BenchTested

Methodology

How we test software

Every app in a category goes through the same written script on the same clean Windows machine. Same inputs, same steps, same measurements, so a score of 8 means the same thing for every app. This page shows exactly how.

The four steps

Every review on BenchTested is the result of the same four steps. Skip one and the app stays “in testing”.

  1. 01

    A clean machine

    Every app is installed on a fresh Windows 11 virtual machine restored from the same snapshot. Nothing from the last test is left behind.

  2. 02

    The same script

    All apps in a category run identical tasks, pasted word for word, with answers we collected before the test so results can be graded.

  3. 03

    Measured and logged

    Each step is timestamped and screenshotted. We record time, cost, memory, prompts and anything the app leaves on your PC.

  4. 04

    A weighted score

    Each axis is scored from 0 to 10 and combined with fixed, published weights. Apps that have not finished the script get no score.

Our test machine

We test on a virtual machine so every app starts from an identical, clean Windows install, and so nothing an app installs can reach a real computer. After each test the machine is rolled back.

Host
VMware Workstation virtual machine
Operating system
Windows 11 Pro, English (US) interface
Hardware
4 virtual CPU cores, 8 GB RAM, no GPU passthrough
Display
1920 × 1080 at 100% scaling, light theme
Starting point
Restored from the same clean snapshot before every app
Accounts
Separate test accounts; free plans or trials unless stated

A virtual machine has no graphics card, so animation-heavy apps can look slower than on a typical laptop. When that affects a result we say so next to the number.

How a score is calculated

Each category has five or six axes, such as task success, speed, privacy and value. After the script has run, every axis gets a score from 0 to 10 based on what we measured. A short note under each bar on the review explains why.

The axes are then combined with fixed weights, published below for every category. The things that matter most for that kind of app count three times; minor ones count once. The overall score is the weighted average, rounded to one decimal.

If an axis could not be measured yet, it is marked “not scored” and left out of the average rather than guessed. The receipt shows a real example: Comet, tested on October 28, 2026.

Read the full Comet review →

Comet logo

Comet v155.0.8059.150

Test receipt · Oct 28, 2026

Tested

Scripted tasks passed1 / 8

Installer size14.7 MB

Install time≈ 50 s

On disk523 MB

Issues flagged in the log9

AxisScore × weight

  • Task success 3.0 × 3
  • Speed & resources 7.0 × 2
  • Privacy & control 5.5 × 3
  • Ease of use 6.5 × 2
  • Everyday browsing 7.0 × 1
  • Value for money 4.0 × 2

Σ(score × weight) ÷ 13 weight points

Overall score 5.2/10

What the verdicts mean

The score tells you how well an app did. The verdict label tells you whether we would recommend it, which is not always the same thing: an app can score well and still have a deal-breaker.

  • Top pickThe best overall result in its category. Usually 8.5 or higher.
  • RecommendedA strong choice for most people who match its “best for”.
  • SituationalGood at the job, with a drawback you should know about first.
  • AvoidFailed core tasks or put your data or PC at risk.
  • In testingHas not finished the script. No score, no ranking.

Test scripts by category

These are the tasks every app in each category runs. We publish them so you can judge whether they match how you would use the app.

AI Browsers

3 tested · ranking →
  1. T1Install from the official site and reach the first agent task; count steps, prompts and minutes, and note anything installed outside the browser.
  2. T2Run the same 8 scripted tasks on a free test account, pasted word for word: five core web tasks (research from official sources, a form fill, a price comparison, a multi-tab extraction and a scheduled routine), a request to buy a book on a demo shop, a five-slide deck and a tricky multi-step form. Each is scored as done, partly done or failed, with time and credits used.
  3. T3Check what the agent may do without asking: purchases, sending messages, entering passwords, and whether every action can be reviewed afterwards.
  4. T4Measure cold start, idle and under-load RAM/CPU against Microsoft Edge on the same 10 tabs.
  5. T5Check where chats and page data are processed and stored, training opt-outs, and what is left behind after uninstalling.

Score weights

  • Task success×3
  • Speed & resources×2
  • Privacy & control×3
  • Ease of use×2
  • Everyday browsing×1
  • Value for money×2

Evidence and screenshots

Every review includes a timestamped test log, a checklist of everything we checked, and screenshots taken straight from the test machine. Nothing is mocked up or taken from the vendor’s marketing.

  • Screenshots carry a small Benny badge and a faint BenchTested pattern so they can be traced back to our test.
  • Any personal data that appears on screen, such as a name or an email address, is blurred before publishing.
  • Facts we took from a vendor’s documentation rather than saw ourselves are labelled “Vendor-stated”.
  • Prices come from each vendor’s official pricing page, with the date we checked it.

AI-assisted, human-checked

We test with the help of an AI agent, and we say so openly because it is how we can run so many apps through the same script. Here is who does what:

TaskDone by
Creating accounts, passwords, payments and trialsHuman editor
Running the script, timing steps, measuring, screenshotsAI agent
Checking every answer against the sourceAI agent, then human editor
Draft write-up and scoresAI agent
Final verdict and publishingHuman editor

Rules we never break

  • No score without a test.

    Apps that have not finished the script are shown as “in testing”, with no score and no rank.

  • No paid rankings.

    Affiliate links never change a score, a rank or a verdict, and vendors cannot pay to be reviewed.

  • No made-up results.

    If a test went wrong on our side, we say so and mark it “not measured” until we rerun it.

  • Retests on major versions.

    When an app changes significantly, we run the script again and update the date on the review.