Every power tool likes to show a big green percentage. The question is what it's being compared against — a lot of them quietly compare your savings to a made-up "average PC" that has nothing to do with yours. Here's how to get a number you can actually trust: compared against your own machine's real history, with an honest label for when it's measured versus estimated.
Quick answer: A trustworthy savings number compares your average watts in a power-saving mode against the watts your own PC actually draws in its most power-hungry mode — not a hypothetical always-maxed device or a coefficient a vendor typed into a spreadsheet. PowerDoze tracks this automatically from real per-mode history and tags every figure Measured (laptop on battery) or Estimated (desktop / plugged in) so you know exactly how much to trust it.
Almost every power-saving tool — Windows included, with its built-in "energy recommendations" — shows you a percentage. What it rarely tells you is what that percentage is relative to. Three ways vendors fudge this:
None of that means the software doesn't work — it usually means the number describing it isn't built from your machine. The fix is comparing against something real: your own PC's history.
PowerDoze logs average watts for every power mode you actually use, over time. When it computes savings, the baseline isn't a hypothetical device — it's whichever mode in your own history drew the most power. Every other mode gets compared against that one number:
| Term | What it means |
|---|---|
baselineWatts | Average watts of the most power-hungry mode you've actually run — not a hypothetical max |
modeAvgWatts | Average watts actually measured/estimated while a given mode was active |
savedKwh | (baselineWatts − modeAvgWatts) × hours ÷ 1000 |
| Money saved | savedKwh × your electricity rate |
This has a direct consequence that's worth being upfront about: if you've never actually run a heavy mode, there's nothing power-hungry in your history to compare against, so the savings figure stays small or doesn't appear at all. That's the honest outcome — the alternative would be inventing a baseline you never touched, which is exactly the kind of number this whole page is arguing against.
Not every watts reading comes from the same place, and treating them as equal would be dishonest. There are exactly two sources, and PowerDoze always labels which one produced a given number:
| Badge | Where the watts come from | When it applies |
|---|---|---|
| Measured | Battery discharge rate — the real whole-machine draw, screen included | Laptop running on battery |
| Estimated | CPU package power + GPU board power, summed in software | Desktop, or any machine plugged into the wall |
The reason plugged-in machines can't get a "measured" badge isn't a shortcut PowerDoze took — it's physics. A PC has no built-in component that reports the wattage flowing through the wall socket. The CPU knows its own package power, NVIDIA GPUs report board power, but the motherboard, RAM, drives, fans, and the power supply's own AC→DC conversion losses are invisible to any software running inside the machine. A laptop's battery, on the other hand, genuinely has to supply every one of those components, so its discharge rate is the true number — no summing, no guessing.
Per mode, a reading is tagged real once at least 80% of its samples came from battery discharge rather than the CPU+GPU sum — a little mixed history (say you plugged in briefly) doesn't immediately flip the label. But the savings report as a whole only shows the green "Measured" badge when every mode being compared meets that bar; if even one mode in the comparison is estimate-only, the whole report is honestly tagged Estimated rather than quietly blending real and estimated numbers into one figure.
Best for: Understanding why the same PC can show "Measured" on battery and "Estimated" moments later plugged in. Note: The badge describes data quality, not whether the savings are real — estimated numbers still reflect a genuine difference between modes, just measured with a less complete instrument.
Want to know how the raw watts figure itself gets found in the first place — with a wall meter, a UPS, or software sensors? See our guide on measuring your PC's actual power draw.
PowerDoze samples power draw every 30 seconds by default (adjustable) and keeps a rolling window of detailed raw samples — 30 days by default — while everything older is folded into small permanent daily totals instead of being deleted. That means the lifetime savings number never resets: it keeps accumulating across app restarts, driver updates, even a full year of use, and the app defaults its analytics view to "All time" specifically because a short window makes real savings look artificially tiny. Per-mode history also feeds a savings table so you can see exactly which mode paid off and by how much, with the Measured/Estimated tag on every row.
Honest note: This is a Pro feature — the engine needs to be running to accumulate the history it's built on, and a fresh install with only one power mode in its logs won't have a baseline to compare against yet. Free tier: power mode creation itself is free up to 2 custom modes; the analytics/savings dashboard that reads their history is part of Pro.
It depends entirely on your machine and how you were running it before — there's no universal number, and any tool that quotes one flat percentage is guessing. The honest way to answer it for your own PC is to compare your average watts in power-saving modes against the watts your machine actually draws in its most power-hungry mode, over enough hours that the average means something.
The baseline is the most power-hungry mode in your own history, not a hypothetical always-maxed device or a vendor-supplied coefficient. PowerDoze logs average watts per power mode over time, finds whichever mode drew the most, and compares every other mode against that: savedKwh = (baselineWatts − modeAvgWatts) × hours ÷ 1000. If you've never run a heavy mode, there's nothing power-hungry to compare against, so the savings figure stays conservative.
Both labels exist and PowerDoze tells you which one you're looking at. "Measured" means the watts came from real whole-system draw — only possible on a laptop running on battery, where the battery's own discharge rate is the true reading. "Estimated" means the watts are a CPU + GPU sensor sum, which is what desktops and any plugged-in machine get, since there's no wall meter built into a PC. The app never blends the two into one unlabeled number.
A plugged-in PC has no component that reports true wall power — the CPU and GPU each report their own package power, but the motherboard, RAM, drives, fans, and the power supply's own conversion losses are invisible to software. Summing CPU + GPU is a reasonable estimate and it's still useful for comparing modes against each other, but it will typically read lower than what a physical wall meter would show.
No — lifetime savings accumulate permanently and survive app restarts. Recent samples are kept in detail for a configurable window (30 days by default), and everything older is folded into small permanent daily totals rather than deleted, so the cumulative number keeps growing for as long as you use the app.
Every 30 seconds by default, adjustable in settings. Each sample records total watts, the CPU/GPU split, which power mode was active, and whether the reading was real battery-based measurement or a CPU+GPU estimate — that per-sample tag is what lets the savings report and the measured/estimated badge stay accurate.
Want savings tracked against your own machine's real history instead of a vendor's guess — with an honest measured/estimated label on every number? That's the analytics dashboard, part of PowerDoze Pro.
Download free for Windows 10/11See also: How to measure your PC's real power draw · Limit CPU power on Windows · All features