How to benchmark a phone in a shop before you buy it

Nothing to install, about a minute, on a demo unit you do not own. What to run, which two numbers to read, how to compare it against the phone in your pocket, and why a demo unit is a worst case rather than a verdict.

This is the situation the test was built for. You are in a shop, the phone on the stand is described as top tier, and you would like to know whether that is true before you pay for it. Nothing needs installing, no account needs making, and a run takes about a minute — which is roughly how long a demo unit is yours.

The steps

Open the browser on the demo phone and load this site.
Tap Phone when it asks what you are testing. The rating means nothing until it knows.
Press Run and leave the phone alone. It will run every core flat out, then draw a heavy scene, then measure the display. Touching it during the run adds work and lowers the result.
Read the tier under the dial, and then the 1% low FPS in the results. The tier is the verdict; the 1% low is how it will feel.
Tap Share this result to another device, send the link to yourself, and open it on your own phone. Run the test there. The two results sit side by side, and that comparison rests on nothing estimated.

The two numbers to read

The tier — Flagship, High-End, Mid-Range, Entry-Level or DEAD TIER — places the phone against what a strong phone scores. If the shop said flagship and the tier says Mid-Range, that is a finding. It is the one figure on the screen that rests on an estimate until enough phones have been tested here, and the line beneath it says which basis is in force and how many have been.

The 1% low FPS is the frame rate during the phone's worst moments, and it is the number that matches how a phone feels. Two phones with the same tier can feel very different, and this is where the difference shows. The guide to that number explains how to read it against the average.

The line that says what the phone claims to be

Beneath the rating there may be a sentence saying the device reports a certain class of silicon — the graphics chip names the processor family, and the families are public — and comparing that with how it measured. A phone that reports flagship silicon and measures Mid-Range is performing below typical for its class. That is the finding this test exists to make.

It is not a finding of fraud, and the site will never call it one. A hot phone, a background app, battery saver, an aged battery and a genuinely weak unit all produce that same gap. What the gap tells you is that the phone in front of you, right now, is not delivering what its hardware should — and that you should ask why before you buy.

A demo unit is a worst case

Bear in mind what a display phone has been through: screen on since the shop opened, a dozen apps left running by a dozen people, possibly warm, possibly low on charge with battery saver on. Every one of those pushes the result down. A poor result on a demo unit is a reason to ask for a second run under better conditions — closed apps, a few minutes to cool, plugged in — not a reason to walk out. A good result on a demo unit, by contrast, is a good result under adversity, which is worth more than it looks.

If the throttling flag appears on a demo unit, read it in that light. The throttling guide explains what the flag actually measures and what the chart's shape means.

What you can trust and what you cannot

Everything in the results boxes was measured on that phone, in that browser, in the last minute, and every box says how. The comparison between the demo unit and your own phone is two measurements side by side, and it needs no scale at all. The tier is the one judgement, it is labelled as provisional while it is, and the guide on what the score means explains exactly what it rests on. Run the test twice if a number surprises you; the guide on repeatability says how much movement between runs is normal.

Measure the device you are holding

About a minute, nothing installed, and it works on a device you do not own.

Run the test

DeadTier