The infrastructure mistakes we find most often
Everything on one box. The site, the database, and the only backup all live on the same server — so the failure that takes the site takes the backup with it. Off-site backups with restores actually tested are the first fix in almost every engagement, not the last.
DNS scattered across a registrar, an old host, and a departed developer’s personal account. Nobody can safely change anything, so changing the website breaks email and nobody knows why. Meanwhile SPF, DKIM, and DMARC sit at defaults, quietly costing deliverability every day and leaving the domain spoofable.
Paying for scale that was never configured — or never needed. Oversized instances, orphaned resources, “enterprise” tiers nobody finished setting up. A cost review is part of every engagement because small businesses routinely pay for capacity they cannot see and do not use.
When shared hosting is actually fine
A brochure site with modest traffic on a reputable managed platform is fine, and we will say so at the review. Migrating it to AWS would add cost and moving parts for nothing — the honest recommendation is sometimes “stay where you are, but fix your backups and DNS.”
The trigger points where that changes: revenue depends on uptime, you run an application backend, you handle customer data with obligations attached, or deploys have become scary. That is when owning the infrastructure — with monitoring, rollback plans, and documentation — starts paying for itself.