Context

When I joined CoinDesk in December 2022, I started learning about Arc XP, the SaaS platform we used to host our website. Arc had two main components - Composer (the CMS) and PageBuilder (the front-end). After spending time with the system, I realized Arc wasn't a great fit for CoinDesk. While it might work well for companies that just publish articles or videos, we needed something better for our mix of news, market data, indices, and event information.

We had a lot of problems with Arc and they affected everyone at CoinDesk. Our editorial team struggled to publish time-sensitive stories, developers couldn't solve technical issues, and business teams couldn't launch new products efficiently. It became clear we had simply outgrown what Arc could offer us.

Challenge

Our problems went deeper than just using a limiting SaaS platform. Our information architecture had grown without a clear plan, resulting in over 80 different modules that made the system unnecessarily complex for both our technical teams and content creators. We also couldn't easily connect different types of content because Arc's structure was too rigid.

The biggest issue was our lack of control. We couldn't fix critical performance problems affecting our Core Web Vitals within Arc's limitations. When hit with scraper attacks, we were essentially helpless - our site would keep working, but our editorial team couldn't publish new stories because the APIs would be overloaded. We needed more control over our own platform.

Approach

Instead of just moving from one platform to another, I saw a chance to really improve things. In March 2024, we began the process of rebuilding our digital foundation with Sanity.io as our headless CMS and creating a modern front-end using React and NextJS on Vercel. This wasn't just a technical change - it was about gaining independence and flexibility.

I focused the first two months entirely on product and design work, knowing we needed to get that right before writing a single line of code. During this planning phase, we simplified our component architecture from 80 modules to just 23. This wasn't about cutting corners - we thoughtfully redesigned everything based on what our content creators and audience actually needed.

The engineering work took another five months as we built the CMS schema, developed the front-end, and moved all our data. Throughout this process, we stayed committed to creating something simpler, faster, and more connected than what we had before.

Outcomes

When we launched the new platform in November 2024, we saw immediate improvements. Our Core Web Vital scores improved across all page types, making the site faster and more responsive for our users. Shortly after launch, we faced a scraper attack - the kind that would have crippled us on Arc. Instead, we blocked the attacker at the WAF level within minutes, something we couldn't have done before.

The most significant change has been our development speed. Projects that once took months now take weeks, and tasks that took weeks can be done in days. This isn't just about working faster - it's about being able to respond quickly to new opportunities and better serve our audience's changing needs.

Key Takeaways