An independent benchmark of 15 WordPress hosting providers tested between January and April 2026. Every plan purchased with real money. No sponsored content, no press accounts, no host influence over findings. TTFB tested from 28ms to 612ms. Renewal pricing analyzed across 36-month horizons. CPU hardware verified via SSH. The data tells a story the hosting industry's marketing budgets work hard to obscure.

Executive Summary

I spent 90 days and approximately $2,400 buying and testing hosting plans from 15 WordPress hosts. The goal was to answer the questions that affiliate review sites won't answer honestly: which hosts are actually fast under real traffic, which pricing models trap buyers at renewal, and which infrastructure choices protect sites from failure when traffic spikes.

The findings were starker than I expected. The performance gap between the best and worst hosts is not 20% or even 100%. The fastest host (ScalaHosting, 28ms TTFB) is 22 times faster at server response than the slowest (GoDaddy, 612ms). Both are marketed as WordPress hosting. Both charge approximately $35 to $120 per year for introductory plans. The difference in what you get for that money is a 22-fold performance gap that directly affects Google rankings, user experience, and conversion rates.

The pricing findings were equally striking. The average renewal price increase across the 15 tested hosts is 287%. Three hosts exceed 500% renewal price jumps. GoDaddy's Business plan increases 569% from intro to renewal. This is not disclosed prominently in checkout flows. Most buyers discover the renewal price when they receive an invoice in Year 2, at which point switching costs (effort, risk) are high enough that many renew rather than migrate.

The infrastructure findings matter most for long-term site health. Only 6 of the 15 hosts offer true account isolation (CloudLinux, CageFS, or container equivalents). The remaining 9 hosts run on standard Linux shared hosting where a compromised or misconfigured neighbor account can read your files, consume your server's resources, or cause downtime that isn't your fault. Most buyers do not know to ask about isolation. Most review sites do not mention it.

Part 1: Performance Findings

TTFB Rankings: The Full Spectrum

TTFB (Time to First Byte) measures how long the server takes to respond to a request before any content is sent. It is the server's contribution to page load time. Lower is better. Under 100ms is good. Above 300ms is a problem for Google ranking and user experience.

All tests were run from WebPageTest (Dulles, VA, Cable connection, Chrome) with 9 runs per host. Median taken. WordPress 6.5 installed with Twenty Twenty-Four theme, WP Super Cache disabled. No CDN active. No object cache. The goal was to measure server performance, not cache performance.

The 22x Performance Gap: What It Means in Practice

A 28ms TTFB (ScalaHosting) vs 612ms TTFB (GoDaddy) is a 584ms gap. That 584ms represents the time your server is thinking before it sends a single byte to the visitor's browser. You can optimize your WordPress theme, compress images, enable browser caching, and reduce HTTP requests — and none of those optimizations reduce TTFB, because TTFB is entirely determined by the server.

Google uses TTFB as a component of the Largest Contentful Paint (LCP) Core Web Vitals score. A site with 612ms TTFB has already failed Google's "fast" classification before any content renders. A site with 28ms TTFB starts from a position where "good" LCP is achievable with basic optimization. The practical implication: a site on GoDaddy and a site on ScalaHosting with identical content, identical WordPress optimization, and identical CDN configuration will have different Google rankings because the server response time is different.

Load Degradation: When Traffic Hits

TTFB at idle is one thing. TTFB under 50 concurrent users is what you actually experience during a product launch, a media mention, or a seasonal traffic peak. The degradation pattern is where shared hosting falls apart.

Bluehost at idle: 380ms. Bluehost at 50 concurrent users: 1,240ms. That is 226% degradation — a site that was borderline slow at idle becomes functionally unusable under real traffic. The degradation happens because shared hosting's CPU and I/O resources are finite and shared. When 50 visitors hit the server simultaneously, they are competing for the same CPU cores. A site that uses 0.3 CPU cores per request at idle suddenly needs 15 cores simultaneously at 50 concurrent users — far more than the shared allocation provides.

ScalaHosting's dedicated vCPU cores prevent this. At 50 concurrent users, ScalaHosting moves from 28ms to 31ms — an 11% degradation vs Bluehost's 226%. The same LiteSpeed cache that serves responses from memory at idle continues serving from memory under load. The performance floor is protected by the hardware allocation, not shared with 500 neighbors.

Part 2: The Renewal Pricing Trap

How the Bait-and-Switch Works

Hosting companies use intro pricing to acquire customers, then recapture margin at renewal. The business model depends on switching inertia: once a site is live on a host, the effort and risk of migrating is high enough that most customers pay the renewal rather than move. This creates a structural incentive to offer unsustainably low intro prices and recoup them aggressively at renewal.

The intro-to-renewal increase is disclosed in the terms of service, but not in the primary purchase flow. Most buyers discover the renewal price when they receive a renewal invoice — at which point their site is live, their DNS is configured, their email is on the host, and the switching cost is high. This is the intentional design of the pricing strategy.

Hosts With Zero Renewal Shock

Three hosts in the tested group have no renewal price increase: Cloudways ($50/mo flat), Kinsta ($35/mo flat), and FastComet (minimal increase). These hosts compete on predictability. Their upfront price is higher than budget hosts' intro rates, but their 3-year total cost is often lower when compared honestly against budget hosts' renewal rates.

The pattern is consistent: hosts with zero renewal shock are generally the better-performing hosts. This is not coincidence — both reflect the same business philosophy. A host that invests in infrastructure does not need predatory renewal pricing to recoup margin. A host that uses 2012-era hardware and oversells resources needs that renewal premium to maintain margins on customers who don't leave.

Beyond Renewal Pricing: The Add-On Fee Structure

Renewal pricing is one layer. Add-on fees are the second layer. Several of the tested hosts include features in intro pricing that become paid add-ons at renewal or require separate purchase:

• SSL certificates: Free from Let's Encrypt on most hosts, but GoDaddy charges $5.99 to $7.99/mo for SSL as a separate product
• Backups: Bluehost and HostGator include automated backups in higher tiers only; entry-tier plans require a paid CodeGuard add-on ($3 to $4/mo) or Backup Pro ($3/mo)
• Security: GoDaddy's SiteLock security add-on is $6.99/mo. The checkout flow presents it as a selected option, not an optional add-on, for new buyers who don't read carefully
• Domain privacy: Most hosts charge $0.99 to $1.99/mo for WHOIS privacy, which is free at several registrars

When you total the add-ons that a production business site actually needs (SSL, backups, security), the effective Year 1 cost for budget hosts often exceeds their headline rate by 30% to 80%. ScalaHosting's managed VPS includes all of these at no extra cost.

Part 3: Infrastructure Findings

Account Isolation: Who Uses CloudLinux

CloudLinux is an operating system layer that adds per-account resource limits and filesystem isolation (CageFS) to shared hosting. It prevents one account from reading another's files, prevents one account's runaway process from consuming all CPU, and limits the blast radius when an account is compromised or misconfigured. It costs hosts approximately $8 to $15 per server per month.

Of the 15 tested hosts, 6 use CloudLinux or an equivalent isolation technology:

• ScalaHosting: KVM-based isolation with dedicated vCPU cores on managed VPS — stronger than CloudLinux
• SiteGround: Custom container isolation built on top of their own control panel stack
• Kinsta and WP Engine: Docker/Kubernetes container isolation, more aggressive than CloudLinux
• ChemiCloud: CloudLinux with CageFS on all shared accounts
• FastComet: CloudLinux with LVE limits enforced

The remaining 9 hosts (including Bluehost, HostGator, GoDaddy, iPage, DreamHost, Network Solutions) run on standard Linux without account-level isolation. On these hosts, a compromised account — one that a spammer uses to send email, one that malware infects, one that a developer misconfigures with world-readable permissions — can potentially affect neighboring accounts. Isolation is the infrastructure investment that separates quality shared hosting from bottom-tier shared hosting.

CPU Hardware: The PassMark Correlation

CPU performance determines how quickly PHP processes WordPress requests. Server CPUs are ranked by Passmark (passmark.com/cpu_list). A higher Passmark score means faster WordPress PHP execution. You can verify a VPS host's CPU with lscpu | grep "Model name" via SSH.

Key hardware findings from the 15-host test:

• ScalaHosting: AMD EPYC 7003 series (PassMark single-thread score in top 50 server CPUs) — dedicated cores, no sharing
• Cloudways Vultr HF: AMD EPYC 7402P or similar (PassMark top 100 server CPUs) — burstable, not dedicated
• Kinsta/WP Engine: Google Cloud/AWS modern instances (2021 to 2023 hardware)
• SiteGround, ChemiCloud, FastComet: Mixed 2020 to 2022 CPUs, actual model varies by server
• HostGator, Bluehost, iPage: Intel Xeon E5 and E7 series from 2012 to 2016 (PassMark rank 800 to 1200 in server CPU list)
• GoDaddy: Intel Xeon from 2013 to 2015 (PassMark rank 900 to 1400)

The CPU generation gap between top and bottom hosts is approximately 8 to 10 times in per-core PHP throughput. A WordPress request that takes 40ms of CPU time on ScalaHosting's AMD EPYC takes 320 to 400ms of CPU time on HostGator's 2012-era Xeon. This is the physics underlying the TTFB gap — not configuration or optimization, but raw hardware capability.

Backup Retention: The Policy Comparison

HostGator's backup policy is the most problematic finding in this category. Their terms of service state explicitly that backups are performed "as a courtesy" and cannot be relied upon for data recovery. They are not liable for backup failures. In practice, this means a HostGator shared hosting account has no contractual backup guarantee. A malware infection that corrupts your database, a migration error that deletes files, a ransomware attack — none of these scenarios are covered by HostGator's backup system on entry plans. This is a serious risk for production WordPress sites.

Methodology

Test Environment

All hosting plans were purchased under standard retail conditions: no discount codes, no press accounts, no hosting company contact before purchase. Plans were selected to represent entry-level or mid-tier shared hosting (for shared host comparisons) or entry-level managed VPS (for VPS comparisons).

WordPress installation: WordPress 6.5 with Twenty Twenty-Four theme, Elementor not installed, WooCommerce not installed (separate WooCommerce tests run with a 10-product catalog). Standard plugins: Wordfence (security), WP Super Cache (caching, disabled for base TTFB tests), Yoast SEO (metadata).

Test content: a homepage with 1,500 words of text, one featured image (WebP, 120KB), and a standard navigation menu. No lazy-loading JavaScript, no third-party scripts. This is the minimum viable WordPress page for accurate TTFB measurement.

Speed Testing Protocol

Tool: WebPageTest (webpagetest.org). Location: Dulles, Virginia (Dulles is the default test location for most US hosting benchmarks, providing a geographically neutral baseline). Connection: Cable (30 Mbps/5 Mbps simulated). Browser: Chrome latest. Runs: 9 per host. Metric taken: Median TTFB from the 9 runs (not the fastest, which removes outliers in both directions).

Load testing tool: Loader.io. Protocol: Maintain-clients (not ramp) mode. Test duration: 60 seconds per load level. Load levels: 10, 25, 50, 100, and 250 concurrent users. Success threshold: HTTP 200 responses. Error rate defined as any HTTP 5xx response.

CPU Verification Method

For VPS hosts with SSH access: lscpu | grep "Model name" to get the exact CPU model. Cross-referenced against Passmark's CPU benchmark list (passmark.com/cpu_list) for performance ranking. For shared hosts without SSH access: CPU model derived from LVE Manager data (if CloudLinux) or from host's public documentation. Where documentation was unavailable, PassMark score is listed as "not verified."

Pricing Analysis Method

Renewal prices sourced from: (1) host's own terms of service and pricing FAQ pages, (2) direct account dashboard showing renewal rates, (3) checkout flow for renewals on active accounts. Where renewal prices are shown as ranges, the most common renewal rate for the test plan tier was used. All prices are USD monthly equivalents. Three-year totals calculated as: (intro price × 12) + (renewal price × 24).

Key Conclusions and Recommendations

Recommendations by Use Case

For a solo WordPress blog or portfolio site (under 10,000 visits/month, no revenue attached):
ChemiCloud or FastComet. Both use CloudLinux isolation, modern-enough hardware (2020 to 2021), reasonable renewal pricing (200% to 300% increase), and daily backups included. They are not the fastest hosts tested, but they are fast enough for non-revenue sites and honest enough in pricing to recommend without reservation. Cost range: $200 to $250 over 3 years.

For a small business WordPress site (10,000 to 100,000 visits/month, some revenue attached):
ScalaHosting managed VPS (Build #1). The 28ms TTFB and dedicated vCPU cores deliver a meaningful SEO and conversion advantage that shared hosting cannot match. The Year 1 intro price ($29.95/mo) is reasonable. Included email and daily backups eliminate add-on costs. The renewal price ($82/mo) is higher, but so is the performance — evaluate at Month 11 whether the speed advantage is generating measurable business value. If not, Cloudways at $50/mo is the best alternative.

For a WooCommerce store:
ScalaHosting managed VPS. Dedicated vCPU cores prevent the noisy-neighbor problem that causes checkout slowdowns during traffic spikes. LiteSpeed Enterprise serves cacheable product pages from memory at 28ms. Uncacheable cart and checkout pages benefit from the dedicated core allocation. WooCommerce stores that experience 500ms+ checkout load times lose measurable conversion — the $30/mo incremental cost over a good shared host typically recovers in reduced abandonment within 2 to 3 months.

For an agency managing multiple client sites:
Cloudways. The multi-cloud flexibility, team collaboration features, application-level access controls, and flat $50/mo pricing make it the most capable platform for managing multiple unrelated WordPress sites under one account. ScalaHosting's SPanel does not have equivalent multi-tenant management features. Cloudways wins the agency workflow comparison conclusively.

For a high-traffic authority site (100,000+ visits/month):
Kinsta or WP Engine. Both are fully managed WordPress-only platforms with hardware in Google Cloud or AWS, dedicated infrastructure per site, and expert WordPress-specific support. Their pricing ($35 to $200/mo) reflects the premium, but for sites where server failure means significant revenue loss, the managed-platform risk reduction is worth the premium.

Who should avoid budget shared hosting entirely: Any site running WooCommerce, any site in the top 10 Google results for competitive keywords, any membership site with logged-in user sessions, any site that generates more than $500/month in revenue. At those stakes, the performance and reliability gap between budget shared hosting and managed VPS is not a marginal difference — it is a business-critical variable.

Industry Patterns Worth Noting

Several patterns in the 2026 data deserve attention beyond the individual host rankings:

The Newfold Consolidation Effect. Every Newfold-owned brand (HostGator, Bluehost, iPage, Network Solutions) tested in the bottom half of performance rankings. This is not coincidence — it reflects infrastructure investment decisions made at the holding company level. The hardware refresh cycle for these brands has effectively stopped. The performance gap between Newfold brands and independent hosts widens each year as competitors upgrade hardware while Newfold coasts on existing assets.

LiteSpeed vs Apache performance gap. Hosts running LiteSpeed Enterprise (ScalaHosting, SiteGround, ChemiCloud, A2 Turbo) consistently outperform equivalent hardware hosts running Apache. LiteSpeed's server-level caching eliminates PHP execution for cached pages, serving from memory instead. Apache-only hosts require WordPress-level caching plugins that are less efficient. When hardware quality is equal, LiteSpeed adds approximately 30% to 50% TTFB improvement over Apache. When hardware quality differs (ScalaHosting's AMD EPYC vs HostGator's Xeon), both hardware and server software contribute to the gap.

Managed VPS vs. shared hosting is no longer a premium question. ScalaHosting's managed VPS at $29.95/mo intro is cheaper than Year 2 renewal prices for Bluehost, HostGator, GoDaddy, and SiteGround. You can get managed VPS performance — dedicated CPU cores, faster hardware, better isolation — for the same or less than shared hosting renewal pricing. The historical assumption that managed VPS is significantly more expensive than shared hosting is no longer accurate for Year 2+ total cost.

How to Use This Report

Use the TTFB benchmark table to anchor your performance expectations. If your current host is in the bottom half of the table and your site has any business significance, the data quantifies exactly what you are losing in server response time. Run WebPageTest on your own site (webpagetest.org, Dulles VA, Chrome, 3 runs) and compare your TTFB against the benchmarks here.

Use the renewal pricing table before your intro period ends. Check where your current host falls and calculate your 3-year total cost. If you are in Year 1 of a budget shared plan, you are 12 to 24 months away from a renewal price shock. Plan now whether to absorb it, negotiate it, or migrate — migrating before renewal while your account is stable is easier than migrating reactively after an unwanted invoice.

Use the infrastructure findings as a checklist. Ask your host: Do you use CloudLinux? Are automated daily backups included in my plan or paid extra? What CPU model is running on my server? If your host can't answer these questions through their documentation or support, that is itself a data point about their transparency.

Download the Full Report

Transparency and Methodology Notes

This report contains no sponsored content and was not funded by any hosting company. No host paid to be included, ranked, or excluded. ThatMy.com is independently run and funded by its author through affiliate commissions (when readers purchase hosting through links on this site). Affiliate relationships do not influence benchmark methodology or test results. The data was collected under identical conditions for all 15 hosts.

Limitations: testing was performed from a single geographic location (Dulles, VA) using a single test tool configuration. Real-world performance varies by visitor location, time of day, current server load, and WordPress configuration. The benchmarks represent a reproducible baseline, not a guarantee of performance for any individual site. Hosting performance changes over time as hosts upgrade or degrade infrastructure. This report reflects testing conducted in January to April 2026.

How to Cite This Report

If you reference this report in writing, publication, journalism, or research, please cite as:

Supe, M. (2026). State of WordPress Hosting 2026: Annual Industry Report. ThatMy.com. Retrieved from https://thatmy.com/research/state-of-wordpress-hosting-2026/

For questions about methodology, data, or findings: mangesh@thatmy.com. Response within 24 hours.

Data Appendices

• WordPress Hosting Benchmarks — Full TTFB and load test data with raw results
• Hosting CPU Rankings — PassMark cross-reference for all tested hosts
• Hosting Renewal Pricing Database — Intro vs Year 2 analysis with TOS citations
• CloudLinux and Account Isolation Guide — Detailed explanation of isolation technology
• How to Check Your Host's CPU with PassMark — Verification method guide