Knowledge platform that cut content discovery time by 45% and extended sessions by 30%

The situation

A publisher with a catalogue of digital books and educational material was distributing content through Google Drive folder links shared via WhatsApp and email. There was no discovery mechanism — if you didn't know the title, you couldn't find it. New publications were announced in a group chat and quickly buried. There was no way to track which content was being read, what was popular, or whether users were actually engaging with material once they accessed it.

Authors uploading to the platform had no self-service option — they emailed files to an admin who manually organised them into folders. When a reader on a mobile device opened a PDF, their browser would often download it rather than displaying it inline, disrupting the reading experience entirely.

The publisher had considered a generic document-hosting SaaS tool, but none of the options let them keep their own branding, control access per title, or see any usage data — the trade-off was giving up ownership of the reading experience entirely. They needed something that behaved like a real content platform, not a file drop, while still being simple enough for a small editorial team to run day to day.

What we built

Structured content library

A catalogue with categories, tags, author profiles, and publication metadata. Each title has a landing page with description, author info, table of contents, and a reading preview. Content is organised in a taxonomy the publisher controls through an admin panel, with the ability to feature titles on the homepage and in category listings.

Elasticsearch-powered search and filtering

Full-text search across title, description, author, and table of contents. Filters by category, topic, language, and reading level. Search results return in under 200ms. Search terms are logged (anonymised) to inform content acquisition decisions — the publisher can see what users search for but don't find.

In-browser reader

A PDF.js-based reader that renders documents in-browser on all devices without downloading. Supports bookmarks (stored to user profile), text highlighting, and notes. Reading progress is saved automatically so a returning user resumes from where they left off. Mobile-optimised with pinch-to-zoom and page-by-page navigation.

Author upload portal

A self-service portal where verified authors upload manuscripts, enter metadata, and submit for review. Admin reviews and publishes in one click. Files are stored in AWS S3 with a presigned URL delivery system — direct access to S3 paths is not possible, protecting the content from being shared or scraped directly.

Social features

Users can share titles via a canonical URL (not a file link), write short reviews, and follow authors for new publication notifications. A "reading list" feature lets users save titles they want to read later.

Analytics as a content-strategy input

Every meaningful interaction — searches, opens, completion rate, bookmark saves — feeds an analytics view the publisher's editorial team checks weekly. Instead of guessing what to commission next, they can see which categories are searched for but under-stocked, and which published titles are opened but abandoned early, which shaped several acquisition decisions in the months after launch.

Results

• ~45% improvement in content discovery speed
• ~50% reduction in search latency (vs. manual Browse + Google Drive)
• ~30% increase in average user session duration
• Author upload time reduced from 48-hour admin turnaround to same-day self-service
• Content protection — direct file links eliminated, all access via authenticated sessions

Where this fits

This platform grew out of our broader web development work, applied to a content-heavy, search-driven product rather than a marketing site. Publishers and training providers with a similar library problem are often better served by custom software than a generic CMS, and the same discovery and access-control patterns carry over well to education platforms with large content libraries.