Insights / Hosting
VPS Pricing for New Hosting Providers: How to Avoid Underpricing
Price for support, abuse, and customers who leave — not a copied $4 price list. Build a cost stack, then offer three plans you can actually deliver.
A VPS (virtual private server) is a virtual machine with a set amount of CPU, RAM, and disk that a customer rents, usually by the month. For a new hosting provider, how to price VPS hosting is the first product decision most get wrong. They open a large competitor’s order page, copy the RAM steps, cut a dollar, and publish. That is not a pricing model. It is a hope that someone else’s automation will appear in your ticket queue.
Running VPSGrid — U.S. Linux and Windows VPS — makes one rule hard to ignore: the price on a plan has to survive a real person answering a real ticket. It also has to survive a fraud review, an IP address that gets listed, and a customer who leaves in month two. If your price list cannot absorb those costs, you are not cheap. You have not finished the model.
This is VPS pricing strategy for people building a hosting company, not shopping for a $4 server. If you want a copied catalog, you already know how to do that. If you want a company that is still there in year two, price the work.
A public price list is not a cost model
A public price list is a marketing page. It does not show oversell ratio — how many virtual machines you pack onto one physical server. It does not show how many of those machines are actually sold. It does not show what the provider pays for IP addresses, or how many minutes of support sit behind a $6 plan. Copying it copies the advertisement, not the business.
Your catalog has to match who you can serve. A one-person shop with a helpdesk open 12 hours cannot price like a company with 24/7 chat and a dedicated abuse desk. Customers who need a human will find you. Price as if they will.
Write the buyer before you write the RAM. A developer who wants a cheap lab box is not the same as a small company that needs Windows, a backup they can restore, and someone to answer when mail stops. Those two buyers do not belong on the same “unlimited” story.
If you cannot say, in one paragraph, who the plan is for and what happens when they open a ticket, you are not ready to publish a price. More plans will not fix that.
Build the cost stack on paper this week
Do this before you talk to a datacenter salesperson. One sheet, three plans, every line that will hit you in month one. If a line is unknown, put a conservative number — not zero.
Start with the node or wholesale VM: CPU, RAM, disk, port. Then add what the order form never shows.
- IPv4. Addresses are not free, and they are not optional for most VPS buyers.
- Bandwidth. Included GB plus a realistic overage. “Unmetered” still has a cost. You pay for the port. The customer does not see that line.
- Backups. Snapshot storage, off-node copies, restore tests. A backup you have never restored is a slogan.
- Billing software. Licence, modules, payment gateway fees, chargebacks.
- Support minutes. Tickets, chat, the 40-minute “why is my WordPress slow” on a CMS the customer installed.
- Abuse and fraud. Reviews, WHOIS noise, IP reputation, the afternoon you spend on a spam complaint.
- Churn. Setup time you never recover when they cancel on day 20.
Allocate support and abuse per plan, not as a company overhead you hope volume will wash out. Volume of cheap plans increases tickets. It does not automatically increase skill or hours in the day.
| Cost line | What to estimate | Common mistake |
|---|---|---|
| Compute | Your cost for the VM or slice of the node | Treating list RAM as if it were sold at 100% with no oversell risk |
| IP | Monthly cost per v4, including idle addresses | Giving extra IPs away to win the sale |
| Bandwidth | Included transfer plus a likely overage | Advertising unmetered on a contended port |
| Support | Minutes per customer per month at your loaded hourly rate | Assuming tickets will be “rare” |
| Fraud and abuse | Review time, listings, replacements, chargebacks | Zero, until the first bad week |
| Payments | Processor percent plus fixed fee plus disputes | Pricing as if every invoice is a wire |
Loaded hourly rate means what an hour of your time is worth when you are the support desk, the abuse desk, and the owner. If you would not sell that hour to a client at $X, do not pretend the hosting plan gets it for free.
Finish the sheet for three plans by Friday. If a plan cannot cover support minutes plus a modest margin, raise the price or kill the plan. Do not launch it “to be competitive.”
Three plans, not twelve
Entry, working, and “this customer will actually pay for a human.” That is enough for version one. Twelve SKUs look like a catalog. They are usually twelve ways to confuse a buyer and twelve ways to mis-price support.
The working plan should be the one you want most customers on. Price and name it so the middle is the default. The cheap plan exists so you have an on-ramp — not so you can win a race to the bottom against companies that do not answer tickets the way you will.
Windows is a different product. Licence cost, image maintenance, and the tickets are different. Do not hide it as a checkbox on a Linux price. Reseller is a different business: you inherit someone else’s customers and someone else’s abuse. Do not launch reseller in week one because a WHMCS screenshot had a “resellers” tab.
Annual discounts are a cash tool, not a branding exercise. If you do not have the operations to survive a year of that customer, do not lock the rate. Monthly honesty first. Annual later, when you know churn and ticket load.
Add-ons should be things you can deliver without a project: extra IP, extra backup retention, extra traffic. “Managed” is not an add-on. Managed is a product with a response time and a price that funds it. If you are not staffed for it, do not sell it.
Name plans for the job, not for the hardware. “Lab,” “Site,” and “App” tell a buyer more than “VPS 1 / VPS 2 / VPS 3” copied from a wholesale portal. Hardware specs still belong on the page. They are not the offer.
When cheap is a trap
Cheap is a trap when the customer who buys it needs more help than the margin can buy. That can describe some very low-priced plans, especially when a new company has no automation and a founder answering tickets at midnight.
Cheap is a trap when it trains the market that you are the discount option. The buyers who find you on price comparison charts will leave for a dollar. Your acquisition cost walks out every time a larger brand runs a promo.
Cheap is a trap when it forces oversell you cannot see. You fill the server. Performance falls. Tickets rise. You spend the margin you never had on apologies. A slightly higher price with fewer, quieter neighbors is often the more honest product.
Cheap is not always wrong. A tightly scoped lab plan, sold as a lab plan, with a hard cap on tickets, can be a funnel into the working plan. The trap is pretending lab economics are company economics.
If you are using price to compensate for a weak offer — unclear location, no restore story, no support hours on the site — raise the offer before you cut the number. A clear Dallas Linux VPS with backups and a published support window is easier to defend than a vague “unlimited” at two dollars less.
Too cheap, sustainable, or greedy — check your own catalog
I will not quote a competitor’s live prices as if they were a cost model. Public prices move. What you can do is check your own catalog against three price levels.
| Price level | What it usually means | What to do instead |
|---|---|---|
| Too cheap | You are buying customers you cannot support, or you omitted IPs, abuse, and tickets from the stack | Kill the plan or raise it until a bad ticket week still leaves margin |
| Sustainable | The working plan covers cost stack plus a margin you would still accept if growth is slow | Keep three plans; put operational energy here |
| Greedy | Price far above what your location, support, and reliability can justify, with no story | Add a real difference — support hours, backups, Windows, a region — or come down |
Sustainable is not “as high as the market will bear.” It is “I can still answer this customer in month six.” Greedy without a real difference just sends buyers back to the large provider. Too cheap without automation sends them into your evenings.
Location is a difference. U.S. VPS is a commercial choice, not a moral one. If your buyer needs that region, price it. If you are selling a region you cannot explain, you will compete only on the number in the corner.
Support hours on the website are a difference. “We reply” is not. Write the window. Price as if people will use it.
Churn, abuse, and the customer you should refuse
Churn is a pricing input. If a typical cheap customer lasts two months, your setup and KYC time must be earned in those two invoices. That usually means the cheap plan cannot include hand-holding, custom ISOs, and same-day restores as if they were included forever.
Abuse is a pricing input. Crypto miners on a “fair use” CPU, phishing kits, and buyers who want a host that will ignore complaints will find an underpriced VPS. Your acceptable use policy is not decoration. It decides whether you still want the company after the first wave.
Refusal is a product feature. If you will not host a use case, say so before the order. Every hour you spend arguing with a customer you should not have sold is an hour you did not spend on a customer who pays for the working plan.
Fraud and payment fees belong on the stack because they are not rare. Card testing, stolen cards, and disputes show up when you sell low-friction VPS to the open internet. Stronger checks slow sign-up. That is a trade. Price and process together — do not only cut the advertised number.
Read this next to how to start a hosting company if you do not yet have an offer, and WHMCS for a new hosting company if you are about to encode a bad catalog into a billing panel.
A pricing worksheet you can finish this week
Block three hours. No shopping for servers until the sheet is done.
- Write one sentence per plan: who it is for, and who it is not for.
- List RAM, CPU, disk, port, included IPs, backup policy.
- Fill every cost-stack line above. Unknowns get a conservative number, not a blank.
- Estimate tickets: quiet month and bad month. Convert minutes to money at your loaded rate.
- Set a price. Check contribution after fees. If the bad month is red, raise or cut scope.
- Write support hours and a one-paragraph AUP you are willing to enforce.
- Delete any SKU you would resent supporting at 11 p.m.
If you cannot complete the sheet, you are not missing a datacenter quote. You are missing a product. Get the product on paper. Infrastructure is a later decision. It is a direction, not a shopping list.
I do not install your panel or procure your nodes in a consulting package. Hosting Launch Call ($199) is a 60-minute working session. Hosting Business Blueprint ($499, featured) adds a concise written plan. Hosting Launch Advisory ($999) is that blueprint plus three 60-minute sessions within 45 days. If you need someone to click through WHMCS with you, that is a separate contract.
Book through the consultation page. Fees are non-refundable once paid, unless I cancel and we cannot agree a new date.
Price like you will still be the person answering the ticket. That is the whole method.
Questions people ask
How should a new provider set VPS hosting pricing?
Build a cost stack per plan — IP, bandwidth, support minutes, fraud, payment fees — then add margin you can defend. Three honest plans beat a copied $4 price list.
Why is cheap VPS hosting a trap for new companies?
Low advertised prices attract ticket-heavy customers a giant can serve with automation you do not have. One abuse case or a week of tickets can erase the month.
What costs belong in VPS pricing besides the server?
IPs, bandwidth overage, backups, billing software, payment processing, chargebacks, fraud review, and hours of support and abuse handling. Hardware is often the smallest line.
How many VPS plans should I launch with?
Three: an entry plan you can support, a working plan most customers should buy, and a plan priced for people who will actually open tickets. Windows and reseller are separate products.
Should I match a competitor’s public grid?
Use public grids as a ceiling check, not a cost model. You do not know their wholesale, oversell, support load, or how many of those $4 nodes are actually profitable.
When is a higher price the honest move?
When your support hours, abuse handling, or location premium are real. Customers who need a human should not be on a price designed for self-serve hyperscale.
Does consulting include setting up my billing or servers?
No. Hosting packages are strategy and a written blueprint where included. Server configuration, WHMCS work, and procurement are not in the fee unless separately contracted.
How do I pay for a hosting session?
Book through the hosting consultation page. Fees are non-refundable once paid, except if I cancel and we cannot agree a new date.
Not legal, tax, or accounting advice. Entity, sales tax, and acceptable-use questions belong with licensed counsel in your jurisdiction.