MailChimp is good at iterating, and our Knowledge Base team has taken that same iterative approach to educational content and structure. We’re good at reacting to change, but we also want to plan ahead, anticipating change before it happens.
To guide us toward a better, more consistent product, the Knowledge Base is developing new procedures and frameworks to facilitate interdepartmental communication, on-going data analysis, and the like. Here’s a look into what we’re working on, and how we’re planning for the future.
Soliciting and Organizing Feedback
The ultimate goals of our Knowledge Base are to support our customers and reduce the number of tickets in our support queue. To meet these goals, we reach out to users and talk to our customer Support agents to understand what users struggle with, what processes don’t make sense, and how Knowledge Base content can help.
Improving Existing Content
We include a short user feedback survey on every Knowledge Base article to get a snapshot of which articles are the most successful. We also get constant feedback from all over the company—Support, Research, Marketing, Partnerships, Data Science, etc.—about how to improve article language, linking, or searchability.
This feedback is partially automated and tagged through homegrown web applications, but a large portion is also analyzed by actual humans. These Knowledge Base team members are liaisons to other departments and prioritize issues through a single project management funnel—a huge help for keeping large amounts of feedback in order.
We also use a lot of spreadsheets. We’ll run exports from our CMS (Hippo) and Google Analytics to pull article titles, categories, last modified dates, and web traffic. Trends in traffic and topics across the entire Knowledge Base help us identify which articles need simple updates, a metadata overhaul to improve searchability, or other structural changes. Then, we can organize our sheets in all sorts of helpful ways. We can sort pages for highest traffic, cross-reference for older last modified dates, make notes or questions for teammates, assign or prioritize with color coding—and on and on.
When feedback doesn’t come with the institutional knowledge of a team liaison, the CMS and analytics data help us prioritize where and when to tackle content projects. When we’re ready to make changes, we move relevant pieces of feedback to Asana, our project management application, and assign tickets to our writers based on their areas of responsibility.
Creating New Content
Leading up to releases of new or updated MailChimp features, the Knowledge Base’s editorial process changes a bit, so we can make sure the appropriate instructional content is published simultaneously. Mini-teams—comprising an engineer, designer, QA tester, and writer—work together to ensure we document new features effectively. We use a smattering of tools, including Pivotal Tracker, Basecamp, and InVision app, to monitor conversations about feature updates and to share drafts.
Aside from being a relatively simple way to manage assignments, these tools keep multiple departments in the know, reducing bottlenecks and surprises. Combined with defined focus areas and interdepartmental teams, we’re able to stay connected and on top of current projects during what can be a busy, confusing time.
Our articles must be consistent and easy for users to follow—a standard which becomes harder to maintain with each new writer we hire. Our growing team needs meta-documentation about how we work to help onboard new writers and provide guidelines for procedures and consistency. So, we created a handbook for our department that includes policies, procedures, focus areas, editing and revision tips, style and formatting guidelines—and a heavily used Word Bank.
To build the Word Bank, writers discuss commonly used terms and phrases on the fly and agree on the shortest, clearest way they can be presented to users. When we aren’t sure about terminology or a best practice, we hash it out in our biweekly Writers’ Meeting, and update the Knowledge Base handbook.
The Word Bank and handbook are always growing, along with MailChimp. We’re training ourselves to keep it available for constant reference to encourage consistency every day.
Cataloging and Updating Content
With tactics like interdepartmental teams, spreadsheets, and handbooks in place to support the everyday management of content, we can actually get down to the nitty gritty of content creation and improvement.
Using a combination of exports from our CMS, statistics from Google Analytics, and the feedback we get from all over the company, our writers can prioritize and review articles for accuracy and clarity. Sometimes we have to combine, restructure, or even remove articles to better organize content for users. If we find a concerning trend in our docs, these tasks might blossom into full blown audits.
The Next Steps
Last year, as part of a CMS migration from Expression Engine to Hippo, we templated four article types—Reference, Tutorial, Troubleshooting, and Quick Answer—to build content forms for the new system. But we provide a lot more types of content than that. Every time we shoehorn content into a form or framework that isn’t quite right, we fall off the wagon in terms of consistency and efficiency.
To better keep us on track, the next task for our handbook is to re-evaluate our content types and create solid frameworks for each of them. We’ve been drafting ideal, sample articles for each article type we use and breaking them into components that writers can use as a starting point. Ideally, these formulas will push us to categorize and structure content the right way the first time.
We’re also working on a persona project to define the types of users who interact with the Knowledge Base most. We’ll rally another interdepartmental team to gather data about these users, and crunch some numbers down into digestible action items we can use to improve the way we approach our content.
Of course, this approach to shaping and organizing content isn’t one size fits all—you’ll have to investigate each component and tailor the process to your team and needs. Since we iterate all the time, these current projects and processes will see their share of changes. But they’re working for us for now, and they’re helping us act instead of react.