Engaging your customers, members and/or constituents with Sitecore’s OMS ○Prepared by Carla Brown, National Wildlife Federation ○and Amanda Shiga, non~linear creations ○April 2011 ○ Please blog and tweet about our presentation - #Dreamcore
Engaging your customers, members and/or constituents with Sitecore’s OMS
○Prepared by Carla Brown, National Wildlife Federation○and Amanda Shiga, non~linear creations
○April 2011
○Please blog and tweet about our presentation - #Dreamcore
Who We Are
Carla Brown, Senior Manager Online Production
National Wildlife Federation○ Largest conservation non-profit in the U.S.○ Protecting wildlife, addressing global warming &
getting people outside in nature
Amanda Shiga, WCM Practice Area Lead
non~linear creations○ Founded in 1995, 70+ employees○ Sitecore partner since 2007○ Offices in Ottawa, Toronto, Calgary, New York○ Digital strategy & marketing○ WCM, search, systems integration
Poll: What Excites You the Most about OMS?
○Creating personalized content
OR
○How it handles multivariate tests
OR
○What it will teach you about your users?
What we want to share with you today
○Starting with small steps: our foray into getting the most out of OMS
○Lessons learned
○Tips & tricks
What is the Online Marketing Suite (OMS) ?
Online Marketing Suite
Omniture or Google Analytics
Page Visits, Unique Visitors Yes YesTrack an individuals route through your site
Yes Yes
Track how a group of people – based on a persona – routes through your site
Yes No
Test content based on personas Yes No
The OMS changes visitors…
Into visitors with session data
Where to start?
Sitecore Story of www.nwf.org
January 2010 – Launch in Sitecore CMS
October 2010 – Persona implementation
Now – Multivariate tests
Engaging Stakeholders
Positives○ Market to specific audiences
○ Easy to put persona scores on content
○ Less complicated than Google Website Optimizer
Challenges
○ Our conversion pages mostly not in Sitecore
○ Analytics reports challenging to manipulate
○ Client data is stored client-side – makes marketers nervous – how to hook to CRM?
Initial Setup: Lessons Learned
○ Be aware of MaxMind costs
○ Exclude robots – 60-70% of traffic○ Workaround for firewall IP
How reports look without MaxMind(Note: No Company Name or Location)
How reports look with MaxMind(Note: Company Name, Location – allows personalization
based on location)
Initial set-up: More Lessons Learned
○Database tuning is critical
○Have a developer on hand for query writing
Sample query – “average number of pages per visit by persona”select y.Name, AVG(y.cnt) as average from (
select x.Name, s.SessionId,count(p.IndexableUrl) as cnt from (
select s.GlobalSessionId, pkd.Name, sum(pk.Value) as score, RANK() over (partition by s.globalsessionid order by sum(pk.Value) desc) as rnk
from Session s
inner join Profile p on p.SessionId = s.SessionId
inner join ProfileKey pk on pk.ProfileId = p.ProfileId
inner join ProfileKeyDefinition pkd on pkd.ProfileKeyDefinitionId = pk.ProfileKeyDefinitionId
inner join GlobalSession g on g.GlobalSessionId = s.GlobalSessionId
where Name <> 'Chris' and Name <> 'Sandra'
and g.VisitorIdentification < 900
group by s.GlobalSessionId, pkd.Name
) as x
inner join Session s on x.GlobalSessionId = s.GlobalSessionId
inner join Page p on p.SessionId = s.SessionId
inner join GlobalSession g on g.GlobalSessionId = s.GlobalSessionId
where x.rnk = 1
and p.IndexableUrl not like '/layouts/%'
and p.IndexableUrl not like '/sitecore/%'
and p.IndexableUrl not like '%Error-Page.%'
and g.VisitorIdentification < 900
group by x.Name, s.SessionId
) as y
group by y.Name
order by average desc
Our Personas
Wildlife Enthusiast Outdoor Enthusiast Policy Enthusiast
Our Personas
Parent / Caregiver Educator Green LifestylesEnthusiast
Step 1: Persona Implementation
○ Easy to put the persona scores on content
○ Great lesson – we don’t write to one audience often!
○ Persona scores do not propagate
○ Challenging to get marketers to do audiences, not actions (Wildlife Enthusiast vs. Donor)
Step 2: Think like a web analyst
○ Profiles ran for 2 weeks
○Captured millions of records of raw data
○How to extract meaning beyond basic traffic analysis?
–How are profiles behaving differently?
–Tie back to audience engagement goals
How hard was it to extract this information?
○The OOTB reports were great, but we needed to dig deeper.
○We spent considerable time developing a set of complex SQL queries.– We got to know the database structure extremely well.
○How best to provide results to marketers? Excel or BI tool.
Translate into recommendations: basic Qs
○ Is level of content aimed at each profile congruent with organizational goals?
“Total number of sessions as highest-scoring persona”
Who’s spending the longest on the site?
Why don’t wildlife enthusiasts stick around?
Who is looking at the most pages?
○Why do Outdoor Activity Seekers hit the highest number of pages?– Review top pages – are there clear links to more content?– Compare to top exit pages – opportunity to link elsewhere, or logical
end point?
The “aha” moment…
○How can we get high-traffic visitors to act more like engaged visitors?
User engagement: Passive role to active role
Next Steps: Personalization and MV testing
○ Target calls to action based on personas
“Give wildlife a fighting chance” for wildlife enthusiasts
“Take action to defend polar bears” for policy enthusiasts
○ Track conversions to test effect of personalization
Getting a little more technical…
oKey architectural considerations oBalance content granularity with template flexibilityo Ensure marketers can fully administer OMS features through the content author interface(s)
Content granularity
○As best you can, plan for future testing and personalization scenarios– Multivariate tests vary the contextual data source item of a
sublayout, not the sublayout itself – Beware the freedom of the rich text area– Beware data source assignment in code-behind
Beware site-wide elements!
○ MV tests work best on individual sub-layouts
○ Enable marketers to administer tests!
○ Site-wide elements (such as a footer) may be configured on each template
○ Get creative – code-behind or multiple tests Using Page Editor to add a test
to a sub-layout
Storing MV tests
○When setting up a MV test, you’ll need 2 or more test variable items
○Where is the best place to store them?
Find inspiration with Wildlife Watch
The challenge
Storing MV tests
○Two possible solutions:– A subfolder under the root – organized in one place– A subfolder under the original item – easier to locate
External conversions
○Solution: set up a special redirect template
Summary of tips
○ Start simple and ask basic questions– Tie back to website/audience strategic initiatives
○ Profiles are a great place to start– Move on to personalization and testing, depending on your
organizational goals
○ Have a reporting framework in place○ Prepare infrastructure properly
○ Architect build
○ Get ready for DMS 2.0!
Q&A
○Thank you!
○ Please blog & tweet (#Dreamcore) about our presentation!
This is Carla’s tree-hugging daughter, Nora