Information architecture: How to structure, store, connect data

From DevSummit
Jump to: navigation, search

Information Architecture Space

Icebreaker: Name + Are you struggling with organizing data?

  • Wordpress development for social good
  • Salt stack. Getting everyone to adopt the same plan. Understand and buy into the plan.
  • Conflicting ways of passing around data
  • Design Action Collective
  • Reinforce Action Network. Who are you writing for and why when you don’t have control of who writes everything.
  • Alternative Technologies - Berkeley
  • Anthropology PhD Student at UC Irvine building organizational archives for 3 groups in Nairobi on Drupal site (
  • Kenyon consulting

Defining Information Architecture Coming up with ways of structuring data that make it findable. Improves user experience by making content discoverable, findable.

Info Architecture disasters: New site that had been operating for 10 years as Joomla site then Drupal then WordPress. No Info Architecture pass. Lots of years of organic growth in their systems. Single use taxonomy. Not effective to have geography, theme, author, style of writing, etc. These have to be broken up into different kinds of things. What taxonomies would we use. Take all of the place names out of tags and put into a geography taxonomy. They were listing the kind of article – op ed, news analysis, etc. Content type taxonomy. This is important for SEO. Google will lower your ranking on SEO if you mix them. Google keeps ramping that up.

Taxonomy – any way of labeling.

Heirarchical structures. In the age of blogging we used to organize content in hierarchical or linear fashion. In chapters in a book. Folders in filing cabinet, date by date. No topical relationships. Then focus on tags and categories to horizontally organize within that flow. All info architectures are a bunch of decisions between do you arrange hierarchically and which one do you organize in which way. Date focused, tag focused data, categories.

Meta data – not content but we are saying something about the content. In a relational database, to relate similar content into groups.

You have to ask yourself what the use case is for all of your data. Who uses this data? How do they use this data? Who will search for things? Search interface. Presentable to your users. Dropdown that says I want this country in this year in this way. You have to put yourself into user’s shoes. How does user interact/find the data? Term has to be useful. Taxonomy and term. 1200 articles tagged Hilary Clinton; 2 with Hilary Clinton’s email issues. Not effective if you create a folder that is too narrow. The ghost town effect. If you have places with no content in it; looks like website is not very busy or effective. Reuse a structure that people are familiar with. Reusing a metaphor that people who are not technologists. Makes it easier for them to accept the system. Finding an abstraction that everyone can agree on. These are my “inventory items” these are the transactions. They will just continue to keep growing, etc. You have to agree on the models of how things are held together. Don’t need to much granularity. 72  6 categories. When you break things down into taxonomies. Set some of the to be mandatory. Hard-wired it into the system. They can’t opt out. One way to break these giant collections of mismatched terms into mandatory terms. How to build a shared vocabulary – ontology – what is the process to get everyone on board to share the same understanding of what the topics are: Framed as “what Google SEO needs” Data definition document – this is when you should choose article vs op ed. Sometimes we get together with the team and create the info architecture. Everyone agrees. But then there is not enough documentation captured to the meeting. What often happens, passing of time, they forget, you forget or new person/team comes in and is in charge of entering the content so all that work gets lost. Every user can add tags to anything. So if someone writes something that they have a trigger for. They can tag that. Huge indexing system. There is so much metadata. Can be a good place to put your chaotic thing and then see what patterns emerge out of that. Google analytics to see what people were searching for on the site. Most were trying to find an author’s name. So the navigation was failing the users. They have to go to search. Better to have a side bar. Problem: getting too granular – there is a sweet spot Something too broad like “politics” vs too specific – “hilary clinton’s emails” Writing different lists of criteria: Is it too general or too vague? Does it create an archive that is interesting and fresh? New tag that you know will be important later? (e.g. Ferguson) Is it a tag that Google understands? Related ones – 30 different related terms. Go through, collect and merge. Google Drive – way of getting things out and making it searchable. Make a main page and list out content – Make a hand-maintained index. Google Drive has bad search. Make a wiki page. Different tools for wikis. You can have links to each other. Negatives of hierarchical structure – cannot see everything inside of everything. Wiki will auto generate table of contents for your drive. Airtable – website. Cool UX in front of a database. Can put links and tag and play with. How do I use my information – what am I struggling to find. Does the date matter more than X? Put all details in folder on the book name, or if author matters more? Geography can be a good hierarchical category – can nest cities in state and state in country. Allows user to drill down. Web dev – is this a taxonomy or a content type? Is op ed - content or taxonomy? Easier to have content type that uses a different template. Content type if it has different display. Content type becomes another category. Show me all blog posts that are associated with this topic vs all content associated with this topic. E.g. article, video, authors (author bio), art, Special things about how I want to display it. Most modern you can create custom fields – staff profile might include their position, etc. Content type vs taxonomy. Content type has different fields. Author content type might have their twitter handle. Categories and tags – something should only be in one category. Tags are way more amorphous. Category are radio buttons – there can be only one. Tags can be as many as you want.

Discovery process: Facilitated process. Tease things out and impose strategy. Ask people – what is the goal of your website. Why? Who comes here? What do they want? Do they have any audience data? Google analytics? What do people search for when they get there? Try to identify fake audiences – service providers, audience is our clients, etc. You might be serving people on the street but those are not the ones going to your website, maybe it is their family members. What is our goal, who is our audience, in the case of each audience, what are they looking for? You can get someone to help you get consensus. Story-based strategy. Sometimes hard to be asked this. Bunch of exercises and tools that helps them identify their goals and their users. Get outside the box and start dreaming about alternatives. Professional services. It can happen in a day. RESOURCE: Center for story-based strategy. Orgs have a tendency to organize their public facing data the way they organize it for themselves. E.g. climate department and forest department. All internal facing is like that. Bunch of content that didn’t match the categories. Changed info architecture to reflect content we have. Hopefully write content that makes sense for audience and stops reflecting “us.” Don’t use acronyms as tags, or name of campaign. Unless you know what the out of fashion does, you have no idea that is what you are looking for. If you have “deforestation” as your term, then people can go look for it. Google tag manager – shows you clicks. Taking important real estate.

Advice from S on tags: Will this tag make a clear sign post. Don’t want more than 6 options.

Discovery Plan: What are your organizations goals? What are your website goals? Sort the content in ways that dovetails with the goals and startegies.

Best Practices: Make some categories mandatory for the user to tag. Document how the terms are agreed on and what goes where and make sure that the documentation is shared with all stakeholders TO DO: Scott to share the guidelines document with the group. RESOURCE: Center for story-based strategy can help! Resource: Google tag manager – shows you clicks. Taking important real estate.