Uptime 99.9% — How Many Minutes of Downtime Does That Actually Allow per Month?

One month has approximately 43,200 minutes. At 99.9% uptime that allows roughly 43 minutes of downtime per month. At 99% uptime it allows over 7 hours per month. At 99.99% it allows approximately 4 minutes per month. Each additional decimal point matters enormously in practice. When comparing hosting providers, always ask whether the uptime guarantee is written into the contract and what the compensation is if it is not met.

Why Do Hosting Quotes Vary so Much When the Service Seems to Be the Same?

Because 'hosting' or 'VPS' as a label says nothing about what is actually inside the package. A Thai provider at ฿250/month before VAT may include a local-language support team available during business hours, while a DigitalOcean droplet at US$4/month is bare metal you manage entirely yourself. The price difference is a reflection of different service levels, not of one provider being better value. Ask precisely what is included before comparing prices.

My Website Has Never Gone Down in Years — Does That Mean My Hosting Is Good Enough?

Not necessarily. A website that has never been under real peak load has never actually been tested. It is like electrical wiring that has never tripped because nothing was running at full capacity — it may be fine, or it may not be. Check pagespeed.web.dev periodically, even when everything seems to be working. Waiting for a crash is an expensive way to discover the answer.

What Should I Ask Before Signing a Hosting Contract or Hiring a Developer?

At minimum, five things: (1) is the uptime SLA a written guarantee, and what is the exact percentage? (2) What is the compensation if the uptime commitment is not met? (3) How often are backups taken, and how quickly can a backup be restored? (4) What are the support hours — business hours only, or 24 hours? (5) Is the renewal price the same as the first-year price? Without clear answers to all five, you are comparing incomplete offers.

If My Website Goes Down During a Big Promotion, Who Is Responsible for the Lost Revenue?

That depends entirely on the SLA wording. Most hosting providers compensate for downtime proportionally — you get credit for the time the service was unavailable, not for the revenue you lost during that time. Expecting a provider to compensate for lost sales is usually not realistic. This is why prevention matters far more than compensation: notify your hosting team before any major promotion date so someone can monitor the server in real time on the day.

How Can I Tell Whether My Website Is Ready to Handle Promotion Traffic Before It Actually Crashes?

Run a check at pagespeed.web.dev — it gives you both lab performance data and real user data from Google’s CrUX database. If your scores are poor under normal traffic, they will be worse under a traffic spike. Then notify your web and hosting team at least a few days before the promotion so someone is actively monitoring the server on the day itself, rather than waiting for customer complaints to arrive first.

Your Website Crashed on Your Biggest Promotion Day Because Hosting Wasn't Enough: How to Read a Hosting Quote Before You Sign

I have a friend who runs a small restaurant in an old shophouse. The building was wired by the previous owner back when nobody had air conditioning in every room. On ordinary days, it was never a problem — fans, a fridge, an electric stove all running at once and the power held fine. But then came the biggest promotion night of the year. The place was packed, every air conditioner running, multiple electric grills going full blast at the same time. That night, the breaker tripped in the middle of service. The restaurant went dark with customers still sitting at every table. My friend said: "But the power's never gone out before." Which was true. It had never gone out before because there had never been a night where everything in the restaurant was running at full capacity at the same time. The same wiring that handled ordinary days had never actually been tested against the worst-case day.

Hosting for a business website works exactly like that. It's the wiring behind the wall that the website owner never sees. On a day when only a few dozen people visit the site, the cheapest hosting on the market and the most expensive hosting on the market perform identically. The difference only shows up on the hardest day: the day hundreds of people click through from an ad at the same time, the day a post goes viral and gets shared widely, the day of the biggest promotion of the year — like my friend's restaurant.

Test Your Website Right Now — Takes Under 3 Minutes

Before getting into the technical side, open pagespeed.web.dev and paste in your own website URL. This Google tool gives two sets of results that most people don't notice are different from each other. The first is a lab score — a simulated single page load in a controlled environment. The second is real-user experience data — collected from people who have actually visited that site over the past 28 days. The two don't have to match, and hosting is one of the factors that can cause them to diverge — though not the only one.

This article won't repeat the basics we've already written about elsewhere. If you don't yet know what hosting is, or how shared hosting differs from cloud, read our article How to Choose Web Hosting So Your Business Website Loads Fast and Ranks in SEO. And for domain names, which often get confused with hosting, there's a separate article at What Is a Domain Name? Here I'm only going to cover what hasn't been written about elsewhere: when it actually breaks, and what to ask before signing.

What Actually Breaks When Hosting Isn't Enough

Type One: Complete Crash

This is the most visible outcome. The server has a limit on how many simultaneous visitors it can handle. When more people click in than the server can support, the page returns an error instead of showing the products or services. Customers who were ready to buy or were about to reach out to you find an error page instead. Most of them don't come back the next day to try again.

Type Two: Doesn't Crash, but Loads So Slowly That Customers Leave

This one is more dangerous than it looks, because the website owner won't see any error. The site still opens — it just loads far more slowly than usual during periods of high traffic. Customers who click through from an ad or a shared post, if they wait too long, quietly close the tab without the owner ever knowing how many were lost that day. There's no error to see. Just sales figures lower than they should have been.

An example I want to share directly is our own website, TumWebSME. We've invested heavily in performance optimization, and the Lighthouse lab scores on mobile for the homepage, /system, /package, and /blog pages climbed from 51, 52, 62, 66 to 88, 88, 95, 95 respectively. But when we checked real-user experience data on the same day — September 16, 2026 — we found LCP still at 3,780 milliseconds, while the "good" threshold Google sets at web.dev/articles/lcp is under 2,500 milliseconds. INP was at 534 milliseconds versus the good threshold of under 200 milliseconds per web.dev/articles/inp. Only CLS passed cleanly at 0.00. I'm sharing this not to make ourselves look bad, but to show how different a polished lab score can be from what real users actually experience — even on a site that's actively maintained every week.

Why Hosting and VPS Prices Can't Be Compared Directly

Let me share some real figures I checked today, September 16, 2026. One Thai hosting provider, ThaiDNS, lists a Shared Hosting package starting at ฿500 per year for 2GB of storage on an Asian server. Their VPS packages from the same provider start at ฿250 per month, or roughly ฿3,000 per year — six times more from the same provider. See their pricing at thaidns.co.th/pricing.html. Their hosting package page also notes that all displayed prices exclude 7% VAT.

Now compare that to DigitalOcean, a global cloud compute provider. Their smallest Droplet (what they call a VPS) starts at just $4 per month, per their pricing page at digitalocean.com/pricing/droplets. At a rough estimate of 35 Thai baht per dollar for illustration only — not an actual exchange rate or card fee — that's about ฿140 per month, lower than ThaiDNS's VPS starting at ฿250 per month (before 7% VAT). But these two numbers are calculated on different tax bases, so they're only comparable in a rough sense, and a lower number doesn't mean DigitalOcean is better or worse value. They're simply different products. A $4 Droplet gives you a bare server with no Thai-speaking support team, no automatic backup included at that price. If the server has a problem at 2 AM, you fix it yourself or hire someone who knows servers. A Thai provider's VPS price typically bundles basic support and monitoring into the price.

The price difference between two hosting quotes isn't because anyone is overcharging. It's because they're selling different things that happen to share similar names. Two quotes that both say "hosting" or "VPS" can differ by several times without either party being dishonest. The number on a quote never tells you what's inside. You have to ask.

How to Read a Hosting Quote Before Signing

Questions I recommend asking every time before making a decision — regardless of which hosting provider or web development team, including us:

Is the uptime guarantee written down in a document, or just a verbal assurance of "definitely stable"?
If the site actually goes down, what's the compensation? Is it in the contract?
How often are backups made? How many days are they retained? Can you restore yourself immediately or do you have to wait for the team?
What hours can you reach support? If the site goes down at 2 AM, will someone answer?
Is the first-year price the same as the renewal price for subsequent years?

The last one matters more than most people realize — because the attractive rate used to bring you in is often only for year one.

The uptime number itself carries more meaning than vendors usually explain. A quick calculation: one month has about 43,200 minutes. A 99% uptime guarantee means the site is allowed to be down for roughly 7 hours per month. 99.9% means about 43 minutes per month. 99.99% means about 4 minutes per month. Each decimal place added to that number means something much more significant than it looks.

"Never Crashed" and "Can Handle Peak Load" Are Not the Same Thing

I understand that if a website has never gone down in years of operation, it feels like the current hosting must be fine. I won't argue against that — it might genuinely be fine. But like my friend's restaurant breaker that "had never tripped before" — "never crashed" and "can handle any situation" are different things. If the site has never seen a day where that many people came at once, you don't actually know yet whether it would hold up. As I mentioned earlier, even our own site still has real-user data that doesn't pass Google's thresholds, despite ongoing work to improve it. That's why checking actual data periodically matters more than trusting that no problems have come up so far.

This connects to what we wrote in Why Run a Business Website on Cloud? — about why scalable infrastructure matters. The businesses that grow steadily often don't see the traffic spike coming. A big promotion day, a news feature, a viral post — they tend to arrive without warning. The more the infrastructure is ready before that day comes, the less you have to decide anything mid-crisis while the site is on the verge of going down.

What an SME Needs at Minimum — Without Enterprise-Level Infrastructure

An SME doesn't need infrastructure that can handle hundreds of thousands of simultaneous visitors like a large corporation. Three things are enough for most situations: infrastructure that can scale up on its own when traffic spikes, without someone needing to sit and manually add resources; automatic backups with a written schedule and retention period; and a real support contact that has actually been tested and works — not just a phone number in a contract that's never been called. Investing in solid infrastructure from the start is part of the overall cost of owning a website, something we've worked through at a break-even analysis level versus platform fees in our article on the break-even point of owning your own website.

Our own packages start at ฿79,000 as a one-time payment, with custom code on a cloud infrastructure that can scale on its own. If you'd like to see exactly how each package handles these things, the details are on our packages page.

A 30-Day Plan to Check Your Own Hosting

Days 1–3: Test your site at pagespeed.web.dev and take screenshots of both the lab score and the real-user experience data as a baseline.
Days 4–7: Pull out your current hosting contract or quote and check it against the five questions above. Anything without a clear answer — send an email to your provider and ask directly.
Days 8–14: If you have a major promotion or ad campaign coming up soon, let your web or hosting team know in advance — and watch how the site responds on the actual day.
Days 15–21: If backup is not a clearly documented automatic system, manually export your site's data or key content and store a copy yourself.
Days 22–30: Run the pagespeed test again and compare it to the baseline from day one.

How to Keep Monitoring After This

Come back to pagespeed.web.dev once a month — that's enough. What to look for isn't just the lab score going up, but the trend in real-user experience data — whether it's moving toward the "good" thresholds or not. If the numbers stall or worsen without any changes on the website side, that might be a signal worth taking back to your hosting provider.

The wire behind the wall never makes the restaurant run better. But the day it has a problem, it's the only thing anyone will remember.

All of this may sound highly technical, but what I really want business owners to take away is just one thing: "never crashed" doesn't mean "can handle it." The only way to know is to look at the real numbers before the hardest day arrives. If you'd like to talk through your business website — whether you're just starting to think about it or looking to upgrade what's already there — feel free to reach out and tell me what you're working with.