Cost formula published in full Prices checked July 2026 Speed benchmarks not measured yet
Badri Dutta
Tap a tab — or drag the bead along the bar.
23 guides · 4 sections
Developer · 8 years technical SEO · No paid placements

Build it, host it, get it found.

Hosting is written about by people who never had to rank a page. SEO is written about by people who never had to deploy one. This site covers the chain that connects them: what you build sets your performance, performance sets your visibility, and each decision has a number attached.

Model your bill — free Browse the hosting guides

Test methodology published with every comparison.

ZeroFabricated figures
One formulaPublished on the page
Platforms tested AWSDigitalOceanHetzner VercelNetlifyCloudflare RailwayFly.ioLinode Vultr
15 yrsBuilding for the web
8+ yrsTechnical SEO practice
4Hosting tiers modelled
ZeroPaid placements accepted
The chain

Three problems, usually written about by three different people

They are not separate. A hosting choice becomes a performance number, a performance number becomes a visibility problem, and almost nobody writing about one has done the other two.

01

Host it

Infrastructure priced from a published model, not a pricing page.

  • 4 tiers modelled live
  • 12 stacks mapped live
  • Cost calculator live
  • Migration sequence live

Open the hosting guide →

02

Build it well

Design and build-quality decisions that show up as Core Web Vitals numbers.

  • INP, hosting & rendering live
  • Core Web Vitals writing
  • Render-path diagnosis live
  • Accessibility & security headers planned

Open the design guide →

03

Get it found

Technical SEO for people who read code — and the newer question of AI citation.

  • Answer-engine visibility live
  • llms.txt, evidence-checked live
  • Structured data live
  • Ranking case studies planned

Open the visibility guide →

04

Watch what's changing

AI notes filtered through one question: does this change hosting, crawling, or how software gets built.

  • How coding agents pick files live
  • Model & tooling notes irregular
  • Not a news roundup by design

Open the AI notes →

What is actually here

Eight guides, four sections, one standard.

A section only enters the navigation once it holds five published guides — not when the folder is created. Three of the four are still below that line, and the panel says so rather than rounding up to look finished.

editorial standard

Every hosting bill is arithmetic. Publish the arithmetic and the recommendation defends itself.

Badri Dutta

Coverage

Counted from what is published, not planned

Current
Guides live
8
across 4 sections
Hosting tiers
4
modelled end to end
Stacks mapped
12
runtime to tier
Paid placements
0
since 2026
By sectionGuides
  • Hosting — in navigation 4
  • Visibility — below threshold 2
  • Design — below threshold 1
  • AI notes — publishes irregularly 1
$0To read any of it
5 postsBefore a section ships
What makes it different

The numbers come from a deployed app, not a pricing page

Cost modelling

Headline prices hide 80% of the bill

A managed platform advertises $20 a month. At 250,000 views the modelled invoice is $102.24 — the difference is bandwidth overage, described only in a pricing footnote.

  • Every line item itemised, including overage tiers
  • Modelled at a stated request volume, not "typical usage"
  • Image weight treated as a cost variable, because it is one
  • List pricing dated, and re-checked rather than left to rot
invoice · managed platform
Base plan Pro, 1 seat
$20.00
Bandwidth overage 548 GB @ $0.15
$82.24
Invocations 95k of 1M included
$0.00
Image optimisation Included
$0.00
Modelled total$102.24
Scaling

The gap widens with every gigabyte

A managed platform's base tier hides the problem at low traffic. Past roughly 163,000 views at 5.6 MB pages, bandwidth overage starts compounding while the VPS price stays flat until the next tier.

  • No cost break-even — a VPS is cheaper at every volume
  • The managed premium buys operational time, not savings
  • Page weight is a cost variable, not just a speed one
  • Change any assumption and re-run it yourself
modelled cost · managed vs vps
25k views$20 / $12
100k views$20 / $12
250k views$102 / $28
600k views$432 / $44
1M views$810 / $80

Bar length = managed platform cost. List pricing July 2026, 5.6 MB median page weight.

Multi-stack coverage

Not just WordPress.
Every major stack.

Framework choice constrains hosting more sharply than traffic does. Each stack is paired with the tier its runtime actually needs.

12
Stacks mapped
4
Hosting tiers
1
Published formula
Jul 26
Pricing verified

JavaScript & Node

Next.js

SSR · ISR → VPS or managed

Remix

Edge or Node → managed or VPS

Astro

Static output → CDN

Eleventy

Plain files → CDN

Python & JVM

Django

Long-lived process → VPS

Flask / FastAPI

ASGI, lightweight → small VPS

Spring Boot

JVM heap → VPS, 4 GB min

Docker

Anything containerised → VPS

PHP & content platforms

WordPress

PHP workers → shared or managed WP

Laravel

PHP-FPM · queues → VPS

Rails

Puma · jobs → VPS

Hugo

Build-time only → CDN

$74
Modelled monthly gap · 250k views

One formula. Published in full. Re-runnable today.

Every cost figure on this site comes from the same stated model applied to published list pricing — no proprietary data, nothing you cannot verify against a provider's own pricing page. Disagree with an assumption and you can change it in the calculator.

How to use this site

From question to a decision in under ten minutes

Model your bill

Put your traffic, page weight and server-rendered share into the calculator. It applies the overage tiers the pricing pages bury and returns all three hosting models side by side.

~2 minutes

Check your framework

Confirm the tier the calculator points at can actually run your stack. A Node runtime rules out shared hosting no matter what the price comparison says.

~3 minutes

Read the migration sequence

If you are moving an existing site, follow the DNS, TLS and redirect ordering. That sequence is what decides whether the move costs you rankings.

~5 minutes
Where you sit

The right answer depends on what your hour is worth

Cost is only half the decision. The other half is how much maintenance time you actually have — and that changes the answer completely.

Side project

Under 20k visits

A blog, a portfolio, or something you are still validating. Hosting cost barely registers at this size.

Static + CDNOptimise for least effort, not lowest price.
Solo developer

20k – 150k visits

A real product with real users, run by one person who can spare an hour a month on servers.

VPS + CDNCheapest and fastest at this range, by a wide margin.
Small team

150k – 1M visits

Shipping features matters more than shaving $60 off infrastructure. Nobody wants to be on call for a server.

Managed platformPay the premium, buy back the ops time.
Scaling up

Above 1M visits

Bandwidth economics dominate. Egress pricing decides the bill, not the compute tier you picked.

VPS + heavy CDN offloadCache strategy is now the cost strategy.
Free tool

Most sites overpay for hosting. How much are you losing?

Move the sliders. The model applies published list pricing including the bandwidth and invocation overage tiers that make headline prices misleading.

Modelled egress: 1,572 GB · VPS tier: 4 vCPU / 8 GB

How this is calculated

This is a model built from published list pricing, not a measurement. Every figure is reproducible from the formula below — no proprietary data, nothing you cannot check against the providers' own pricing pages.

egress = views × pageWeightMB ÷ 1024 × 1.15 managed = $20 + max(0, egress − 1024GB) × $0.15 + max(0, invocations − 1M) ÷ 1M × $0.60 vps = tier price + max(0, egress − 5TB) × $0.01 static = $0
The 1.15 multiplier
A 15% allowance for assets that miss the cache and hit the origin.
VPS tiers
$12 to 100k views · $28 to 300k · $44 to 800k · $68 above, each with 5 TB transfer included.
Static at $0
Major CDN free tiers do not meter bandwidth for static files.
Not modelled
Reserved-instance discounts, enterprise agreements, regional price differences, and your own time.
Managed platform$20 base + 548 GB over @ $0.15 $102.24
VPSFlat tier price, 5 TB transfer included $28.00
Static + CDNOnly if you can drop server rendering $0.00

A VPS is $74.24/month cheaper — $890.88 a year.

Runs entirely in your browser. Nothing is stored or sent anywhere.

Sample output

What is the cheapest way to host a small production app?

A virtual private server is the cheapest option at every volume once overage tiers are modelled — $12 a month under 100,000 views against $20 on a managed platform, widening to $28 against $102.24 at 250,000. Managed platforms buy operational time, never money.

Monthly cost and operational burden by hosting type. Costs modelled from published list pricing, July 2026; latency publishes once measured.
Hosting type Cost / month (USD) Median TTFB Ops burden Stops working when
Shared hosting $3 – $10 pending None You need Node, background jobs or real CPU
Virtual private server $6 – $28 pending You patch and monitor it You have no time to maintain a server
Managed platform $20 – $101 pending Minimal Bandwidth or invocations scale past the base tier
Static host + CDN $0 – $5 pending Build pipeline only You need server-side rendering or a database
The difference

Why most hosting comparisons are worthless

Not a swipe at competitors — a description of what the format rewards, and what this site does instead.

● Typical hosting round-ups

Common industry practice

Patterns observed across the affiliate content that dominates these queries.

  • 01 Prices copied, never computed. Headline plan prices lifted straight from a pricing page, with overage tiers — usually most of the bill — left out entirely.
  • 02 Nothing deployed. Rankings written without the author ever running the software, so "fast" and "reliable" are adjectives rather than numbers.
  • 03 Order follows commission. The highest-paying affiliate programme is placed first, and the ranking is presented as a finding.
  • 04 Undated figures. Prices from three years ago presented as current, with no verification date anywhere on the page.
  • 05 No negative findings. Every listed provider is "great for" something, because a bad review ends the partnership.
★ This site

How this works instead

Standards this site is written under. They exist to be checked against.

  • 01 The formula is published. Every cost figure derives from a stated model you can re-run — overage tiers included, arithmetic shown.
  • 02 Assumptions stated up front. Page weight, traffic, cache-miss allowance and what is deliberately not modelled, all on the page.
  • 03 Order follows the maths. Cheapest is cheapest. No provider can pay for placement, and the calculator has no thumb on the scale.
  • 04 Everything carries a date. List pricing verified July 2026, re-checked rather than left to rot, with corrections made in place.
  • 05 Findings that lose the click. The diagnostics guide talks you out of migrating when your real problem is a missing database index.
What is modelled and what is not. Cost figures on this site are modelled from published list pricing and are reproducible today. Latency and cold-start benchmarks are a separate matter — the test method is described in full below, but measured results are not published until the test rig has actually been run. No performance number appears on this site before it has been measured.
Test method

How benchmarks will be produced

This site has no testimonials and no review scores, because it is new and inventing either would be dishonest. What it has instead is a procedure you can repeat and disagree with.

Time to first byte (TTFB)
The delay between a request leaving the browser and the first byte of the response arriving. It isolates server and network latency from rendering work, which is why it is the clearest single signal of whether hosting is the bottleneck.

  1. One application, every provider

    The same reference app — server-rendered pages, a database read, one image-heavy route — deployed unchanged to each provider tested.

  2. Measured from three regions

    Frankfurt, Mumbai and Virginia, 50 requests each, median reported rather than best case. Cold-start behaviour recorded separately.

  3. Cost modelled at real traffic

    Bills calculated at a stated request volume and bandwidth figure, including the overage tiers that make headline prices misleading.

  4. Dated, and re-checked

    Hosting prices move. Every figure carries the month it was verified, and comparisons are revisited rather than left to rot.

Who writes this

Fifteen years of doing it, then writing it down

2011 →

Hand-written front-end

HTML, CSS, JavaScript and jQuery, back when a layout meant floats and a lot of patience.

2016 →

Search work that ranked

Eight years of on-page and technical SEO, including small business sites taken to page one and held there.

2019 →

Backend and cloud

Java and Python in production, AWS infrastructure, and the cost discipline that comes with running it.

2022 →

Applied AI

Building with modern model APIs, and tracking how answer engines change what visibility means.

Read the full background →

Guides

Published so far

Every guide states its test conditions and the date the pricing was checked.

Questions

Common questions

How are hosting recommendations on this site tested?

Every hosting recommendation comes from a deployed test application, not a spec sheet. Each provider runs the same reference app, measured for cold-start time, time to first byte from three regions, and total monthly cost at a defined traffic level. Test conditions and dates are published with each comparison.

Does this site use affiliate links?

Yes, some hosting links are affiliate links, and a disclosure appears above the first affiliate link on every page that contains one. Rankings are based on measured results only. Providers cannot pay for placement, and negative findings are published for affiliate partners the same as anyone else.

Which topics does this site cover?

The current focus is hosting and deployment: choosing infrastructure by framework, cost modelling, migrations, and performance tuning. Web design and build quality is the next area, followed by technical SEO and answer-engine visibility. Sections are published once they hold genuine depth.

Is hosting choice actually a ranking factor for Google?

Hosting is not a direct ranking factor, but it determines server response time, which feeds Core Web Vitals and crawl efficiency. A slow origin raises Largest Contentful Paint and reduces how much of a large site Google crawls per visit. The effect is indirect but measurable.

How accurate is the hosting cost calculator?

The calculator applies published list pricing including overage tiers, which covers the majority of real bills. It does not model reserved-instance discounts, enterprise agreements or regional price differences, so treat the output as a close estimate rather than a quotation.

Free tool · No account · No signup wall

Stop guessing what hosting costs.
Model the real bill in two minutes.

Published list pricing, including the bandwidth and invocation overage tiers that turn an advertised $20 plan into a $102.24 invoice. Three hosting models, side by side.

Every line item Overage tiers included 3-region benchmarks No paid placements
Covers Managed platforms VPS Static + CDN Shared
Worked example · not an average
$890
Per year · 250k views · 5.6 MB pages
Open the calculator Or read the hosting guide first →
Runs in your browser Nothing stored or sent List pricing July 2026 Method published in full