Recommended
More Related Content
What's hot
What's hot (20)
Similar to Architecting a CMS for a content centered website
Similar to Architecting a CMS for a content centered website (20)
Recently uploaded
Recently uploaded (19)
Architecting a CMS for a content centered website
- 1. Architecting a CMS for a content centered website Or ‘you really do learn a lot at these conferences’ Kristin Rowley IA/UX University of Colorado Denver | Anschutz Medical Campus
- 3. The old/current CMS Photo credit: Fibreciment on deviantart
- 5. 1. Switch from http to https 2. Update the site and URL structure 3. Update the architecture and content of sites as they move into the new CMS 4. Add a new search engine 5. Implement a taxonomy for tagging content 6. Break the university website into two campus specific top-level domains 7. Move to new templates (responsive) and themes (standardized) 8. Integrate with authoritative sources for content 9. Create new customer journey sites for prospective and current students on both campuses 10. Improve accessibility 11. Improve site security and performance Project Goals
- 6. 1. The structure of the site should make content easier to find by search engines 2. The structure of the CMS should make content easily sharable and reusable (content types and a taxonomy) 3. Improve the site navigation and the findability of content (site maps and wireframes!) My Goals
- 7. 1. The structure of the site should make content easier to find by search engines 2. The structure of the CMS should make content easily sharable and reusable (content types and a taxonomy) 3. Improve site navigation and the findability of content (site maps and wireframes!) My Goals
- 8. The structure of the site should make content easier to find by search engines
- 10. When asked, almost 87% of prospective students say they use search engines to find university websites, program information and applications. Ruffalo Noel Levitz 2017 E-expectations report 87%
- 11. Once on the university website, high school students and their parents use search engines more than site navigation to find specific information on the site. Ruffalo Noel Levitz 2017 E-expectations report
- 12. People generally scan search engine results in the order in which the results appear and then fixate on the results that rank highest, even when lower-ranked results are more relevant to their search. From Robert Epstein and Ronald E. Robertson’s paper “The search engine manipulation effect (SEME) and its possible impact on the outcomes of elections(2015).
- 13. The inspiration
- 14. Jorge Arango IA Summit 2016 ATLANTA, GEORGIA May 4-8, 2016 Lands, Hubs, and Wienies https://jarango.com/2016/0 5/23/lands-hubs-and- wienies/
- 16. The solution
- 18. The benefits of hub and spoke •Better SEO •Topics are included in the URL •We’re creating content clusters
- 19. The benefits of hub and spoke •Shorter, learnable URLs •/offices/hr •/offices/it •/offices/facilities
- 20. The benefits of hub and spoke •Navigation and content that serve a specific audience •We no longer have to have pages that are everything to everyone •Students can go to /students to find information and postdocs to /postdocs
- 21. The structure of the CMS should make content easily sharable and reusable (content types and a taxonomy)
- 22. The inspiration
- 23. Ren Pope IA Summit 2016 ATLANTA, GEORGIA May 4-8, 2016 Ontology Dojo: Learn How to Use Ontology to Define Your Information And Supercharge Your Deliverables https://blueprintdigital.com/ia- summit-2016/ren-pope/
- 24. • A specification of a conceptualization • (Stanford: http://www-ksl.stanford.edu/kst/what-is-an-ontology.html) • A branch of metaphysics concerned with the nature and relations of being Ontology deals with abstract entities • Merriam-Webster: https://www.merriam-webster.com/dictionary/ontology •[A]n ontology is a formal naming and definition of the types, properties, and interrelationships of the entities that really exist in a particular domain of discourse. • Wikipedia: https://en.wikipedia.org/wiki/Ontology_(information_science)
- 25. The solution
- 26. Step 1: I evaluated the website for content that - could be structured, is used by multiple sites (no one-offs), and needs to be consistent. • Bios/profiles • Academic programs • Events • FAQs • Newsrooms/articles/bylines • Testimonials • Alerts • Tuition & fees • Lists
- 27. Academic program content type Degree program title Program level Degree earned Associated school or college Campus where program is delivered Type of classroom Area of interest Number of credits Description Event content type Title Description Date Time Location (building address) Campus Image Target audience Step 2: Once I had that list, I looked at each type of structured content and broke it out into properties (the individual pieces of content that make up the content type)
- 28. Building Building name Building code Address Campus Person Email address Name Academic degree Title Phone number Address (building) Campus Step 3: Properties that were shared by more than one content type, were broken out into related content types so there would be a single source for this information. Authoritative data Authoritative data Authoritative data Manual
- 29. Building Building name Building code Address Campus Person Email address Name Academic degree Title Address (building) Step 4: I started diagramming all this out as an ontology Event Title Description Date Campus Bio Name Title Short bio Campus Testimonial Name Quote School/college Campus Academic program Name Program level Degree earned Campus Has a Has a May refer to…
- 30. Step 5: I created specification documents for each content type that would help the developers and designer bring these content types to life.
- 32. Site search
- 33. Questions
- 34. Credits • A big thanks to Jorge Arango and Ren Pope for some really great ideas and inspiring sessions. • Images of the university are courtesy of the University of Colorado Denver | Anschutz Medical Campus • Other images (unless otherwise noted) are used under the Creative Commons license from https://pixabay.com/
Editor's Notes
- Let me start off by introducing myself. My name is Kristin Rowley and I work in the IT department at the university of Colorado Denver which has two campuses (one medical, one traditional). I consider myself an IA who also does UX. While I work primarily on the university’s public-facing website, I also work on other IT projects around the university, helping to improve the user experience of web apps and other user-focused applications– this can mean anything from doing UX reviews to creating taxonomies or writing understandable error messages.
- The current university website was set up more than 10 years ago, and while there have been some localized redesigns, the plumbing of the site hasn’t changed in that time. Every time a new web browser, update or security patch is released, everyone in the department holds their breathe waiting to see if something will break. And for discussion at another session (or over drinks) the CMS that is currently running most of the university websites is SharePoint 2010. – that’s the reason we’re having so many issues And while updates to the CMS are dangerous, one of the biggest issues is that due to customizations to the CMS, to make SharePoint workable as a public website, the system has become overly complex and fragile. For example, back in 2015, the web team released responsive templates -but some sites have still not be able to migrated to them because too many of their pages break when they switch, so we have large sections of our website that are not mobile friendly. And while the website may have started out fairly straight forward when it was first built, over the last almost 10 years it has ballooned to over 35K pages, and finding sites or content is difficult or impossible.
- Then, after much begging we convinced senior staff that a new CMS was required, sooner rather than later, and after a new one was selected (by committee of course) we were given about 6 months to set it up before any sites needed to start moving in.
- A set of project goals was determined that range from switching to https, adding a new search engine, to creating new templates and then all the standard things like: taxonomies, improved customer journeys), with the over-arching goals of improved usability, accessibility and security.
- As the only IA/UX person on the team, I helped out a quite a few of the overall goals, but I choose to focus on these three things… I picked these three because: At the university, we have a dispersed website management model, where of the 300+ websites the make up the university website, only a few of the very top level sites are centrally managed. Most are individually managed by departments or schools. Centrally, we can control the top-level site structure, the features and functionality available within the CMS, and the templates and themes, but the content and how pages are layed out are up to individual units. And most of the people managing these unit websites are not web professionals (they may know very little about HTML, or web best practices). So with the new CMS, I really want to build a lot of best practices into the CMS itself.
- And while the third goal is always interesting (who doesn’t lover site maps and wireframes), I’m going to focus on the first two and the assist I got from this conference when planning the solutions
- The first problem I had to solve was how to structure the site so that content was easier to find.
- So first a little background on why I thought this was so important to the university - this chart is from research done last year on what tactics are most affective when trying to improve SEO. And while content, keywords and links are still at the top [CLICK], website structure is #4. So that is something that I could do that will make and appreciable difference to search engine rankings.
- Also, Being findable is very important to the university because we’re a public university up against private and for-profit universities that can afford large ad campaigns (and we usually can’t) and stats like this one emphasis the importance to our overall university recruitment goals, for ranking in search.
- And then there is research like this that shows that students and parents are using search engines not just find sites, but navigate them. So, no more deeply nested subsites, or sites that are org chart driven. The new university website need to be structured in a way that makes finding content easier – our bottom line depends on it.
- And while being found is important, Being found first is even more important From Robert Epstein and Ronald E. Robertson’s paper “The search engine manipulation effect (SEME) and its possible impact on the outcomes of elections(2015). They used eye tracking studies and determined people are most likely to click on one of the top three results, regardless of the relevancy to their search or the credibility of the website. Their conclusion was that: This happens because people trust search engine companies to assign higher ranks to the results best suited to their needs, even though these same people generally have no idea how search engines worked. They assume that google and bing are looking out for their best interests. So when viewing search engine results, if a for-profit university is listed above us, prospective students may consider it more credible or relevant.
- So how was I going to set up a structure that would help CMS users make better, more searchable websites?
- So this will be my first of two paraphrases of previous IA Summit sessions – and hopefully, I’m close to the bullseye, but just incase I’m off – please go online and view it for yourselves. Back in 2016 at the IA Summit in Atlanta, Jorge Arango did a talk about place-making, and creating environments that users can understand/learn and can easily move around in. He used the architecture of Disneyland as an example, speaking about how you want the environments that you create to be able to grow and expand with your users over time.
- This talk really stuck with me. A year later, when thinking about how to architect the new website, I revisited the concepts he presented along with analyzing some of my favorite ‘well organized’ websites, (like the BBC site), [CLICK] and I made the recommendation that we move to a hub and spoke architecture for the university. A hub and spoke architecture is basically what you see here – there is a central hub, surrounded by spokes – I’ll get into the benefits of this structure in a minute.
- So, how did I work out the details…
- To come up with a model, I: Looked at all our current websites and broke them into groups. I found there were five general types of sites: School/college sites Campus specific sites Audience specific sites (for groups that don’t identify with a specific campus or college) Verticals (research, healthcare..) ‘Collection of’ sites (admin offices, centers, clinics, we’re even planning a policy spoke) I took Jorge’s presentation quite literally in the initial version of the diagram (see here), I overlaid my recommendations for the university website on top of a map of Disneyland to help make the concept more easily understood by stakeholders, like deans and chancellors who are not web professionals and don’t know anything about sitemaps
- The benefits of moving to the hub and spoke model include: Better SEO (of course). And some of the reasons for this are that one – each spoke uses keywords to identify the ‘type’ of site in the URL For example – On the current website, we have information about research at the university in (at least) three different subsites, with three different site goals, diluting our ability to rank high for research terms (we’re competing against ourselves in a way), and we’re creating confusion about where in-bound links should point. with hub and spoke, all research related sites will all be located in a single spoke, their URL will contain the keyword ‘research’ that helps describe the content – but because we’re a university, we have sites named things like: CCTSI or COMIRB – designating them as research focused sites (using the word ‘research’), will help make them more findable Also, grouping them together allows for them to combine their SEO juice. This is the concept of ‘content clustering’ to get more SEO weight. Content clustering works because Instead of searching for keywords, people have started searching for questions or concepts, thanks in part to voice search and Siri and ALexa. Creating content clusters (also referred to as ‘pillars’) has proven to increase search results because it provides a better semantic understanding for search engines of the ‘aboutness’ of their content when taken as a whole. (Hubspot and Moz.com has been doing a lot of research around this concept).
- Shorter, learnable URLs For example, all administrative offices will be under /offices – need to find HR- easy, it’s at /offices/hr. Need to find the Office of the Chancellor - /offices/chancellor. And because sites are no longer deeply nested, only 1 or 2 levels down at the most, shorter URLs are also an SEO win
- The third benefit of moving to hub and spoke is that; Navigation and content serves a specific audience – it no longer has to serve all audiences We can now have website sections that focus on specific types of users and their tasks When talking about this point – I always call it the ‘multiple front doors’ benefit. In our current website, there is one front door – the top level homepage (it’s our primary landing page with 17% of all entrances). Most users come through this single entry point at some time, and then have to find their way to the content they really want. Historically, this has meant that the home page needs to address all audiences, throw in the politics of homepage content and this page has had everything to everyone (visit a university website and notice how much navigation they have on their homepage) Primary navigation, audience based navigation, quick links, in-page navigation, fat footers While not all universities do this, quite a few still do Moving to this new architecture allows us to create multiple, topic or audience focused front doors that will take the pressure off the homepage. And hopefully over time, search engines will start to recognize these new doors as well, sending users to those landing pages instead. For these front-door spoke sites: Their navigation is specific to their primary audience, their content answers those needs and helps that audience complete their tasks – it’s focused and simplified (hopefully). And, Already hub and spoke has been paying off in other ways, it has made building out sites in the new CMS a lot easier. Sites can move from the old CMS to a spoke in the new CMS with fewer dependencies. For example, because centers and clinics now have their own spoke, they can move over any time without having to wait on the school/college site to move first. And, when the university has a new initiative that it wants to promote, something like ‘healthcare’ , this can become it’s own spoke instead of having to figure out where to ‘fit it in’.
- The second goal I wanted to work on was around sharable and reusable content. Currently, most of our webpages are being built as one-offs, with little or no shared content (mostly because it’s very difficult in the current CMS to do). So the CMS itself is making it difficult for our websites to rank because of three things: Duplicate content (people copy/paste if they can’t find shared content) Confusing or inaccurate content – because content gets out of date or isn’t being pulled from authoritative sources Hard to reuse content (pulling in a tuition & fees table can overwhelm a page, so content managers aren’t doing it – they link to a tuition table on a different site, dropping the user into unknown territory)
- We new all along that we were going to need to set up content types and a taxonomy in the new CMS, but where to start…
- Again at the IA Summit in Atlanta, Ren Pope did a session on ontologies and basically told us not to be scared of using them to diagram relationships. He was very convincing. At this time, the university was in the process of selecting the new CMS, so it was in the back of my mind – and I had a ‘light bulb’ moment in Ren’s session. Creating an ontology was the perfect way to work out what we needed to do. For those that may have missed his talk, you can watch it online, but in the meantime, just to catch everybody up
- While there are multiple meanings – for IA’s - an ontology is basically a way to model a system or environment using objects that have properties, and their relationships. Of these definitions, the Wikipedia one probably makes the most sense, but I do love the top one from Stanford – I think I’m going to start using that phrase as my elevator pitch when someone asks what an IA does – we create specifications of conceptualizations.
- So how do you create a set of content types that allow CMS users to easily share, reuse and repurpose content when you’re starting from scratch?
- First I looked at our current websites for content that: - could be structured - was used by multiple sites (no one-offs) - but I also considered content that needed to be consistent everywhere it appeared, even if it was only in a few places The list of content types that I came up with include: Bios/profiles Academic programs Events FAQs Newsrooms/articles/bylines Testimonials Alerts Out-of-the-box content types were: Lists Blogs That we also used. And since this initial list, there have been additional types added – but this is the list that got us going.
- This was initially a big excel spreadsheet with a tab for each content type and rows for each property(field). I added additional information around the fields where necessary – like example text, to help everyone on the team understand what was included. Once I had a list of fields– I ran them by a stakeholder group for feedback to get the final list of content types and properties.
- Next I did analysis. First I looked for blocks of content (meaning multiple fields) that were used by multiple content types that could be pulled out or refactored and made into their own related or child content types. [CLICK] For example, I noticed that: Address (which included building, street address, city, state, zip) is used by: department, event, and person was pulled out on it’s own and used as a related content type so that there was a single source for that information that could be easily updated. And by doing this, it could also be used to create a directory of university buildings using the new address content type Person information is used by: bio and byline and on its own could be used to create a directory of people Second, [CLICK] I looked for single fields (not groups of fields) that were used across multiple content types to see if they could become part of the CMS’s tagging system. This would allow these values to have a single source and use a controlled vocabulary Common fields that needed a controlled vocabulary included: Campus School/college Role/audience Department Academic program And finally – I looked at all fields and added an attribute for if the value for that field should come from an authoritative source or if it would be manually entered. This allowed us to start working with the data team on getting that information imported into the CMS.
- Now I had objects and properties, I just needed to map out their relationships. [CLICK] While this can get a bit complicated, it was very helpful in determining if I need to do more refactoring for related content types, and it allowed me to evaluate relationships that weren’t definite (may refer to, may teach…) as places where I needed to make sure that the tagging functionality could tie those content types together. For example a testimonial may refer to a program. These are not related or parent/child content types (because they have no fixed relationship), but they need to be able to be associate together, so that was done by making sure that testimonial used the ‘academic program’ name tag. This tag wasn’t part of the initial content inventory, but was added to create this association After diagramming the content types as an ontology, I realized that some tags should go on most, if not all content types (tags like ‘campus’ and role’). This way, these tags could be used to create focused pages based on their value. For example if I created a Faculty page – I can display faculty focused content from the newsrooms, the events feed, or bios Having global tags has paid off big time, when we recently got the directive that we now have two top level domains, one for each campus, and we were going to be breaking our site into two– tagging insured that content could still live centrally or be managed by a single office, but on the campus sites, only the relevant content would show. The ontology not only helped define content types, but it also helped to create the taxonomy that tied it all together. Had there not be the ontology diagram, I’m not sure I would have been as thorough in the evaluation of our content.
- AS the final step, I needed to create a document that the developers could use to build these content types and the ontology was overwhelming. [CLICK] for this, I created a specification document. It took all the information about a single content type and recorded it in a single document. Additional information was added around presentation of the content using wireframes. This document also made it easy to vet the content types with stakeholders without confusing them with an ontology
- At the end, this has helped us to solve the three issues: [CLICK] Frequently used content was standardized into content types that could be displayed on multiple pages, but with a single source for the content (no more duplicate content) [CLICK] Important, high value content like tuition could be pulled in from an authoritative source, keeping it up to date (no inaccurate content) and now there were multiple views of the content so only relevant information needed to be shown (not everything) [CLICK] Using tagging our CMS users could easily relate the content types together creating more focused pages (no more hard to reuse content).
- Add as an added bonus - we were also able to pull the content types and tagging into our site search as facets, to help users filter their search results.
- ------- Thanks you all for taking the time to be here, especially on a Sunday afternoon, it is much appreciated. The information and ideas that are shared at these summits is really great stuff that we can all use to solve real-world problems. And I would like to thank Jorge and Ren, and everyone else who has presented over the years for sharing their ideas, this event really does help to move our industry forward.