The first time you open an AWS bill, it is genuinely confusing. Dozens of line items, cryptic usage-type codes, numbers that do not match your mental model of what you are running, and no obvious "here is what to do about it." I remember staring at mine trying to figure out why "data transfer" cost anything when I had not transferred anything on purpose. Here is the tour I wish someone had given me: how to read an AWS bill, find the big rocks, and know which numbers actually matter. Do not start on the Bills page. Start in Cost Explorer. The raw Bills page is a flat list, accurate but hard to reason about. Cost Explorer is where you actually understand the bill, because it lets you group and filter. Open it and set the view to the current month. The single most useful thing you can do: group by Service. This instantly shows you the shape of your bill, which handful of services make up most of the cost. Almost every bill is dominated by a few services, and everything else is rounding. The usual big rocks For most accounts, the top of the list is some order of: EC2 (compute): usually the biggest. Instances you are running by the hour. RDS (managed databases): often second, because databases run 24/7 and are easy to over-size. S3 (storage): grows quietly as data accumulates. Data transfer: the one with no resource behind it (bytes moving between AZs, regions, or out to the internet). Surprising the first time because you never "provisioned" it. Load balancers, NAT gateways, and other networking: small each, add up. Knowing the top five is 80% of understanding your bill. Do not try to understand all forty line items on day one. Understand the big rocks first. Decoding the confusing parts A few things that trip up everyone at first: Usage types. Line items have codes like BoxUsage:m5.large (EC2 instance hours) or USE1-DataTransfer-Out-Bytes (egress from us-east-1). The prefix is often the region (USE1 = N. Virginia). Once you know the pattern they stop looking like noise. "Charges" vs Cost Explorer numbers. The Bills page shows what you are billed including credits and tax. Cost Explorer defaults to a specific cost metric (unblended cost). They can differ. For understanding where cost comes from, Cost Explorer is right; for what you actually pay, the Bills page total is right. Credits and Free Tier. If you have credits, your actual charge is lower than the gross usage. Do not panic at a big usage number if credits are covering it, but do watch when credits are about to run out, that is when the "free" thing suddenly is not. The three questions to ask your bill Once you can see the shape, ask: What is the biggest line, and do I know why? If EC2 is the biggest, can you name the instances? If not, that is where to look first. Is anything here I do not recognize? A service you do not remember using, a region you do not operate in. Unrecognized cost is often forgotten or orphaned resources. What is growing? Change the Cost Explorer view to the last few months. A line trending up faster than your usage is the thing to investigate, growth matters more than the absolute number. Turn reading into action Reading the bill is step one. The follow-through: Big EC2/RDS line? Check whether those instances are right-sized (Compute Optimizer, free) and whether non-production ones need to run 24/7 (they usually do not). Surprising data transfer? Group by usage type to see which kind (cross-AZ, NAT, internet egress) and trace it to the architecture. Unrecognized cost? Hunt for orphaned and idle resources, the unattached volumes, idle load balancers, and forgotten NAT gateways that bill for nothing. Set a budget alert. AWS Budgets can email you when spend crosses a threshold, so the next surprise reaches you before month-end, not after. As your account grows, reading the bill by hand every month stops scaling, and that is when continuous cost tooling earns its place (it is the category ZopNight is in). But the manual Cost Explorer pass is the right way to build the intuition first, and it is free. The take An AWS bill is not as scary as it looks once you stop reading it as a flat list and start reading it as a shape: open Cost Explorer, group by service, find the few big rocks, and ask what is biggest, what is unfamiliar, and what is growing. Everything else is detail you can learn as you go. The goal of your first read is not to understand every line, it is to know where the money is, so you know where to look next. What confused you most about your first cloud bill? For me it was data transfer, a real charge for something I did not think I was doing, and understanding it changed how I designed things afterward.