Before you renew

Vendor scorecards

A changelog tells you that a price moved. This tells you what the cloud is like: how often it raises a published price, how big a typical increase is, and how big the largest one has been. That is the part of a renewal conversation you can prepare for.

How these numbers are worked out

Every recorded change in the archive is counted, ungrouped — a cloud that repriced forty meters on one day really did move forty published prices, and collapsing them would score a broad repricing the same as a single one. Suspected listing corrections are excluded — a placeholder price becoming a real one is not an increase, and counting it would rank whoever publishes the most placeholders worst. Rates are per observed year over the span actually in the archive, never extrapolated from a shorter one; where the span is too short for a rate to mean anything, the cell says so instead of printing a number. Typical increase is the median, with the mean beside it, because these distributions have long tails.

See the changes themselves →

Vendor scorecards

Every recorded change between 16 Aug 2026 and 22 Aug 2026 — 7 days observed

For each vendor: how often it raised a published price, the typical size of an increase, and the largest one recorded.
Cloud	Increases / year	Rises · falls	Typical risesampled	Largestsampled	Spreadsampled
Azure132,678 recorded increases	too soon to say	132,678131,643	+314.7%mean +22225.7%	+605885.2%	1,871 SKUson 2 days
GCP1,011 recorded increases	too soon to say	1,0111,472	—	—	—
AWSno increase recorded	too soon to say	090	—	—	—

Counts and rates are exact over the span above. The three columns marked sampled rest on the 2,000 most recent of 133,689 recorded increases — a median needs the individual values, and reading every one of them would be a whole-archive scan on each page view. Suspected listing corrections are excluded throughout: a placeholder price becoming a real one is not a vendor raising a price, and counting it would put whoever publishes the most placeholders at the top of this table. Typical rise is the median, because these distributions have long tails and a handful of very large moves on sub-cent meters drag any mean somewhere useless. These are published list prices, so a scorecard describes what a vendor does in public — not what it does to your account at renewal, which is negotiated and private.

A scorecard is a summary of what is already on the changes page, so every figure here can be traced back to the dated rows it came from. Nothing on this page is a forecast.