AWS Cost Explorer, Honestly: What It Shows and Why Needing It Is a Warning Sign

The Day I Met Cost Explorer (And Wished I Hadn’t)

It was 2022. I was on a client project — not one of my own, one of the ones that actually paid me — and the whole thing lived on AWS. Not my choice. It was there before I arrived, the way AWS usually is, like asbestos in an old building.

One morning the finance person messaged the channel. The bill for the month was tracking about 40% higher than the last one, and nothing had shipped. No new features, no traffic spike anyone knew about. Just a number climbing quietly while everyone slept. The instruction I got was four words: “Open Cost Explorer, find it.”

So I opened it. And I want to be honest with you, because this whole site is built on honesty about tools: AWS Cost Explorer is genuinely good at the one job it has. I found the leak in about twenty minutes — a NAT Gateway someone spun up for a one-off migration and never tore down, sitting there charging by the hour and by the gigabyte for traffic that didn’t need to exist. Twenty minutes to find it. But here’s the thing that stuck with me, the thing this whole post is really about: I only needed those twenty minutes because we were on AWS in the first place. On my own projects, that leak was structurally impossible.

Let me teach you the tool properly. Then let me tell you why I hope you never have to use it.

---

What Cost Explorer Actually Is

Cost Explorer is the dashboard AWS gives you to look at your own bill. That’s it. It’s not a cost cutter, it’s a cost viewer — a set of charts and filters sitting on top of your billing data, showing you where the money went and, with some ML, where it’s probably going next.

The good news, and it’s real good news: the Cost Explorer user interface is free. You log into the AWS console, open it, and browse your spend at no charge. AWS updates the data at least once every 24 hours, so you’re looking at yesterday’s truth, not last month’s. You can slice about 13 months of history, and since a late-2025 update the forecast now reaches 18 months out. For a tool this central to not getting financially ambushed, free is the correct price, and AWS got that part right.

What you’re looking at is a big stacked chart of cost over time, plus a pile of filters and a “group by” control. That group-by is the whole game, and most people never touch it properly.

---

How to Actually Read It

The mistake everyone makes is staring at the total. The total tells you nothing except “too much.” The skill is in the breakdown.

Start by changing Group by. Group by Service and you’ll instantly see whether it’s EC2, S3, RDS, data transfer, or some Lambda thing eating your money. Nine times out of ten, one or two services are 80% of the bill, and one of them is doing something you didn’t intend. Then group by Usage Type to go one level deeper — “EC2” is a category, but “EC2 NAT Gateway data processing” is the actual culprit with a name.

A few controls worth knowing:

• Granularity — monthly, daily, or hourly. Monthly is for board meetings. Daily is what you want when hunting a leak, because a runaway cost shows up as a step change on a specific date, and that date is your clue. (Hourly granularity exists too, but it’s a paid add-on — more on that irony below.)
• Filters — narrow to one linked account, one region, one tag. If you tag your resources by project or environment, you can answer “how much does staging cost me?” in one click. If you don’t tag anything, this is where you learn why people beg you to tag things.
• Forecast — Cost Explorer will project your month-end and future spend from your trend. Useful for a gut-check, not gospel. It assumes the future looks like the recent past, which is exactly the assumption that breaks the moment a bot finds your endpoint.

That’s genuinely most of it. Group by service, drop to daily, find the date the line jumped, filter down to the usage type, go turn that thing off. It’s a good tool. I’m not being sarcastic. It did its job in twenty minutes.

---

Budgets and Alerts: The Part That Matters More

Cost Explorer is reactive. You open it after you suspect something’s wrong. What you actually want is to be told before you’d ever think to look — because the whole horror of a cloud bill is that it grows while you’re not watching.

Two free tools do this, and you should turn both on the day you touch AWS.

AWS Budgets lets you set a number — say, $50 a month — and get emailed the moment your actual or forecasted spend crosses a threshold you pick (50%, 80%, 100%, whatever). Monitoring and notifications are free. You only start paying if you use “action-enabled” budgets that automatically do something when tripped — and even then your first two are free, with additional action-enabled budgets running $0.10 per day each. For a solopreneur, plain alert budgets cover you and cost nothing. Set one. Today.

AWS Cost Anomaly Detection is the smarter sibling, and it’s completely free. It uses machine learning to learn your normal spending shape and pings you when something deviates — often within hours, instead of the weeks it’d take you to notice on an invoice. This is the one that would’ve caught my NAT Gateway on day one instead of at the finance person’s mercy three weeks in. It’s free. There is no reason not to have it on.

If you want the alarms to fire into a dashboard, CloudWatch billing alarms are the older, clunkier way to do the same thing. (CloudWatch is also AWS’s whole monitoring product — metered per metric and alarm, as I break down in AWS monitoring.) Budgets superseded most of that. Start with Budgets and Anomaly Detection.

---

The Cost Tool That Costs Money

Now the part I find genuinely funny, in the dark way that AWS is funny once it’s not your money on the line.

The Cost Explorer UI is free. But the moment you want to pull that data programmatically — to build your own dashboard, to feed a Slack bot, to automate anything — you hit the Cost Explorer API, which charges $0.01 per request. One cent per call.

A penny sounds like nothing until you understand pagination. If your query returns a big result set, the API hands it back in pages, and each page is a separate billable request. A script that naively loops through paginated results across a few accounts, running every hour, can quietly rack up its own line item. Custom billing views make it worse — you’re charged $0.01 per source in the view, so a view stitched from five accounts is five cents a request.

Read that back. The tool for controlling your costs has a meter on it. To programmatically watch your bill, you pay per look. And that same hourly-granularity view I mentioned earlier? Also a paid add-on — billed at $0.00000033 per usage record, which rounds to about $0.01 per 1,000 records a month. Small, sure. But it’s the shape of the thing that matters: on AWS, even the smoke detector runs on a coin slot.

This is not me dunking on AWS for sport. It’s the single cleanest illustration of the entire AWS mindset. Everything is a meter. Everything, including the meters.

---

The Hard Truth: If You Need This, You’re Already Too Deep

Here’s the part I really want you to sit with.

I’ve just taught you Cost Explorer honestly, because knowledge is power and you might inherit an AWS account like I did. But step back and look at what we’re actually doing. We’re building tooling, setting alarms, writing anomaly detectors, and paying per API call — all to answer one question: is my infrastructure about to bankrupt me in my sleep?

That question only exists because the underlying meter can run away. AWS bills you for what you use, with no ceiling, forever. There is no “stop” button unless you build one. Cost Explorer, Budgets, Anomaly Detection — this entire apparatus is a fire brigade for a building made of matches. It’s impressive engineering. It’s also a tax on a choice you didn’t have to make.

The Hard Truth: the best cost control is the cost control you don’t need. If you’ve built a whole watchtower to guard against your own hosting bill, ask why the bill can spike in the first place.

On my own projects, I don’t open Cost Explorer, because there’s nothing to explore. My frontend sits on Vercel with a hard billing cap set — the one setting that turns a usage-based meter into something with a ceiling. My backends run on flat-fee boxes: a $5 Railway instance, a predictable Koyeb machine, a Fly.io app. Bandwidth-heavy stuff goes behind Bunny, which starts at a dollar a month. My database is Neon on a free tier that pauses when idle. When I need “a real server that likes developers” at scale, it’s DigitalOcean, not AWS. (If the AWS free tier is what tempted you here, read what it actually includes in 2026 first — it’s not what it used to be.)

Every one of those is a machine where the number I pay is roughly the number I agreed to. A bot can hammer my Railway box all night and the bill at the end is still $5, because I rented the box, not the electricity. There’s no NAT Gateway to forget about. There’s no anomaly to detect, because there’s no meter to run away.

That’s not me being cheap for its own sake (okay, partly it is). It’s that a fixed bill is one less thing that can wake you up at 3 AM. I wrote the whole tier list of where I actually deploy in 7 cheap places to deploy your SaaS, and AWS is firmly on the “just don’t” list — not because it can’t do the job, but because doing the job requires exactly the kind of vigilance we just spent a whole post learning. The same trap, scaled up, is what makes an observability tool like Datadog eat a bootstrapper alive: per-GB billing with no natural ceiling.

---

The Bottom Line

If you’re already on AWS — a client project, a job, a startup that made the choice before you arrived — then use Cost Explorer well. Group by service, drop to daily, find the date the line jumped. Turn on a Budget alert and Cost Anomaly Detection the same afternoon; they’re free and they’re the difference between finding a leak in hours versus on a statement. And if you ever automate against the API, watch the pagination, because the cost tool has a cost.

But if you’re starting fresh, building your own thing on your own dime? The real lesson of Cost Explorer isn’t how to read it. It’s that a stack designed so you’ll never need to open it is a better stack. Put your money on flat-fee platforms where the meter physically can’t sprint, cap the one place that can (Vercel), and spend your attention on distribution instead of on watching a graph of your own potential ruin. I broke down exactly what a lean SaaS actually costs per month — and Cost Explorer isn’t on the list, because AWS isn’t.

The cheapest AWS bill is the one you never started.

This is the Broken Engineer Guide — I over-engineer everything, fail at business, and share the scars so you can skip the part where finance messages you at month-end. Go build something with a ceiling on it.