Why has it become so difficult to find timely documentation on elastic.co?

Meta Elastic
1

Hello Elastic team and community,

I want to open this thread because I'm honestly frustrated (and I think I'm not alone) by how complex it has become to access specific documentation for key Elasticsearch features, such as ILM, Reindex, or the APIs for use in Devtools.

Since the implementation of the ESRE-powered search engine, the search experience on elastic.co is no longer clear and straightforward. For example, when searching for reindex, the first results mix Curator documentation (already obsolete in many cases), blog posts, examples for specific clients (such as JavaScript), and even documentation for serverless features that aren't relevant if I'm running an on-premise cluster. All of this appears without an effective way to filter content by version, product, or type (such as REST APIs, YAML examples, etc.).

More than 300 results, the vast majority of which generate more noise than value. And don't even think about searching for something like ILM retry failed steps or how to manually move ILM to next phase. The information exists, but it's buried amid scattered or duplicate references.

I understand the value of having a modern semantic search engine, but don't you think it should prioritize technical accuracy and context over "related content"? Where's the focus on clear, version-navigable technical documentation like there used to be?

I suggest (or beg):

  • A classic search mode by version and product tree (like the old documentation).
  • More useful filters: by Elastic version, by content type (API, Beats module, etc.), and not just "documentation" or "blog".
  • The ability to ignore serverless content, if my stack doesn't use it.
  • Or at least a clear link from the main bar to the classic document tree.

My experience as a certification candidate

I'm currently training for the Elastic certification exam, and if this same search experience is transferred to the exam environment (where documentation is allowed), it's really frustrating.
Having to filter through irrelevant results when what I need is direct access to an API definition or valid command parameters is a critical waste of time.

This directly impacts the quality of the study and creates insecurity even in experienced users.

Elastic has amazing products, but finding the right information is becoming unnecessarily complex.

Does anyone else feel the same? Are there any tricks you're using to avoid falling into this result overload?

Thanks for reading.

3 Likes
2

Thank you @iTiago for sharing your experiences and frustrations with the documentation. Is this off the back of the recent documentation changes?

The reason I ask is that there is an ongoing thread here on new documentation issues that this feedback could be relevant to. Do let us know if this is the case as, if so, connecting your feedback here on that thread would be useful.

Hope that helps!

2 Likes
3

As Carly said post/follow on the mentioned thread. Until some thing don't change in the documentation, the old documentation is moooostly under Reference.

1 Like
4

@iTiago Thank you for taking the time to share your thoughtful and detailed feedback. I want to start by acknowledging your frustration. It’s absolutely valid and we’ve received similar sentiments from others as well.

We recently launched a new information architecture for Elastic Docs that organizes content around user goals rather than individual products. Our aim was to make it easier for users to accomplish tasks without needing to understand Elastic’s internal product boundaries. However, we recognize that this change impacted users who rely on product-specific navigation and search patterns, particularly those who are deep in the ecosystem, studying for certifications, managing specific clusters, and more.

To directly address the issues you've raised:

  • We’ve recently added product-specific metadata to each docs page in the new information architecture to improve search relevance.
  • We’ve also worked with the website team to implement product-based filters in the global search experience. This gives users a way to narrow down search results based on specific products like Elasticsearch, Kibana, Beats, and more.
  • Version filtering is an area we are actively exploring next. We agree it’s critical, especially for users managing long-lived clusters or studying for certification exams.

We’re continuing to refine the experience, and your examples, such as trying to search for “ILM retry failed steps”, are incredibly helpful. We’re using feedback like this to guide improvements to the search experience.

Your idea of a "classic" navigation or search mode is also compelling. We’re discussing ways to bring back more of that tree-based clarity while preserving the benefits of a goal-oriented structure.

Thanks again for helping us make the docs better. We’re listening and actively working on it.

3 Likes
5

Reminds me of:

(not to make light of the issue, but one has to laugh!!)

3 Likes