and where SEO falls in that journey
SEO User Behaviour
How users behave onsite differs depending on which keyword they arrive upon.
Examining branded multi-touch behaviour in study for a client, we found:
- Brand Searchers touched website ~10% more often than non-brand
- Brand Searchers average number of touches before conversion were ~150% less than Non-brand
Can be used to make judgements on which visitor type hits a page from (not provided)
Click Through Rates
- As keywords change position, amount of traffic fluctuates. (i.e if ranking goes down, traffic goes down)
- If observed keyword changes rank, check landing URL for projected change in traffic.
- CTR from GWT and industry benchmarks.
- Requires historic record of keywords and positions
Page Based Reporting
With loss of SEO metrics by keyword, SEO metrics by URL page helpful.
Good SEO fundamentals provide inference of SEO keyword behaviour:
- Client Study showed average of 70% of keywords hitting URL were also in URL's
(not provided) % split
- Take all keywords that are not (not provided), find % split distribution
- Apply to (not provided) traffic, estimate what % of traffic are attributed to other keywords
Will only work with big samples of keywords, which are dwindling to 0.
Use GWT and other data to find split in future.
URL Split of (not provided)
- Take landing page URL for SEO keywords
- Download title tag, metrics per URL
- Find % split of (not provided) for each URL
Benchmark against pre (not provided) era, to find % of keywords that are in title tag
- Useful metric for focusing title tags for SEO
- Narrows down list of possible keywords to landing URL
Google Webmaster Tool API
- Only way to download Search Queries is currently Python tool.
- Allows download of Top Queries, Top Pages
- Lacks Keywords per Top Page.
- Runs every week for historic archive
SEO keyword Forecast
- Apply to historic data to find seasonal trends.
- Apply to all SEO traffic or individual keywords
SEO keyword CTR Prediction
- Apply to Impressions and ranking keywords to find projected traffic
- Unreliable under position 10
- May have different CTR distribution depending on query
Monitoring SEO Rank changes
Predict traffic changes according to CTR and rank change
- Monitored keyword increases from position 5 to position 1
- CTR expected to raise by factor of 5.5
- Repeat for every SEO keyword pointed at URL in rank checker
- Calculate overall traffic change
- Compare with actual traffic change to that URL
SEO keyword Clustering
- Apply Machine Learning to find user behaviour
- e.g. Split brand vs non-brand behaviour
Putting all above together
- Shortlist what keywords are likely to hit URL.
- Compare with keywords that URL is ranking for
- Infer traffic split of keywords
- Monitor changes to rankings, project changes to (not provided) traffic for displayed URL
- Account for seasonal and forecasted traffic volumes
Custom Channel Grouping & utm tagging best practices
Assisted conversion report for Social Paid is broken
Because Default Channel Grouping is wrong
Here is a Private Channel Grouping fixed example
Expected GA utm_medium for social are
- sm (like cpc)
- social (like organic)
- social network
- social media
Social Paid should be...
Social Organic should be...
Social Referral should be...