Taxonomies, filters and search after moving to WordPress
An IT expert explains how to rebuild categories, filters and site search in WordPress after converting an AI-built website.
Convert2WP – convert your AI website to WordPressClick here
From components to structure
AI-built sites often display lists with filters that look dynamic but are really hardcoded arrays in a React component. The design shows tags and categories, yet there is no real data model behind them. When the site becomes WordPress, those lists turn into static HTML. They look right, but they cannot grow.
The real value of WordPress appears when you rebuild that structure properly: content types, taxonomies and a search that understands them. That is the step that turns a converted design into a site your team can expand for years.
Choosing between categories, tags and custom taxonomies
WordPress offers categories and tags out of the box. For a simple blog that is enough. For anything else I usually create custom taxonomies, for example industry, product type, location or difficulty level. Each one answers a question a visitor actually asks when looking for content.
I keep the number of taxonomies small and the terms within them deliberate. Ten taxonomies with three terms each confuse editors. Two or three well-thought-out taxonomies with clear rules make content easy to classify and easy to filter.
Custom post types where they make sense
If the old site had sections like case studies, team members or products, these become custom post types. Each gets its own fields, templates and archive. Fields can be added with a fields plugin or registered in code. I prefer code for structure and a plugin interface for editors.
The converter gives you the visual templates. Wiring those templates to real post types and fields is a manual step, and it is worth doing early, before editors start adding content in the wrong place.
Building filters that do not slow the site
Filtering on taxonomies is fast because WordPress indexes term relationships well. Filtering on metadata, like price or date ranges stored as post meta, is much slower at scale. When a site needs heavy filtering, I either move values into taxonomies or use a dedicated filter plugin that builds its own index table.
I also make sure filter results have clean, crawlable URLs only where they add value. Every possible filter combination indexed by search engines creates thousands of thin pages. I allow indexing for a few meaningful combinations and add noindex to the rest.
Why the built-in search is not enough
Default WordPress search looks for words in titles and content and sorts by date. It ignores custom fields, taxonomies and relevance. For a small site that may be acceptable. For a knowledge base, a shop or a multilingual site it quickly frustrates users.
A relevance-based search plugin that builds its own index is the first upgrade. It can weight titles higher than body text, include fields and taxonomies and handle partial words. For very large sites I connect an external engine such as Elasticsearch or a hosted search service.
Search in multiple languages
Multilingual sites need search that stays within the language of the visitor. A Dutch visitor should not get English results first. With most multilingual plugins this works when the search plugin is configured to respect the language, but I always test it explicitly in every language.
Non-Latin scripts deserve special attention. Languages without spaces between words, such as Thai or Chinese, need proper tokenization. External search engines handle this much better than database-based search, which is one reason I recommend them for sites in many languages.
Measuring what visitors look for
Once search works, it becomes a source of insight. I log search terms and especially searches that return no results. Those zero-result queries tell you exactly which content is missing or which words your visitors use that your pages do not.
Reviewing that list once a month is one of the cheapest content strategies I know. It often reveals simple fixes: a synonym to add, a page title to adjust or a short article to write.
Keeping the structure healthy
Taxonomies drift over time. Editors create near-duplicate terms, use tags inconsistently and forget to assign categories. I write a short classification guide and review terms every quarter, merging duplicates with redirects so no URL is lost.
The converted design gives you the starting point; the structure behind it is what you build on top. Small mismatches between the old filters and the new taxonomy are normal at first and can be corrected step by step.