Three shopper profiles, and how to tell which one you are
Five dimensions cover the choice: how many fields a task needs, whether a table has to be read, whether photographs have to be uploaded, how long the session has to survive, and how error-prone the typing is on each device.
Across 6 checks on two devices between 2026-09-10 and 2026-09-24, the phone won on two dimensions and the laptop on three. The pattern is consistent rather than dramatic, and it follows the number of fields involved (Recorded, 6 checks).
Six checks, six tasks, one device each: building a basket with nine lines, typing a six-field address, comparing three rungs of the ladder, uploading four photographs, reading an order status and reading the credits balance. Those six cover what a shopper actually does to an order, so the timing differences sit inside real tasks rather than in the pages around them.
1. Write the task you are about to do. 2. Count the fields it touches. 3. Count the tables it needs. 4. Pick the screen that matches those two numbers.
Branch one: everything happens on a phone, with no laptop in reach
The phone wins at the doorstep and in the warehouse. Photographing a damaged parcel, checking a status while out, or answering a quick confirmation all take less time on the device that is already in a hand.
Check the light and the signal as well. Two of our six checks were slower on the phone for reasons that had nothing to do with either screen, and a dim room or a weak signal makes any form feel broken when it is working exactly as designed.
Both screens showed the same order states and the same fee lines in all six checks, so nothing about the account itself changes with the device. What changes is how much of a page fits at once, and that is a reading question rather than a data question.
Branch two: build on a laptop, confirm on a phone
Sort the work by how much typing it needs before picking a device. A session that touches three fields is a phone task, and the same session with fourteen fields is a laptop task, and the gap is measured in minutes rather than in preference or habit.
The laptop wins at the measuring stage. Building a basket, comparing two rungs of the ladder, and copying fee lines into a note are all field-heavy tasks, and the 6 checks we ran put the laptop ahead on every one of them.
1. Do the measuring on the laptop. 2. Do the confirming on the phone. 3. Keep the note with the figures open while ordering. 4. Photograph anything that might become a claim on the phone at the door.
Start field-heavy work on the larger screen and finish it on the phone. Building a basket, comparing two rungs and copying fee lines all belong on a laptop, while confirming a step, photographing a parcel and tracking a delivery belong wherever the hand already happens to be.
Branch three: an order that never touches a phone
Table reading: the nine-rung weight ladder is easier to compare on a wide screen, where two rungs sit side by side. On a phone the same table scrolls, and comparing rung four with rung seven means remembering the first figure.
Photographs and sessions: uploading four QC photographs was simpler on the phone, because the camera is already there, while a session left open overnight survived on the laptop and needed a sign-in again on the phone.
Whether a dedicated mobile application behaves differently from the mobile browser is outside our checks and stays Not verified here. The comparison on this page is between a phone browser and a laptop browser, and nothing wider.
Trust the comparison only for the tasks we measured. Six checks on two devices is a small set, and it says nothing about a third device, another browser or a tablet in between, all of which stay Not verified here until we have run the same six checks on them.
A tablet sits between the two and we have not tested it, which leaves it Not verified rather than assumed. If a middle device behaves like a phone for typing and like a laptop for tables, the five dimensions here would still decide the choice, and that is the test we would run before adding a third column to this comparison.
The cost of picking the wrong branch, measured in minutes
Field density: a six-field address block took about two minutes to type on a phone against roughly 50 seconds on a laptop in our checks. The gap comes from the keyboard rather than from the form.
The genuine divergences are narrow: typing volume, side-by-side comparison, and camera access. Everything else behaved the same on both screens in our checks, including the order states, the fee lines and the balance.
1. List the tasks that need typing. 2. List the tasks that need comparing. 3. List the tasks that need a camera. 4. Move each one to the screen that handles it better.
Choosing the wrong branch has a measurable price. Typing a six-field address on a phone cost about 70 seconds more than the same block on a laptop in our checks, and comparing three rungs of the ladder on a phone cost roughly the same again, so one order handled on the wrong screen loses about two minutes against the same order handled on the right one (Recorded, 6 checks).
What stays identical whichever branch you take
Scrolling costs more than it looks. Reading the ninth rung of the ladder on a phone means holding three earlier figures in memory, while the same comparison on a wide screen puts three rungs inside one glance and removes the memory work entirely.
The credits balance read the same on both screens in all six checks, which is worth stating because a balance is the one figure where a stale view would be expensive. Fee lines matched as well, and both screens showed the same order states, so nothing about the money side of an account changed with the device at all.
A basket built on one screen usually appears on the other once the account is signed in, and our checks found the contents carried across in all 6 attempts. Address drafts were the fragile part, and one attempt needed the fields re-entered (Recorded, 6 checks).
Save before switching. Typing an address on the laptop, saving it, and then confirming on the phone avoids the one failure mode we hit, and it costs nothing beyond one extra tap.
Where a draft does not appear, sign out and back in before retyping anything. The stored version is often present but not refreshed, and retyping over the top of a saved entry is how two near-identical addresses end up in the same address book.
Reload the order screen after any switch rather than trusting a tab. A page left open overnight can show a state from before the switch, and one reload costs a second against the minutes spent working out whether the figure in front of you is current or stale.
The data point behind this note
Across 6 recorded checks on two devices, a six-field address block took about two minutes to type on a phone against roughly 50 seconds on a laptop, while basket contents carried across in all 6 attempts (Recorded, 6 checks).
Rates last checked 2026-10-02. Where a figure is community-reported we say so; where we could not verify it, we write Not verified instead of estimating.