This page contains the details of a technical writing project accepted for Season of Docs.
- Open source organization:
- Technical writer:
- Project name:
- SEO / Site Analytics & Docs Site Updates
- Project length:
- Standard length (3 months)
To improve search engine visibility, understand user behaviour, and drive future content improvements, I propose a bottom-up optimization strategy for DVC.
In terms of search engine optimization, “bottom-up” means using data from current search results and existing content to direct updates and start a positive feedback loop of improvement. This strategy focusses on results to build upon actual progress instead of relying on assumptions about what users will search for, or how they will use it. I’ve used this approach effectively for multiple SEO clients and it’s known to be effective for current search engine behaviours.
The goal of the process is to develop a feedback loop as follows:
- What pages and search terms get search results?
- What are related to these terms? Are we answering the questions searchers have? Whats missing in the doc?
- Update the existing doc or identify new documents that should be created (if it makes more sense).
- In areas where the organization wants to get results (but has none), look for evidence of competitor searches or user analytics before building content.
- Start again at 1.
I propose the following high-level project plan (with further details on implementation in the Q&A below):
Week 1 — Initial set up for Analytics tools and tracking. Run SEO audit and create issues to fix metadata or clear technical issues. (This could even begin in the warm-up period). Week 2 — Identify docs that are already ranking for key terms. Identify related terms to expand content around and audit the docs for other improvements. Create issues on individual document scale to plan updates. Start updating/publishing the docs. Week 3 — Continue monitoring search results to identify new opportunities and continue working through the planned update backlog. Week 4-10 — Observe changes to search results for newly updated docs, and continue monitoring and updating the backlog. Week 10+ — While it’s definitely out of scope for this project, once there is a comfort level with the rate of change and methods, the same principles and feedback loop could be used to drive revisions to the DVC use cases and docs home page. In my opinion, a bottom-up approach is more likely to be effective for those projects as well.
Here are my direct answers to each question listed in the project idea:
Q. What tools should we employ? (e.g. Google Analytics, etc.)
A. The essential tools are Google Analytics, Google Search Console, and Google Data Studio (to aggregate data between tools for reports). Google Tag Manager is useful for tracking some specific click or page events (for example, YouTube embedded video tutorials). I’d also use an SEO audit tool (I use Ubersuggest) to flag issues and track competitive and related search terms across the docs site. While the DVC site seems quite fast, we’d need to make sure using PageSpeed insights as it’s also critical for SEO.
Q. What trends and reports do we need to focus on?
A. The key SEO metrics are clicks, impressions, click-through rate, and position. However, the challenge is that these are trailing indicators and don’t provide much insight on what to do to improve. For that purpose it’s critical to track and monitor what happens before and after searches: the search terms people use, and what happens when they visit the site. The search terms in use are essential to directing productive work on content creation and updates (as described above). Ensuring that users who arrive via search results successfully find what they are looking for makes the difference between ranking well or not at all, as returning to the same search (a bounce) tells the engine that the page what not a good result. Measuring user engagement on the site is a more complicated task, but the essential metrics we’d need for documentation are bounce rate, session duration, and pages/session. (For sites where the goal is to have user’s acquire/buy/contact, conversion rate against the goal is also a key metric.)
Q. What kinds of users do we have and what interaction flows do they each follow?
A. If set up to do so, Google Analytics will track a user’s path through the site, page timing, click URLs, user agent properties, and attempt to identify them on returning visits (there’s more, but these are the basics). It will take time to identify and understand patterns that define user types, but looking at popular interaction flows are a good place to start. Starting from the most popular landing pages, we’d look for obvious trends in second, third, and further pages. Then we could propose user models against these (which should also help inform the key use cases). Working from there, we can further refine or validate the models/use cases by correlating other data: search terms, anecdotes, surveys, interviews, etc.
Q. Can we semi-identify these users and/or cross-examine their data with DVC usage analytics?
A. Based on the usage analytics documentation, DVC is using a truly random identifier (uuid4) and sending data through a proxy. Assuming this won’t be changed, cross-examination would be limited to viewing the volume trend for each command event against the usage patterns for the docs site. This would help us identify discrepancies between docs usage and command usage in the aggregate, but won’t provide user-level insights. So we could probably answer the question “for which commands/docs are people using the doc and command at the same time, or not?” While that is basic, it would provide basic validation for assumptions (e.g., if a particular use case is well matched it should lead to increased usage on the key commands) and identify opportunities (e.g., if a command is not being used, but the docs for it are (or vice versa) what’s wrong?)