Most "we need better hosting" requests are picking the wrong tier for the product. Here is how to pick between shared, Vercel, AWS, VPS, and your own server.
Software Development Solutions
Looking for a software development partner?
We build domain-led systems tailored to your industry and workflow. 12 years. 2,100+ engagements.
Almost every founder building a customer-facing product runs into the same fork in the road within their first few releases. "Where should this actually live?" Sometimes the answer is easy because their friend on Twitter told them Vercel is the only sensible pick. Sometimes the answer is easy because a serious-looking blog post said real companies use AWS. Sometimes the answer is easy because the cheapest shared hosting plan came bundled with the domain. All 3 of those "easy" answers are sometimes right and mostly wrong. The right hosting depends on your product, your traffic, your team, and your customers, not on what somebody louder than you already picked.
So how do you actually pick? That is the point of this piece. You will see what hosting actually means in plain English without the buzzwords, the 5 tiers your product can pick between (from the cheapest shared plan to your own physical server), the 3 questions that decide which tier is honest for your product right now, the 4 things that break hosting decisions after launch and quietly compound into bigger bills, a simple shape that keeps your product portable so a host change is a swap and not a rewrite, and the 3 signs it is time to move up to the next tier. All of it is written for the buyer making the call, not the infrastructure engineer setting it up, because the buyer is the person who has to sign off on the monthly bill.
Why does this matter more than a few years back? Because the number of good hosting options has multiplied and their pricing shapes are wildly different. Shared hosting still costs almost nothing and still covers most small business sites. Modern platform hosts like Vercel are cheap at small scale and expensive at big scale in ways you cannot see coming. Managed cloud like AWS looks affordable in the demo and becomes eye-watering the moment your traffic patterns are not the ones the demo assumed. And a well-run VPS at Hetzner or DigitalOcean now runs circles on price against the giants for a lot of workloads. The right answer moved. Most founders' mental model of hosting did not move with it.
5
Tiers your product can pick between: shared hosting, Vercel-style platforms, managed cloud, a VPS, and your own server. Most founders only compare 2.
3
Questions that decide which tier is honest for your product right now: traffic shape, operational load, and where your customers physically are.
4
Things that quietly break hosting decisions after launch: traffic spikes, egress bandwidth bills, region mismatch, and provider outages.
1
Portable shape that keeps your product loose from any one host. Skipping it is why a hosting change becomes a rewrite instead of a swap.
The rest of the piece walks the answer in the order the questions show up during a real hosting conversation. What are these tiers? Which one fits you now? What decides the pick? What breaks after launch? How do you build so a move is cheap? When is it time to move up? Boring on purpose, because boring hosting decisions age better than clever ones.
Hosting Explained Without the Buzzwords
Hosting is the answer to a simple question: which computer, sitting where, actually runs your product for your customers to use? Every option below is a different flavour of that answer. They differ in how much of the work you do versus how much the host does, how easy or hard the pricing is to predict, and how much the setup costs at small scale versus large scale. Once you strip away the marketing words on each provider's homepage, all 5 tiers are just different trades between price, control, and how much of your team's time goes into keeping the thing running.
Shared hosting is the cheapest and simplest end. Your product runs on a server that hundreds of other websites also run on; the host handles almost everything, and you get a control panel like cPanel to click through the basics. It fits small business sites, brochure sites, WordPress installs, and low-traffic e-commerce shops. It becomes uncomfortable the moment your traffic outgrows the shared resources or your product needs anything the host does not offer out of the box.
A platform host (Vercel, Netlify, Render, Railway) is one step up. You push your code and the host figures out how to run it, scale it, and put it in front of your users. It fits modern web apps, especially anything built on Next.js, React, or similar frameworks. It is a joy to use at small scale and can get expensive fast at big scale, especially when your traffic is spiky or you serve large files. This tier is what people mean when they say "serverless" (your product runs on demand, spinning up and down as traffic arrives, without you managing any actual servers).
Managed cloud (AWS, Microsoft Azure, Google Cloud) is where the biggest products live and where the widest range of tools sits. You get almost every capability imaginable, from raw computing power to specialised services for machine learning, video encoding, or global content delivery. You also get pricing that is genuinely hard to predict until you have run for a few months, and a control surface that assumes a full-time infrastructure engineer is on your team. For most small and mid-sized products, this tier is more powerful than they need and more complex than they should be running.
A VPS (short for Virtual Private Server, meaning a slice of a real server that behaves like your own machine) sits between shared hosting and managed cloud. Providers like Hetzner, DigitalOcean, Linode, and Contabo give you a full server for a fixed monthly price, and you decide what to run on it. It is cheaper per unit of computing power than the managed cloud giants, and predictable in a way managed cloud usually is not. The trade is that you (or your team) manage the operating system, the security patches, the backups, and the scaling yourself.
Your own physical server, sometimes called "on-premises" or "on-prem", is the endgame for products with strict data-sovereignty needs, compliance requirements the cloud cannot meet, or legacy systems that were never designed to leave their original building. For most products, this is not the answer. For some, it is the only answer, and no cloud provider is going to close that gap.
The Plain-English Test
If your product is a small business site or a WordPress install with modest traffic, shared hosting is probably enough. If it is a modern web app with predictable traffic, a platform host or a VPS is usually the sweet spot. Managed cloud starts making sense when you actually need one of its specialised services, not just because "serious companies use AWS". And your own server is a rare answer that solves a specific compliance or sovereignty problem, not a general hosting choice.
Which of the 5 Tiers Fits Your Product Right Now?
Which of the 5 tiers is honest for your product today? Almost every "we need better hosting" conversation resolves into one of them once you push on the actual product. Knowing which tier you are in changes everything: the monthly bill, how much of your team's time goes into ops, whether you can scale under a traffic spike, and what happens the day the host has an outage.
5 Tiers
Where Your Product Can Actually Live, Cheapest to Most Powerful
Tier 1
Shared Hosting / cPanel
Hostgator, Bluehost, Namecheap. Small business sites, WordPress installs, brochure sites, low-traffic e-commerce. Cheapest, simplest, most limited. Fine forever for a lot of products.
Tier 2
Platform Host (PaaS)
Vercel, Netlify, Render, Railway. Modern web apps and startups. Cheap and easy at small scale, expensive at big scale, especially with spiky traffic or heavy files.
Tier 3
VPS / DIY Cloud
Hetzner, DigitalOcean, Linode, Contabo. Cost-conscious teams and mid-scale products. Cheapest computing power per unit; you or your team manage the server yourself.
Tier 4
Managed Cloud
AWS, Microsoft Azure, Google Cloud. Scale-stage SaaS and enterprise. Widest capabilities, hardest pricing to predict, assumes an infrastructure engineer on your team.
Tier 5
Your Own Server (On-Prem)
A physical server in your building or a colocation facility. Rare answer, right for strict data-sovereignty, compliance, or legacy-system needs that the cloud cannot meet.
How to Sort Yourself
Ask: what does your product actually do, how much traffic does it get, and how much operational work can your team absorb. Most small business sites are honestly tier 1. Most modern web apps sit at tier 2 or tier 3. Only a small share of products genuinely need tier 4, and even fewer need tier 5.
Why does the tier matter so much before you sign up? Because a tier-1 product hosted on tier 4 is spending more per month than the product will ever earn back in convenience. A tier-4 product hosted on tier 1 will fall over the first time real traffic arrives. And a tier-2 product hosted on tier 3 is fine on the bill but eats your team's time every week for work the platform host would have done automatically. Getting the tier right is the single decision that decides whether the next year of hosting feels invisible or feels like an ongoing tax.
3 Questions That Decide Where Your Product Should Live
Once you know the tiers exist, how do you honestly pick between them? The 3 questions below are the ones that keep separating the right answer from the loud answer. Every other input (fees, dashboards, developer experience, brand recognition) is downstream of these 3.
01
What Does Your Traffic Actually Look Like?
Steady, predictable traffic (a small business site, a private tool, a mid-sized SaaS with a stable user base) fits a fixed-price tier cleanly. Spiky traffic (viral content, launches, seasonal peaks, event-driven products) either fits a platform host that scales for you automatically or a well-sized VPS with headroom. Bursty and hard-to-predict traffic on the wrong tier is where the surprise bills live. Ask honestly what your product's traffic looks like across a normal quarter, not just on launch day, and pick the tier whose pricing shape matches that pattern.
02
How Much Operational Work Can Your Team Actually Absorb?
Every tier trades money for operational load. Shared hosting and platform hosts do almost everything for you; you pay more per unit of computing power in return. A VPS or your own server saves money per unit but only if your team has the skills and the time to keep it patched, backed up, monitored, and secure. Managed cloud sits awkwardly in the middle: it gives you the tools but assumes you know how to use them. If your team's time is going into product work rather than ops work, that is a design choice, not an accident. Pick the tier whose operational load matches how much of your team you want touching servers.
03
Where Are Your Customers Physically?
A product serving mostly one region is happier on a host with a strong presence in that region. A product serving customers across multiple continents needs a host with global reach or a content-delivery layer in front. Latency is a real user-experience feature: a page that loads in one region and crawls in another is a page half your customers do not use. Pick a host whose region map matches where your customers actually pay from, and if you serve multiple continents, plan for a content-delivery layer from day one, not as an emergency later.
The Order Matters
Answer the traffic question first, the operational-load question second, the geography question third. Teams that answer them in reverse order often pick a globally distributed managed cloud setup for a product whose traffic could have fit on a single VPS in one region for a fraction of the price.
4 Things That Quietly Break Hosting Decisions After Launch
What actually goes wrong with hosting decisions once they are live? The 4 below show up in almost every hosting choice that felt fine on day one and started causing arguments a few weeks in. All of them are survivable if the decision was made with them in mind. Skipping them at launch is why so many hosting choices end up as a rebuild conversation.
01
Traffic Spikes Turn Into Bills You Did Not See Coming
Your product gets picked up by a newsletter, a competitor tweets about it, a video goes viral, a marketing campaign hits harder than planned. On a platform host that scales automatically, that spike is a bill. On a VPS or shared plan, that spike is a downtime. Neither is a surprise if you designed for it. Both are unpleasant if you did not. Ask what happens on your host when traffic goes 10 times normal for a day; if the answer is "we will find out", you have not planned for it yet.
02
Egress Bandwidth Is the Bill Nobody Reads the Fine Print On
Every byte your host sends back out to your customers is bandwidth you pay for on most cloud tiers. Video, large images, downloadable files, and heavy pages compound this fast. The managed cloud giants charge the most for this; shared hosting and many VPS providers include a generous amount for free; platform hosts sit in the middle. Products that serve heavy content on a tier with expensive egress are why the mysterious bill increases happen. Look at your product's real bandwidth pattern before you pick a tier, not after.
03
Regional Mismatch Bites Once You Expand
You hosted in one region because that was where your first customers were. You expanded to a new market and your new customers experience the product as slow, laggy, or unreliable, while your old customers still see it as fine. The host that fit you well in one geography does not automatically fit the next. Plan the second region into the design before you enter it, not after the reviews start coming in from customers you cannot yet talk to.
04
Provider Outages Land Right When You Cannot Afford Them
Every host, every tier, goes down sometimes. Shared hosts have quiet outages you may not even notice. Platform hosts have public outages that hit thousands of products at once. Managed cloud has region-wide outages that make the news. The right question is not "will my host ever go down"; the right question is "when it does, how bad is that for my customers, and can we route around it". Products with a real backup plan for their host survive outages cleanly. Products without one discover on the day of the outage that they never had one.
Why These Feel Sudden
None of these 4 arrive with a warning email. All of them arrive as a bill you did not expect, a customer complaint from a region you just entered, or a downtime that lands during your biggest week. Teams that designed for them at launch survive them cleanly. Teams that did not are the ones doing the hosting-rebuild conversation later.
A Simple Shape That Lets Your Product Move Between Hosts Without a Rewrite
So what does a product that can move between hosts actually look like? Not fancy. The shape below is the arrangement that keeps your product code from ever being tied to a specific provider's quirks, so that moving between shared hosting, a platform host, a VPS, and managed cloud is a configuration change instead of a rewrite. Every layer has one job. When the host changes, the change stays contained.
Architecture
A Shape That Keeps Your Product Portable Across Hosting Tiers
Layer 1
Portable Container
Your product runs inside a container (a self-contained package like Docker) so it behaves the same on any host that can run containers.
Layer 2
Infrastructure as Code
The setup that puts your product on a host lives as code (Terraform, Pulumi, Ansible), so switching hosts is a config change tracked in your repo, not a manual click-fest.
Layer 3
Provider-Neutral Interfaces
Storage, queues, cache, and email go through interfaces your product owns, with adapters for each provider. Swapping S3 for another object store is an adapter change, not a product change.
Layer 4
Content Delivery Layer
Static assets and heavy content sit behind a content-delivery layer that is independent of the host, so regional expansion does not require moving hosts.
↓
The Payoff
What This Shape Actually Buys You
Move Between Tiers
Start on Vercel, move to a VPS when the bill compounds, move to AWS when the product genuinely needs one of its services.
Escape Provider Lock-In
The provider changes policy, raises prices, or ends a service. You swap adapters, you do not rewrite the product.
Add Regions Cleanly
A second region is a new container running the same product, not a parallel codebase managed separately.
Why This Shape Survives
Every layer is one job. Moving hosts touches infrastructure-as-code plus adapters, not the product code. Regional expansion touches the content-delivery layer plus a second container. Provider lock-in stops being permanent; it becomes a swap you can do when the numbers say so.
Why go to this trouble if the current host is fine today? Because hosting decisions are one of the few product choices that compound in cost every month, and they get harder to undo the longer you wait. A product that could have moved off Vercel in a bounded piece of work at year one becomes a product that cannot move off Vercel at year three because "payment" and "auth" and "analytics" and 30 other decisions are all quietly assuming the platform's behaviour. The portable shape is what keeps the option open.
3 Signs You Should Move Up to the Next Tier
When is it time to leave the tier you are on? The 3 signals below are the ones that keep showing up. If you see more than one at a time, you have probably outgrown your current tier and the move-up conversation is already overdue.
01
Your Monthly Bill Now Rivals a Full Server at the Next Tier Down
Your platform-host bill is now higher than a dedicated VPS with plenty of headroom would cost. Your managed-cloud bill is higher than running your own set of servers with a partner would cost. This is the moment the buy-versus-run math has tipped. The current tier is charging you a premium for convenience that a cheaper tier could match if your team is ready to do a little more of the ops work.
02
The Features You Now Need Are All "Coming Soon" or Priced Separately
You started on this tier because it was simple and covered what you needed. Now your product needs a specific queue, a specific storage type, a global content-delivery layer with rules, or an integration your current host does not offer. Every request comes back as "coming soon", a paid add-on, or a workaround the community shares on forums. That is the tier telling you it has stopped growing with your product.
03
Your Team Is Doing Ops Work Instead of Product Work
On a VPS or your own server, if your team is now spending real hours per week keeping the server patched, watching monitoring dashboards, and firefighting instead of building product features, the tier is costing you more than the bill implies. Moving up to a platform host or a managed cloud service (where the provider does that ops work as part of the fee) is often cheaper once you count the salary hours the ops work is quietly eating.
The Honest Test
If more than one of the 3 signals above is true, the move-up conversation is worth having now, not later. Every month you spend on the wrong tier is money the wrong tier is quietly taking, or team time the wrong tier is quietly consuming. The move is real work; ignoring the signals is more expensive work spread across every month you stay put.
Frequently Asked Questions
Is shared hosting really enough for a real business site?
For most small business sites, brochure sites, WordPress installs, and small e-commerce shops, yes. The reason shared hosting has stayed cheap for so many years is that it genuinely fits most of what small businesses actually need. It becomes uncomfortable only when your traffic outgrows the shared resources, when your product needs a database or runtime the shared host does not support, or when uptime and support responsiveness matter enough that the shared host's basic tier stops being acceptable. If none of those apply to you, shared hosting is not just "good enough"; it is the honest answer.
Is Vercel really that expensive at scale, or is that just internet noise?
Vercel and similar platform hosts are cheap at small scale and get expensive at large scale in specific ways. Serverless function invocations, edge bandwidth, image optimization, and content delivery all cost more per unit than a VPS or a managed cloud equivalent. For a product doing modest traffic, the price is fine and the convenience is real. For a product doing high traffic, serving lots of images or video, or with unpredictable spikes, the bill compounds fast. The right question is not "is Vercel expensive"; the right question is "does my product's traffic shape fit Vercel's pricing shape". For a lot of products the answer is yes for a long time; for some it is not, and moving off later is more work than picking a different tier first.
Does your team need dedicated infrastructure engineers to run on AWS?
Not always, but often, yes. AWS gives you almost every capability imaginable and a control surface that assumes you know what you are doing. Small teams that run on AWS without dedicated infrastructure expertise usually either overprovision (paying too much for capacity they do not use), underprovision (running fragile setups that break under load), or misconfigure security in ways that only become obvious after an incident. If your team does not have someone whose job includes AWS, a tier below AWS is almost always the honest choice, or a managed service on top of AWS that handles the infrastructure decisions for you.
What about Cloudflare? Where does it fit in these 5 tiers?
Cloudflare is not one tier; it sits in front of whichever tier you pick. Its role is the content-delivery layer plus security features (rate limiting, bot protection, distributed-denial-of-service protection) plus, more recently, its own platform-host offering (Cloudflare Workers, Pages, R2 storage) that competes with Vercel and Netlify. Almost every product benefits from putting Cloudflare in front of the host, regardless of which tier the host itself is. If you also want to run your product code on Cloudflare's own platform, that is a separate decision and it fits into tier 2 (a platform host) with a different pricing shape than Vercel.
How much does the database decision depend on the hosting decision?
More than most founders expect. Shared hosting usually bundles a database (MySQL, PostgreSQL) at low or no extra cost. Platform hosts either run their own managed database (which is convenient but often expensive) or expect you to bring one from a third party (Supabase, Neon, PlanetScale). Managed cloud gives you every database option under the sun with pricing to match. A VPS lets you run any database you want and manage it yourself. Picking a database that is not aligned with your hosting tier is one of the more common ways hosting bills quietly grow, because the database ends up being the expensive part while the compute cost stays predictable.
Should you host in India, the US, or Europe? Does it really matter?
It matters more than most founders think. Latency (the delay between a customer's browser asking for something and getting a response) shows up as a slow-feeling product for anyone far from the host, and slow-feeling products lose users. Hosting in the region where most of your customers live keeps them happy. Hosting elsewhere and putting a global content-delivery layer in front is the middle path that works for products with genuinely global audiences. Data-protection rules also come into play; some regions require customer data to be stored inside their borders, and picking a host that supports the right regions from the start is easier than migrating later. If you are unsure, host where your first thousand customers are and add regions when the second and third markets earn them.
Can Entexis help you pick and set up the right hosting for your product?
Yes. Entexis designs and runs hosting for custom software products where the hosting decision needs to fit the product rather than the other way around. That work starts with the sorting conversation (which of the 5 tiers is honest for your product today, and which one it is likely to grow into), reviews your traffic shape, your team's operational capacity, and where your customers live, then designs the portable shape (container, infrastructure-as-code, provider-neutral interfaces, content-delivery layer) that keeps you free to change hosts when the numbers say so. We also handle the actual migration when a move up or across is the right call. Reach out with what your product does, roughly how much traffic it sees today, and where your customers live, and we can walk through what the right hosting looks like for your specific product.
So where does that leave you? Most "we need better hosting" conversations end with a different answer than the one the founder walked in with. Sometimes the answer is stay put on shared hosting because it genuinely fits. Sometimes the answer is move from a platform host to a VPS because the bill grew past what the convenience is worth. Sometimes it is the opposite: move from a VPS to a platform host so your team can stop babysitting servers. And sometimes it is a considered move into managed cloud because your product finally needs a service only the giants provide. Get the tier honest first, put the portable shape in place from day one, and the next hosting change becomes a swap instead of a rewrite. Teams that do this find their hosting fades into the background as an invisible cost that scales with the product. Teams that do not find it becomes a recurring line item that grows faster than the product does.
Want Hosting That Fits Your Product Instead of the Loud Answer?
At Entexis, we design and run hosting setups for products across every tier from shared hosting to your own server. We start with the sorting conversation to make sure you are on the right tier for where your product is today and where it is heading, design the portable shape (container, infrastructure-as-code, provider-neutral interfaces, content-delivery layer) that keeps you free to move when the numbers say so, and handle the migration when a tier change is the right call. Your monthly bill matches what your product actually earns from the hosting, your team stops firefighting infrastructure, and your product moves at product speed instead of ops speed. Start the conversation with Entexis.
Need Custom Software Built?
From web apps to enterprise platforms, we build software that fits your workflow, not the other way around. Tell us what you need.
We'll get back within one business day.
Thank You!
We've received your message and will get back to you within one business day.
