kref.altervista.org, reimagined.

A first-time visitor sees only a site title and a wall of links organised into categories. There is no sentence explaining who this is for, what level of Korean learner it suits, or why this resource is better than Duolingo or Talk To Me In Korean. The value proposition is entirely absent.

A beginner and an advanced learner land on the same undifferentiated page. There is no suggested starting point ('New to Korean? Start here'), which likely causes bounce as visitors feel overwhelmed or unsure the site matches their level.

'Korean Language Reference' is descriptive but forgettable. There is no logo, tagline, or indication of the depth of content (grammar, numbers, dates, dialogue, 24 flashcard sets). The breadth of what exists is invisible at first glance.

A learner who finds the conjugation or honorifics pages has demonstrated clear intent. Without a newsletter signup or even a simple 'notify me of updates' form, there is no mechanism to bring them back. The site has been running since 2014 and has built zero owned audience infrastructure.

Offering a zip download for offline use solves a real problem (slow hosting), but the solution destroys any chance of returning visits, ad impressions, or future upsells. A progressive web app or a paid offline access option would solve the same problem while keeping the relationship.

Language learning is a proven monetisation category — premium flashcard decks, a PDF grammar guide, a donation button, affiliate links to Korean textbooks, or even display advertising would generate revenue from existing traffic without changing the content at all.

Obfuscating the email address to avoid spam is understandable, but it also prevents anyone from clicking to contact. A simple contact form would lower friction and open the door to feedback, partnership, or commission enquiries.

The site already has a TTS page and audio in the dialogue section. An AI conversation partner — where a learner types or speaks a Korean sentence and receives a corrected response — would be a natural and highly differentiated feature in this niche, and is now achievable at low cost via API.

The current flashcards appear to be static HTML pages. A spaced-repetition algorithm (like Anki's SM-2) would dramatically improve learning outcomes and increase return visits, as the system tells users when to come back. This could be built as a lightweight web app without migrating all existing content.

These tools show the site already has some dynamic functionality. An AI-assisted grammar checker — paste a Korean sentence, get an explanation — would serve learners at an intermediate level and is the kind of feature that drives sharing and backlinks.

The changelog shows ongoing maintenance: updated avatars, voices, menus, spelling corrections. A static site generator or a simple CMS would reduce the time cost of updates and make it easier to add new content sections without editing raw HTML.

The copyright line says 'Korean Language Reference All Rights Reserved' but reveals nothing about who made this or why they are qualified. A single sentence — 'built by a New Zealand-based Korean language learner with 10 years of study' or similar — would immediately build credibility with a sceptical new visitor.

The version number (2026-07-23) and changelog show genuine, long-term commitment to accuracy. This is a significant trust signal that is buried in a modal. Surfacing '12 years of continuous updates' on the homepage would reassure learners that the content is current and reliable.

There is no indication of how many people use this resource or what they think of it. Even a simple 'used by learners in 40 countries' stat or a handful of emailed thank-you quotes would provide the social proof that distinguishes a trusted resource from an amateur page.

The site itself acknowledges slow server performance. Page speed is a confirmed Google ranking factor and a primary driver of bounce rate. Migrating to a low-cost static hosting service (Netlify, Cloudflare Pages, or GitHub Pages) would be free or near-free and would immediately resolve this without changing any content.

The homepage lists every page on the site in a single long navigation block. Without grouping, visual weight differences, or a recommended learning path, learners have no signal about where to begin. This particularly hurts mobile users who must scroll extensively to find a specific section.

The homepage is a flat list of anchor links with no grid, card layout, or collapsible sections. On a phone screen this becomes an extremely long scroll with no visual relief, likely resulting in high mobile bounce rates in a demographic that primarily learns on mobile.