My wish list (as a web developer that works with clients and web hosts) is pretty much the same as last year.
- True TTFB consistency, not just the best-case scenario benchmarks. TTFB is a huge bottleneck that directly relates to the server, I/O on database connections and then the internet trunk support speeds
- When hosts demo their speed, use real HTML, PHP and Database examples, not just cached demos. We used to put dummy 1MB, 5MB and 20MB files that people could download and test from a host/datacenter - this seems to be gone these days. Show me real-world page loading speed, database connection and retrieval speeds etc.
- Automatic containment if malware is detected. Too many hosts never police their servers and just let exposure run wild. Even a basic Base64 or eval scan can cut this off, and then add specific commands to whitelists for clients etc if the base64 encoding was actually required
- SSH for all users
- WP-CLI is installed by default
- Realtime dashboard showing CPU/RAM/PHP Workers, I/O etc. Transparency seems to be forgotten and in it's place is a sales pitch saying a user needs to update rather than tracking a real issue in the site, fixing it and not needing to update.
- Real Staging sites and preset to not allow spiders to crawl, block Google Analytcis, disable credit card process, disable emailing, and are password-protected by default
- True access to log files (access logs, error logs, php logs) for the client.
I know some of these costs come with a cost increase, but if a host decided to take these items on, they'd be a huge upsell and instead of raising the price, they'd recoup costs through attrition.
The biggest item that has continued to fail is
true transparency. Too many hosts hide behind the "it's working now" answer and do not provide information. If the SSL suddenly stopped working and needed to be installed, WHY did that happen? It's a Let's Encrypt, which auto-renews and was just renewed 14 days ago - why does it have to be force-renewed again? The reality being, it didn't have to be renewed, but that was the generic answer given, rather than saying the cron job failed to run or there was an I/O bottleneck.
Web hosts not taking real accountability sucks - especially when I've been on the other side! And if it's the client that exceeded resources, state it, and back it up with logs/documentation.
