Search is often the first thing people use when they land on our documentation site. If it doesn’t work well, trust erodes quickly. That’s why we’ve been rethinking search — we’re continuously improving search (and navigation) to create experiences that reflect what users actually intend to find.
Understand user intent
We started by looking closely at how people search on our documentation site. Do they type in product names, ask full questions, or use error codes? Do they look for “how-to” guidance or quick reference details? These patterns reveal not just what users are asking, but why they’re asking.
By mapping queries to intents — like setting up a product, troubleshooting an issue, understanding a feature, or comparing options — we can see where content already serves users well and where gaps exist.
Map and restructure content
Once we know the intent, we map and reorganize content so each intent has a clear, trusted answer. We also fill gaps by creating new content where needed and removing duplicates or near-duplicates so users don’t waste time scrolling through competing answers.
This restructuring is an ongoing process. Given the volume of content, it will take time, but every step makes search results cleaner and easier to navigate.
Enhance search
We’ve also updated the algorithm itself. Search learns from user behavior: when many people with the same query find a page helpful, that page moves up in the results for the next person — improving relevancy over time.
We added suggested searches and pages to guide users toward the right content.
Measure what matters
Improving search isn’t just about making changes — it’s about knowing whether those changes actually help. That’s why we measure how well search performs using metrics like relevancy (did the results match what the user wanted?) and precision (how many of the top results were actually useful?).
These signals give us a feedback loop: we test, measure, learn, and adjust.
Design to help you decide
Even the best search results fall short if they’re hard to scan or navigate. That’s why we redesigned the search results page so it’s easier to scan, added filters, and gave people better ways to zero in on what they need.
What’s next
The next leap comes from AI. We’re testing AI models that help us go beyond keywords to understand the meaning behind a query. These aren’t live yet, but they show real promise for surfacing more accurate, intent-driven responses in the future.
Improving search isn’t a one-time project — it’s a journey. By aligning content, algorithms, and design, we’re working hard to make it easier to find what matters to our users. We’re excited about what comes next.