We Self-Host Nothing: Why Every Client Site Runs on Vercel

Every client site we build — ours included, after we moved ekyatech.com off WordPress ourselves — runs on Vercel. Not because it's trendy. Because the alternative means someone on a small team becomes an unpaid, part-time systems administrator, and that job has real costs most SMEs never actually budget for.

What self-hosting actually costs, beyond the server bill

A VM on any cloud provider looks cheap on the pricing page — a few hundred rupees a month for something small. That number doesn't include:

• Patching the OS and every package on it, indefinitely, or accepting the security risk of not doing it.
• SSL certificate renewal — automated with something like certbot, until the automation silently breaks and nobody notices until the site shows a browser warning.
• Monitoring uptime yourself, because nobody else is watching that specific box.
• Being the person who gets paged when a disk fills up, a process hangs, or traffic spikes and the server can't handle it.

None of that shows up on the hosting invoice. All of it is real work that either someone on the team does badly (because it's not their actual job) or nobody does at all, until it becomes an incident.

What "we self-host nothing" actually means

Every deploy is a git push. Vercel builds it, runs it through its own pipeline, and serves it from an edge network with a node in Mumbai — no SSH key ever touches the box, because there's no persistent box to touch. Every deploy also gets its own preview URL before it goes live, so a change can be reviewed on a real, running version of the site before it reaches production, not just read as a diff.

Rollback is instant: point traffic back at the previous build, which still exists, rather than restoring from a backup and hoping it's current. SSL, HTTPS redirects, DDoS-level traffic absorption, and CDN caching are handled by the platform, not configured once and forgotten by whoever set it up two years ago.

The honest trade-off

This isn't free of trade-offs. You're on someone else's platform, subject to their pricing changes and their outages (rare, but not zero). If you need exotic server-side capabilities — long-running background jobs, specific compliance requirements around exactly where data physically sits, or genuinely massive scale where a platform's pricing stops making sense — self-hosting or a different cloud setup becomes the right call, and we'd tell a client that plainly rather than force everything onto one platform because it's what we know.

For the overwhelming majority of SME marketing sites, product sites, and small-to-mid web apps, that trade-off isn't close. The failure modes of "the person who set up the server left the company and nobody knows the SSH key" and "a critical patch didn't get applied for eight months" are far more common, and far more expensive when they happen, than anything Vercel-specific.

Why we apply this to our own site too

We didn't migrate ekyatech.com off WordPress and onto Vercel to have a case study — we did it because we were telling clients to make this exact move while sitting on infrastructure we wouldn't have recommended to them. If we're asking a client to trust a platform choice, it should be the one we actually run our own business on.

If you're weighing whether your own site or app should move off self-managed infrastructure, tell us what you're running and we'll give you a straight read — including the cases where staying put is genuinely the right call.