How the recommended ranking works
Two gigabytes a day
A plan sells you gigabytes. A trip spends days. Whichever runs out first decides what you actually got for your money, and that is the only thing our default order is doing. Here is the whole calculation, running live, with nothing held back.
Work out what the plan can actually deliver before it expires.
Divide the price by that, not by the number on the box.
Cheapest per usable gigabyte goes first. That is the whole ranking.
Start here
Which of these two is better value?
Both are real plans, both cover Japan, and one of them looks ten times cheaper per gigabyte than the other. Have a guess before you read on.
100 GB for $4.75
Valid for one day. That is five cents a gigabyte on the label.
20 GB for $9.90
Valid for thirty days. That is fifty cents a gigabyte on the label.
The second one, and it is not close.
Nobody moves a hundred gigabytes in a single day. Even flat out, a phone streaming video for hours will not get near it. So the first plan is not selling you a hundred gigabytes, it is selling you one day of internet for $4.75. The second is selling you twenty gigabytes you have a month to spend, and you will spend all of them.
Ranked on the label, the first plan wins by a factor of ten and sits at the top of the page. Ranked on what a traveller can actually use, it is $2.38 a gigabyte against $0.50, and it is nowhere near the top. That gap, between the number on the box and the number you can spend, is what this page is about.
The calculation, running
Build any plan you like and watch it get scored
This is not a diagram of the ranking. It is the ranking: the same function the site runs when it builds a destination page, doing the same arithmetic on whatever you set below. Try to invent a plan that games it.
live, the shipped function running on your inputs
- The box says 20 GB, and the plan is valid for 30 days.
- A heavy day of travel uses about 2 GB, so 30 days leaves room for 60 GB.
- The allowance fits inside that room, so every gigabyte counts and nothing is discounted.
- $9.90 divided by 20 GB is $0.4950 per usable gigabyte. That is the number this plan is ranked on, and it is never shown to you as a price.
What each day of this plan can give you
data availableallowance used upplan has expired
2 GB
day 1day 30
The allowance runs out after 10 days, leaving 20 days of validity with nothing left to spend.
- On the box
- 20 GB
- Actually usable
- 20 GB
- Price per GB shown
- $0.50
- Ranked on
- $0.4950
what you are sold
before it expires
the plan's own arithmetic, on its row
price per usable gigabyte
- Every gigabyte on the box counts toward the score
Where your plan lands, against four the method was written for
simify$20.23
- On the box
- 50 GB / 30 days
- Usable
- 50 GB
- Ranked on
- $0.4046
manet$9.90
- On the box
- 20 GB / 30 days
- Usable
- 20 GB
- Ranked on
- $0.4950
Your plan$9.90
- On the box
- 20 GB / 30 days
- Usable
- 20 GB
- Ranked on
- $0.4950
airhub$26.00
- On the box
- 750 GB / 10 days
- Usable
- 20 GB
- Ranked on
- $1.3000
etravelsim$4.75
- On the box
- 100 GB / 1 day
- Usable
- 2 GB
- Ranked on
- $2.3750
| Plan | Price | On the box | Usable | Ranked on | |
|---|---|---|---|---|---|
| 1 | simify | $20.23 | 50 GB / 30 days | 50 GB | $0.4046 |
| 2 | manet | $9.90 | 20 GB / 30 days | 20 GB | $0.4950 |
| 3 | Your plan | $9.90 | 20 GB / 30 days | 20 GB | $0.4950 |
| 4 | airhub | $26.00 | 750 GB / 10 days | 20 GB | $1.3000 |
| 5 | etravelsim | $4.75 | 100 GB / 1 day | 2 GB | $2.3750 |
Compared in US dollars, because that is the currency every plan is converted into before anything is sorted.
Load a real plan
Or try to break it
Each of these is a shape somebody has genuinely tried to sell. Pick one and read the verdict above.
The only two numbers we chose
Everything else comes off the price list
Price, allowance and validity are the provider's own figures. Two numbers in the calculation are ours, they are the only judgement in the whole ranking, and both exist to rule out the impossible rather than to tell anyone how much data they need.
2 GB
What one day can hold
Deliberately generous. This is a heavy day with video on it, not a typical one: maps, messaging and a normal amount of scrolling run well under it. We set it high on purpose, because the number exists to catch allowances nobody could spend, not to cap what we think you will use.
- Turn-by-turn navigation, all day
- about 0.05 GB
- Messaging and browsing, all day
- about 0.3 GB
- A few hundred photos uploaded
- about 1 GB
- Two hours of video on a phone
- about 1.5 GB
Rough figures, here to give the number a size. The point is that two gigabytes is a lot for one day, and a hundred is not a number a phone can reach.
30 days
The longest trip we credit
Capping the allowance against the clock fixed one end and broke the other. An unlimited plan valid for 180 days was credited with 2 GB a day for six months, which is 360 GB, and scored about a penny a gigabyte. A plan valid for a year gets 730 GB. Nobody on a trip uses either, and a plan priced by the year is not competing with one priced for a fortnight.
- 25th
- 5 days
- median
- 15 days
- 75th and 90th percentile
- 30 days
91% are 30 days or shorter
Being straight about it
What this method still cannot do
A ranking that only ever explains its strengths is marketing. These are the places where the arithmetic is thin, and you should hear about them from us.
- Cheap long plans still win
- The thirty day ceiling bounds the credit, it does not remove it. An unlimited plan valid for six months at a genuinely low price still scores well, and it should: it is a real offer at a real price. What the ceiling stops is that plan outscoring everything else by a factor of six on months of data nobody on a two week trip will use.
- It does not know your trip
- Two gigabytes a day is a ceiling used to rule out the impossible, not a prediction about you. If you are going for three days and rarely leave wifi, a cheap one gigabyte plan is the right answer and the ranking will not have worked that out. The filters on every board are there for exactly this.
- Price is not the only axis
- Network quality, whether tethering is allowed, whether a plan wants your passport, whether you can top up mid-trip: none of it is in this number. We rank on value per gigabyte because it is the one thing we can measure identically for every provider, not because it is the only thing that matters.
- Providers describe their own plans
- Allowance, validity and price come from the provider. We check units and currencies, we reject what we cannot convert, and we read fair-use caps in the unit the provider counted them in. We do not independently verify that a plan delivers what its feed says it delivers.
The part people are right to ask about
What money does, and does not, do here
We earn a commission when somebody buys through an outbound link, at no cost to the buyer. Here is exactly where that does and does not reach.
- It does not touch the order. The score is computed from each plan's own price, allowance and validity, before anything about a commercial relationship is read. A provider we earn nothing from is scored on identical arithmetic.
- It decides which plans a view puts forward: the top pick marker, the picks shelf, the featured partner rail, and which plans the Recommended tab of a board lists. Those four, labelled where they appear, and nothing else.
- No plan is unreachable. Cheapest first, Most data and Longest validity carry every plan we hold for a destination, in the same arithmetic, whoever sells them.
- The score is never shown as a price. A plan's row shows the provider's own price per gigabyte. The ranking figure is a comparison, explained here, and never quoted as what a gigabyte costs you.
The same commitments, in the language of a contract: our terms
Under the hood
The whole thing, in five lines
There is no model, no weighting and no machine learning anywhere in this. It is one calculation over four numbers a provider publishes, and this is all of it.
credited days = validity, or 30, whichever is smaller
usable data = allowance, or 2 GB per credited day, whichever is smaller
score = price divided by usable data
order = lowest score first, then lowest total price
listed = any plan priced at $0.10 or more- It is pure, which is why you can run it here
- No clock, no database, no request. The same four numbers give the same answer when a page is built and when you drag a slider. That is what lets this page run the real thing instead of a description of it.
- One definition, two consumers
- The page builder ranks with it when a destination page is generated, and a board re-sorts with it when you filter. A second copy of what counts as best would be free to disagree with the first, so there is only one, in a shared package both of them read.
- Unmeasurable sorts last, it is never hidden
- A plan we have no basis to compare falls to the end of the value order on total price. It keeps its row, its price, its outbound link and its place in every filter and every other sort.
- Ties are broken so rebuilds are stable
- Equal scores fall back to the lower total price, and then to the plan's own identifier. Without that last step a board would reshuffle on every rebuild for no reason a reader could see.
Prefer the short version? Six questions, one page: How we rank