After three years of running a WordPress agency, the question I answer most often isn't "how do I fix this bug" — it's "what hosting should I put this client on?" Here's the decision framework I actually use.
The Variables That Actually Matter
Most hosting comparisons rank providers by speed. Speed matters, but it's not the only variable for agency work:
- Site type — WooCommerce, informational, web app, portfolio
- Monthly budget — what the client will pay for hosting as part of a retainer
- Technical access level — do you need SSH access, custom PHP config, non-WordPress software?
- Support responsibility — are you on call if the site goes down at 11pm, or is the host?
- Scale — how many client sites will you manage on this platform?
Run through these five before looking at price or speed numbers.
The Decision Tree
Branch 1: Is this a WooCommerce store with real revenue?
Revenue < $2k/mo: → Cloudways DigitalOcean 2GB ($14/mo). Better performance than shared hosting, affordable for small stores, configurable enough for basic WooCommerce setup.
Revenue $2k–$10k/mo: → Cloudways DigitalOcean 4GB ($24/mo) or Nexcess StoreBuilder. The store is starting to pay for better hosting. Redis object cache for checkout performance.
Revenue > $10k/mo: → Kinsta. The speed lead (312ms TTFB) and Redis on every plan are worth the $35/mo when checkout speed directly affects a meaningful revenue stream.
Revenue > $50k/mo: → Kinsta or Nexcess on higher-tier plans, or an evaluation of managed WooCommerce infrastructure. At this revenue, hosting cost is a rounding error and optimization matters.
Branch 2: Is this an informational or portfolio site?
Budget < $20/mo all-in: → SiteGround or Hostinger shared hosting. Performance is adequate for sites not competing on page speed.
Budget $20–$50/mo: → Cloudways entry plan or SiteGround GrowBig (post-intro pricing). At this budget, Cloudways' VPS resources are a better deal than shared hosting.
Traffic unknown or potentially viral: → Cloudways (no visitor overage charges) over Kinsta (visitor-based billing). A site that could get press coverage shouldn't have an overage charge surprise.
Branch 3: Does the site need custom server configuration?
Custom PHP settings, non-standard stack, Node.js alongside WordPress: → Cloudways (SSH access, configurable PHP settings) or Hetzner unmanaged VPS.
Standard WordPress, want managed everything: → Kinsta or WP Engine.
Developer on the team who can manage a server: → Hetzner CX22 at €4.15/mo for EU audience, DigitalOcean 2GB for US/global audience. Best price-to-performance with proper stack setup.
Branch 4: What does your support structure look like?
You're available for hosting issues 24/7: → Any of the above; budget accordingly.
You're not on call — client's site needs to be someone else's problem at 3am: → Kinsta or WP Engine. Their support teams handle WordPress emergencies, not just server issues. 4-minute median response (Kinsta) vs 11-minute (WP Engine).
You're an agency with an SLA to clients: → Only Kinsta or WP Engine. Shared hosting downtime is your responsibility to explain; managed hosting downtime is the host's infrastructure issue.
Pricing Model Considerations
Visit-based billing (Kinsta): Plans include a monthly visitor cap. Exceeding it triggers overage charges. Good for predictable traffic; risky for sites that could get media coverage.
Resource-based billing (Cloudways): Pay for server resources (CPU, RAM, storage), not visits. No overage charges when traffic spikes. Better for unpredictable or growing traffic.
Fixed plans (WP Engine, SiteGround): Clear monthly cost regardless of visits (within their limits). Simpler to budget.
For agency billing to clients: I use Cloudways for sites where I want to control margins. At $14/mo cost, charging a client $40/mo for "managed WordPress hosting" is a legitimate margin. With Kinsta at $35/mo, the margin is smaller unless the client is on a premium retainer.
The Multi-Site Economics
If you're managing 20+ client sites, the hosting cost per site matters:
| Platform | Cost for 20 sites | Per-site cost |
|---|---|---|
| Cloudways (10 sites/server ×2) | $28/mo | $1.40/mo |
| Kinsta (20-site plan) | $450/mo | $22.50/mo |
| WP Engine (20-site plan) | ~$290/mo | $14.50/mo |
| Hetzner (3 VPS shared) | ~€15/mo | €0.75/mo |
Cloudways' economics at scale are the strongest case for the platform. A 20-site Cloudways setup delivering 387ms TTFB at $1.40/site vs Kinsta's 312ms at $22.50/site — for most small business sites, the 75ms TTFB difference doesn't justify 16× the cost per site.
The Mistakes I See Agencies Make
Putting every client on one platform regardless of needs. A WooCommerce store with $20k/mo revenue and a 3-page portfolio site have different requirements. Not everything goes on SiteGround.
Not charging clients for hosting separately. If hosting is bundled into a flat retainer, you're incentivized to put clients on the cheapest option. Separate hosting charges align your incentives correctly.
Staying on a bad host "until the retainer ends." If a client's site is slow on shared hosting, it stays slow. The migration conversation gets easier when you show a client: "your TTFB is 489ms and Google is penalizing you for it; for $14/mo more we can cut it by 40%."
Choosing based on affiliate commission. The hosts that pay the most don't correlate with the hosts that perform best. Our benchmark data separates performance from marketing.
My Default Stack (As of 2026)
- Small informational sites: Cloudways DO 2GB ($14/mo), 5–8 sites per server
- Growing WooCommerce: Cloudways DO 4GB ($24/mo), dedicated server
- Premium client sites: Kinsta Starter ($35/mo per site)
- Developer/staging environments: Hetzner CX22 (€4.15/mo)
- Enterprise/compliance: WP Engine Professional
This isn't a universal recommendation — it's what works for my agency's mix of clients and my team's technical comfort. The decision framework above is what I'd give you to build your own.