| Workload | Storage $ | Requests $ | Retrieval $ | Egress $ | Total $ | Not storage | |---|---|---|---|---|---|---| | Small app, 100 GB | 2.30 | 0.90 | 0 | 4.50 | 7.70 | 70% | | Backup archive, 5 TB | 18.00 | 1.00 | 5.00 | 45.00 | 69.00 | 74% | | Data lake, 50 TB | 1,150 | 90 | 0 | 180 | 1,420 | 19% | Only the data lake is dominated by storage, so a cheaper per-GB rate matters there and almost nowhere else. For the 100 GB small app, storage is $2.30 at $0.023 per GB, making the whole app's storage an inconsequential cost. One million reads and 100,000 writes add $0.90, revealing that access patterns barely dent the bill. Another 50 GB served to users adds $4.50. **Seventy percent of this bill is traffic**, so the page weight your app sends matters more than the size of the bucket. For 5 TB of backups, cold storage at $0.0036 per GB is $18. But restoring 500 GB costs $5 in retrieval and $45 to send out of AWS. One restore drill therefore costs nearly three times the monthly storage, and it's the bill you pay on your worst day. For the 50 TB data lake with analytics reads, storage is $1,150. Reading 100 million objects and writing 10 million adds $90. Then 2 TB leaving AWS adds $180. Keep analytics tools inside AWS and the $180 largely disappears. ## Hidden S3 costs and limits teams hit That $180 was data leaving AWS. It's one of four charges missing from the per-gigabyte price: egress, request fees on small files, forgotten versions and uploads, and penalties for moving data too early.  ### Egress and cross-region transfer Serve 5 TB of downloads and you pay about $450 before storage is counted: [EgressCost](https://egresscost.com/aws/data-transfer-pricing/) lists the first 10 TB leaving to the internet each month at $0.09 per GB. The next 40 TB runs $0.085, barely a discount.  Only the next 100 TB at $0.07, and anything over 150 TB at $0.05, bring real relief, so most small and mid-size teams never reach the cheaper tiers. Copying data to a second AWS region is billed as transfer too, so a disaster-recovery copy charges you on every gigabyte you replicate. ### Request fees on small files When you keep millions of tiny files, such as logs or thumbnails, **requests can cost more than storage**, because every read, write and list is a billed request. Some storage classes also bill tiny objects as if they were larger, so the cheap rate on the price page overstates savings for small files. ### S3 versioning and incomplete upload charges Turn on versioning and every overwrite keeps the old copy, billed at full price. Large uploads are sent in parts, and a failed upload leaves its parts behind, unseen in your file listing but still charged. A lifecycle rule that expires old versions and incomplete uploads is the cheapest fix.  ### Early deletion and retrieval penalties Delete sooner than the minimum storage period and cold classes charge you anyway. They also bill per gigabyte to read data back, so archiving something you'll open next month can cost more than leaving it in standard storage. Intelligent-Tiering adds a monitoring fee per object, which is why it suits large files better than tiny ones. ## How to reduce your Amazon S3 costs You lower the bill by matching each fix to the charge it removes. Skip any fix whose own fees outweigh the saving, which is the same reason tiny files stay out of Intelligent-Tiering. Doing nothing means paying premium rates for data nobody has opened in a year. Lifecycle rules move or delete objects automatically as they age. They suit data whose access drops off predictably, like logs and old exports. Intelligent-Tiering covers the unpredictable files. Two cleanups pay quickly: abandoned multipart uploads (half-finished large uploads that still bill for stored parts) and old object versions that pile up silently. **Data leaving AWS is usually the biggest charge**. A content delivery network such as CloudFront caches popular files closer to users, so S3 serves each file less often. Running your own servers in the same region as the bucket avoids transfer charges between them. Before changing anything, look: Storage Lens and Cost Explorer show which buckets and charge types drive the total. To see which fix suits you, use the table below, which pairs each fix with the cost it removes and the situation where it costs you instead. | Fix | Cost it fixes | When it backfires | |---|---|---| | Lifecycle rules | Paying top rates for cold data | Files get opened after moving, triggering retrieval fees; short-lived objects hit minimum storage charges | | Intelligent-Tiering | Unpredictable access patterns | Tiny or short-lived objects, where the monitoring fee exceeds the saving | | Multipart cleanup | Storage for abandoned upload pieces | Almost never, unless a slow upload is still running | | Expiring old versions | Accumulated superseded copies | You lose the rollback you were relying on | | CloudFront | Repeated transfer out of AWS | Rarely requested files, which cache poorly | | Storage Lens / Cost Explorer | Not knowing which charge is growing | Reports nobody acts on | Start with the reports, because they tell you which row of this table applies to you. ## S3 compared to Cloudflare R2 and Backblaze B2 Your reports will show whether most of your bill is data leaving AWS or data sitting still, and that decides it. S3 wins when your workloads live inside AWS, and the alternatives win when downloads dominate. Because rates change and differ by region, the table shows how each vendor charges rather than what it charges. What to take from it is the egress column: that's where the vendors differ most, and the **storage rate rarely decides the outcome**. | | Storage per GB | Egress fee | Requests | Free allowance | Min duration | S3 API compatibility | |---|---|---|---|---|---|---| | S3 Standard | Per-GB monthly rate, varies by region | Charged per GB leaving AWS | Charged per request, by type | Limited, aimed at new accounts | None | Native | | Cloudflare R2 | Per-GB monthly rate | None | Charged per request, by type | Monthly free tier | None on Standard | Supports most common operations | | Backblaze B2 | Per-GB monthly rate | Free up to a multiple of what you store | Some request types charged, some free | Small starting amount | None | Supports most common operations | | Google Cloud Storage | Per-GB monthly rate, varies by region | Charged per GB leaving Google | Charged per request, by type | Small always-free tier in select regions | None on Standard | Through an interoperability layer | | Azure Blob Hot | Per-GB monthly rate, varies by region | Charged per GB leaving Azure | Charged per request, by type | Credits for new accounts | None on Hot | Not S3-compatible without a translation layer | Before you move anything, confirm each row against the vendor's own pricing page. If your servers, databases and analytics already run inside AWS, traffic between them and S3 within the same region is typically free. Moving the data out means paying to cross the boundary and to rewrite anything that depends on AWS features. If customers or other clouds download your files again and again, you pay per gigabyte every time on S3. R2 charges nothing for egress, and B2 charges nothing up to a generous ceiling, so a download-heavy bill can shrink sharply there. ## Automating S3 housekeeping with Activepieces Activepieces builds S3 chores as a chain of steps, drawing on 744 integrations, and it can take over the repetitive S3 jobs that people forget, but it can't lower a single AWS rate. Moving less data out of AWS is still the only thing that shrinks a download-heavy bill, and that part isn't its job. ### Chores worth automating Activepieces has an Amazon S3 integration, and it meters the flow run rather than the steps inside it: one credit per run, whether a chore like copying new uploads into a bucket takes two steps or ten. Zapier's pricing page bills per task and Make's per module, so there the same chore costs more each time you split it into smaller integrations. Per-step pricing taxes the habit that keeps a bucket chore dependable, because listing, filtering and notifying as separate steps is exactly what makes a flow easy to debug. On a per-run meter the ten-step version costs what the two-step shortcut did, so you can build it properly. Check the exact list of S3 actions in its documentation before you design around any one of them. The screenshot shows the builder with a test flow open. A schedule trigger labeled "Every Hour" is selected, a panel on the right holds its settings, including a toggle for weekends, and a successful sample run sits beneath that.  The integration selector is open over the canvas, listing spreadsheet tools. The takeaway is how plain the setup is: pick a trigger, pick the apps, test it. A scheduled alert about a bucket's contents, logged to a spreadsheet, is built the same way. Next comes the cost of doing it. Every action the flow takes against S3 is a request, and requests are billed. An hourly job that lists or copies files will show up on your invoice, so a flow that runs too often can add to the bill you're trying to control. ### What stays in AWS settings Inside AWS is where lifecycle rules belong: the settings that move aging data to cheaper storage classes or delete it on a timetable. They run there without a separate tool, and a flow built to imitate them would pay request fees to do what the rule does for free. Stay in AWS too: storage class choices, bucket design and anything about data leaving AWS. Activepieces tidies the workflow around your buckets. The rates, and the access pattern that decides your bill, are still set on the AWS side. ## Checklist for auditing your S3 bill Because AWS sets the rates, your leverage is your own usage, and reading it takes one sitting. The list below is the whole job in order; skipping ahead means fixing a line you haven't measured. 1. Pull last month's bill broken out by usage type. 2. Find the largest line. 3. Set lifecycle and multipart cleanup rules. 4. Check whether versioning is keeping old copies you never read. 5. Price egress against the alternatives table. 6. Set an AWS Budgets alert. 7. Recheck the pricing page, since rates change. Below are the three groups of steps that trip people up. In AWS Cost Explorer, group last month's S3 spend by usage type instead of by service. A single "S3" total hides whether you're paying for storage, requests, or data leaving AWS, and each has a different fix. Your largest line tells you where to spend your effort, so ignore the small ones for now. Once lifecycle rules move aging data to cheaper storage classes, files nobody opens stop costing the full rate. A multipart cleanup rule deletes abandoned partial uploads, which you pay for yet can never use. Expiring old versions stops versioning from quietly keeping every overwritten file forever. Test each on one bucket first, because deleted data is gone. Set an AWS Budgets alert below the amount that would hurt. Without it, a runaway download job shows up only on next month's invoice, after the money is spent. Then put a recurring calendar entry to reread the pricing page, because rates and classes change and your rules should change with them. ## Frequently asked questions Where the pricing page leaves questions open, these are the ones people ask most. Check the answers against the current page, since terms change. ### Free tier and per-TB cost Is Amazon S3 free? No, not beyond a limited allowance. AWS has a free tier for new accounts, but its size and duration have changed over time. Treat it as a trial, and read its current terms before you plan around it.  How much does S3 cost per TB per month? It depends on the storage class, and the storage line is only part of the bill. Frequently used data costs more per gigabyte to hold than archive data, but archive classes charge for retrieval. Multiply the listed rate by your volume, then add requests and transfer out. ### Uploads and comparisons Does AWS charge for uploading data to S3? Moving data into S3 from the internet is generally not charged for bandwidth. The upload requests themselves are billed, though, so millions of tiny files cost more than a few large ones holding the same bytes. Check the request rates for your region. Is S3 cheaper than Google Drive or Dropbox for storage? Not reliably, because the products are priced differently. Drive and Dropbox sell flat plans per user, so your cost is predictable. S3 charges per use, which can be cheaper for data you rarely touch and costlier for data you read often or send out constantly. ### Reading a bigger-than-expected bill Why is my S3 bill higher than the storage price suggests? Usually because of access and egress, meaning data leaving AWS. Retrieval fees, request counts, old object versions you forgot to delete, and downloads to the internet all add up. Open the cost breakdown by line item and find which one dominates. ## Related reading - [Amazon Lambda Pricing: Cost Breakdown for 2026](https://www.activepieces.com/blog/amazon-lambda-pricing-cost-breakdown-for-2026) - [Amazon Bedrock Pricing (2026): Models and Token Costs](https://www.activepieces.com/blog/amazon-bedrock-pricing-2026-models-and-token-costs) - [Dropbox and Google Drive Compared for Cloud Storage in 2026](https://www.activepieces.com/blog/dropbox-and-google-drive-compared-for-cloud-storage-in-2026) ## References - [Amnic](https://amnic.com/blogs/amazon-s3-storage-costs) - [EgressCost](https://egresscost.com/aws/data-transfer-pricing/) - [Calculatorzilo](https://calculatorzilo.com/s3-price-calculator/)**Moving just 200 GB out costs nearly as much as keeping the whole terabyte.