Why Hosting Clients Leave — and How Much Churn Is Actually Normal

The email is usually polite. They’re moving the site, thanks for everything, what do they need to do. Nothing went wrong. You didn’t miss a renewal or lose a backup, and you can’t think of a complaint in the last year.

Most of the churn advice online was written for software subscriptions. It tells you to reduce friction, send a win-back sequence and watch your health scores, none of which describes what actually happens to a book of hosting clients. So here is a different starting point. Most hosting clients don’t leave because of the hosting. They leave because the business closed, the site was retired, or the relationship with whoever built it ended — and the hosting went with them. The part you can influence is smaller than the churn guides suggest, and it’s concentrated in a handful of predictable moments.

This page is about which departures are which, when they arrive, and how to measure your own rate properly rather than borrowing someone else’s. Churn is the assumption that decides whether a hosting business compounds or stalls, and how to start a reseller hosting business shows where it sits in the rest of the model. If you have fewer than ten clients, the number that matters right now is how many you win, not how many you lose — finding your first hosting clients is the page for that.

The short answer

1. Across our reseller base, the median reseller loses about 18% of their client accounts in a year. Your number will be different, and after two years of your own history it’s the better one to use.
2. Most of that isn’t about hosting. Businesses close, sites get retired, and design relationships end — and the hosting goes with them.
3. New clients leave far more often than old ones. The first renewal is the riskiest date on the calendar.
4. Under about thirty clients, your churn percentage is mostly noise. Count departures and name their causes instead.
5. The churn you can reduce clusters around four moments: the first renewal, the redesign, the day something breaks, and the day your contact changes jobs.

What counts as churn in a hosting book

Churn is the share of clients you had at the start of a period who have gone by the end of it. That part is simple. Three things make it less simple for a hosting book.

Clients, not accounts — and revenue, not just clients

Count clients, not cPanel accounts. A client with three sites who closes a campaign microsite hasn’t left you. A client who drops a care plan and keeps paying for hosting on its own hasn’t left either — that’s a downgrade, and it’s often the best available outcome when a retainer ends.

Revenue churn and client churn also move apart. Losing your two biggest accounts in a year can leave your client churn looking healthy while your income takes the hit, so it’s worth glancing at both.

And some clients leave without meaning to. A card expires, the payment fails, the reminder goes to an inbox nobody reads, and the account is suspended. That’s involuntary churn, it’s usually the most recoverable kind, and non-payment, suspensions and getting paid covers how to catch it.

Why small books can’t trust their own percentage

Here is what one departure does to the rate at three book sizes:

At fifteen clients, the difference between a good year and a bad one is a single phone call. It cuts the other way too. With twenty clients and an underlying rate that never changes — say, 15% — ordinary chance will hand you anywhere from one departure to six in a given year. That’s a “churn rate” of 5% one year and 30% the next, from a business that didn’t change at all.

This is arithmetic, not a finding, and it has a practical consequence: with a small book, look at two or three years together before you conclude anything, and pay more attention to why people left than to how many did.

On benchmarks

You will find hosting churn figures quoted online. Almost none of them state a method: not what was counted, not over what period, not whether “customer” meant a billing account or a website. The usable ones come from software subscription businesses, where the product, the price point and the switching cost are all different from a hosting book sold by the person who built the site.

That’s the honest position, and it’s why this page doesn’t open with a number. A benchmark you can’t check is worse than no benchmark, because it gives you a false sense of where you stand. The method further down produces something better in ten minutes.

Why clients actually leave

Most of the top half of that table is life happening to a small business. The bottom half is where the work is, and three of the rows need more than a line.

“Something broke” is about the handling, not the incident. Clients forgive downtime they were told about. They leave over downtime they discovered — from a customer, at the worst possible moment, with no word from you. The difference is almost entirely communication, and handling downtime: a communication plan is how to have it ready before you need it.

The person who hired you leaving is the cause almost nobody writes about and one of the most preventable. The marketing manager who chose you moves on, their replacement arrives with a favourite agency, and within a year the site — and the hosting — is somewhere else. You usually find out from an out-of-office reply. Keep two named contacts at every client, and treat a change of contact as a reason to introduce yourself properly rather than as an admin update.

Moving to a platform isn’t a hosting failure. A client whose needs now fit a hosted builder may simply be right to move. Arguing costs you goodwill that is worth more as a referral than the account was as revenue. The point about departures arriving in clusters applies here too: when a builder gets good at one kind of business, several of your clients in it can move within months.

None of this means churn is out of your hands. It means a meaningful share of every book’s churn isn’t a retention problem at all, and a reseller who treats every departure as a failure will spend effort in the wrong places — usually on discounts for clients whose needs have already moved on.

When they leave

The causes tell you what to fix. The timing tells you when to pay attention.

The first ninety days. Expectations that didn’t match, an onboarding that felt like an afterthought, or a client who was always going to be the wrong fit — including the one who wanted $3 hosting and was never going to stay at your price, which finding your first hosting clients calls the buyer who doesn’t exist.

The first renewal. The first time a client sees the full price on its own invoice, and the first natural moment to ask whether they still need it. A renewal notice that arrives two to four weeks early and says what they got this year does more work here than anything else on this page.

The redesign. When a site is due for a rebuild, the client often puts the design work out to tender, and the hosting goes with whoever wins. For designers this is the largest single exposure in the book, and it arrives on a schedule you can roughly predict from the launch dates in your own records.

After an incident, and after a contact change. You can’t predict either date. You can predict the effect, and both are covered above.

If you’re a designer: hosting churn follows the relationship

If most of your hosting clients came with a website you built, your hosting churn is mostly a lagging indicator of your design churn. The site gets rebuilt by someone else, the retainer stops, a new marketing lead arrives — and the hosting leaves as a consequence rather than a cause.

That changes what’s worth doing. Two things, specifically.

Offer the hosting-only ending before the retainer ends. A client cancelling maintenance they weren’t using often still wants the site they depend on to stay up. If you’ve already named a hosting-only price, a cancellation becomes a downgrade. Hosting inside a retainer vs billed separately sets out the three endings and why to decide on one in advance.

Decide now what happens when the next agency takes over. Some incoming agencies are happy to leave the hosting where it is, especially if moving it is their problem. Make that the easy option — clear access, a named contact, no friction — and you keep recurring revenue from a client you’ve otherwise lost. The reverse case, where you’re the one taking over, is in telling existing clients you’re taking over their hosting.

This isn’t a loss to prevent at all costs. It’s a transition you can price before it happens.

If you run a larger book: churn at scale, and during a switch

Read your book by cohort, not as one number

A book that has been growing for ten years produces a comfortable-looking annual churn rate, because most of its clients are old and settled. Underneath, this year’s new clients may be leaving at a noticeably higher rate, and the blended figure hides both a good retention story and a weak acquisition one.

Group clients by the year they joined and look at each year’s survival separately. If your 2023 clients are still around at the rate your 2019 clients were at the same age, your book is healthy. If they aren’t, you have an onboarding or fit problem that the overall number won’t show you for years. The curve earlier on this page is the shape to compare against.

Churn is also the single biggest driver of what a book is worth to someone else. A buyer is paying for the clients who stay, and selling a hosting book of business covers how that shows up in the price.

The switch is a churn event — plan for it

Changing provider asks every client to notice you, at least a little. Anything that changes on their side — nameservers, a login address, a password, how they pay — is a moment when some of them reconsider whether they need you at all.

The rule that follows is simple: minimise what the client sees. Move accounts without asking clients to do anything where you can. Keep changes away from each client’s renewal date. And tell them before anything visible changes, never after. The mechanics are in migrating client sites to your reseller account without downtime, and how to word the message is in telling existing clients you’re taking over their hosting.

The migration itself rarely causes the loss. Visible change does.

How to measure your own churn in ten minutes

You need your billing system and a list.

1. Export your active clients at the start of the period from WHMCS, Blesta or whatever you invoice from. Clients, not services or accounts.
2. Mark which of them were no longer active clients at the end. Leave out anyone who joined during the period — they’re neither in the starting count nor among the departures. Including them is the most common way to get this wrong.
3. Write a cause next to each departure, using the table above. If you can’t name the cause, that’s a finding in its own right.
4. Divide departures by the starting count. If you have fewer than thirty clients, also write down the raw number and look at the last two years together.
5. Put your number into the reseller hosting profit calculator in place of the default, and see what it does to year three.

Two examples, both invented for illustration. A designer starts the year with 20 hosting clients and loses 3: a 15% rate, which on the arithmetic above is well within the range of an ordinary year and says nothing on its own — the three causes say far more. A small host starts with 150 and loses 18: 12%, and at that size it’s a real measurement worth comparing year on year.

Converting monthly to annual. If your billing system reports churn monthly, don’t multiply by twelve. Work out how many clients you keep each month, compound that over the year, and subtract from one. At 2% a month, you keep 98%; 98% compounded twelve times is about 78.5%, so annual churn is about 21.5%, not 24%. The gap widens as the rate rises — at 3% a month it’s roughly 31% a year, not 36%.

Reducing the churn you can control

Grouped by the moment it matters:

Before the first renewal

• [ ] Send the renewal notice two to four weeks ahead, and say what they got this year, not just what they owe.
• [ ] Set up card-expiry reminders and a failed-payment sequence someone actually reads — non-payment, suspensions and getting paid has the method.

Around a price change

• [ ] Decide the new price and the reason before you write to anyone. Raising hosting prices on existing clients shows how to check what you can afford to lose first.

When something breaks

• [ ] Tell affected clients before they notice.
• [ ] Know exactly what you can restore and how quickly — backups, and who’s responsible when a client deletes their site. 
• [ ] Have a plan for a compromised site before you have one — when a client site gets compromised.

In the relationship

• [ ] Keep two named contacts at every client, and re-introduce yourself when one changes.
• [ ] Name a hosting-only price before any retainer ends.
• [ ] Note each site’s launch date, and get in touch before it’s due for a rebuild rather than after.
• [ ] Ask every departing client one question — “what would have made you stay?” — and record the answer against the cause.

Two things not to do. Don’t discount to keep a client whose business has moved somewhere else; it delays the departure and costs you margin in the meantime. And don’t make leaving harder, which is the next section.

Letting clients leave well

A client who leaves smoothly refers you, sometimes comes back, and occasionally recommends you to the next agency’s client. A client who has to fight their way out tells people about it.

Three rules. Cancel on the date they give you. Hand over backups and domain transfer codes promptly, without being chased. And never hold anything they own — a domain, their files, their email — as leverage over an unpaid balance or a notice period; settle those on their own terms. What your terms should say about exits is in terms of service and AUP for a small host.

Clean exits don’t reduce this year’s churn. They reduce what it costs to replace the client, and over a few years that matters as much.

Mistakes worth avoiding

• Using someone else’s churn number when you have two years of your own.
• Trusting a twelve-month rate on fewer than thirty clients.
• Multiplying a monthly rate by twelve.
• Counting removed accounts as lost clients.
• Treating every departure as a service failure — or as a fact about the market.
• Missing involuntary churn because nobody reads the failed-payment emails.
• Discounting to keep clients whose needs have moved on.
• Making leaving difficult.

Frequently asked questions

What is a normal churn rate for a hosting business? There’s no benchmark worth planning around. The figures quoted online rarely state what was counted or over what period, and the reliable ones come from software subscriptions rather than hosting books. Measure your own over two years and compare each intake year to the last — that’s a number you can act on.

How do you calculate churn for a hosting business? Divide the number of clients who left during the period by the number you had at the start, leaving out clients who joined during it. To turn a monthly rate into an annual one, compound the monthly retention rate rather than multiplying the churn rate by twelve.

Why do web hosting clients cancel? Most often because the business closed, the site was retired, or the relationship with whoever built it ended. Price rises, failed payments and badly handled incidents account for a smaller share, but they’re the part a host can do something about.

Does better hosting reduce churn? It reduces the part caused by incidents and how they’re handled, which is real but not most of it. The larger levers are renewal notices, catching failed payments, and staying close to clients through a redesign or a change of contact.

Your churn rate is the assumption in the profit model most likely to be wrong, and the easiest to correct. Put your own number into the reseller profit calculator and see what it does to year three.

If you’re moving a book of clients across, most of the churn risk is in what those clients see change — and the reseller plans are where the account side of that decision starts.