Which Exchange Rate Should You Use When You Record Travel Spending?

You pay 120 baht for a coffee in Bangkok. You live in London. What goes in your record — 120, £2.79, £2.87, or the £2.91 that lands on your statement eleven days later?

All four are defensible. They are not the same number, and the gap between them is not rounding noise: across a three-week trip it is routinely the price of another two nights' accommodation. If you are keeping a record of what your travel actually costs — not a rough feel, but a record you will still trust in two years — you have to decide which number you are storing, and then be consistent about it.

Here is the honest version of how that works, including the places where our own conversion is wrong.

There is no such thing as "the" exchange rate

At the moment you hand over your card, at least three different rates exist for the same pair of currencies.

The reference rate. The European Central Bank publishes euro foreign exchange reference rates at around 16:00 CET on every working day, derived from a daily concertation procedure between European central banks that normally takes place around 14:10 CET. It is a single daily snapshot covering roughly thirty currencies. The ECB is unusually blunt about what it is for: the rates are published for information purposes only, and using them for transaction purposes is "strongly discouraged". Nobody will sell you baht at this rate. It is a yardstick, not a price.

The network rate. Visa and Mastercard each set their own conversion rate for the day a transaction clears — not the day you spent. It sits close to the reference rate but is not identical, and it moves with the settlement date, which is why a purchase made late on a Friday night can convert at Monday's number.

The rate on your statement. The network rate, plus whatever your card issuer adds on top: a foreign transaction fee, a currency conversion charge, or both. This is the only one of the three that is genuinely what you paid.

European law wrote that hierarchy down. Regulation (EU) 2019/518 requires payment service providers, and anyone offering conversion at a till or an ATM, to express their total currency conversion charges as a percentage mark-up over the latest available ECB reference rate, disclosed to you before you confirm the payment. The regulation deliberately sets no cap on the mark-up. It only forces the mark-up to be quoted against the reference rate — precisely because the reference rate is the neutral benchmark and the price you pay is not.

The one decision that actually costs money

When a card terminal in Lisbon asks whether you would like to be charged in pounds instead of euros, that is dynamic currency conversion, and saying yes hands rate-setting to the merchant's payment provider instead of your card network. Typical mark-ups sit in a 3–7% band over the interbank rate. A study by a Norwegian bank of around 1,500 transactions found dynamic conversion cost an average of 7.6% more than the standard route, with 99.7% of those transactions coming out more expensive. Research published in the Journal of Public Policy & Marketing in 2023 found the option lands hardest on less financially literate customers, and that showing the mark-up as a percentage — rather than as a raw exchange rate — makes people decline it far more often.

For record-keeping this matters more than any choice about which rate to store, because no bookkeeping method recovers a seven percent mark-up you agreed to at the till. Always choose the local currency. Then worry about your record.

What MapMyWorld actually does

Every expense you log carries two things you typed: the amount, and the currency you paid in. Alongside them the app stores two things it worked out: the equivalent in your home currency, and the exchange rate it used, kept to six decimal places, written onto that individual expense row.

The rate is the historical rate for the date of the expense, not today's rate. It comes from a public API serving European Central Bank reference data, cached for twenty-four hours per currency pair per date so a long trip does not hammer the source.

Because the rate is written onto the row rather than recalculated whenever you open the app, your totals do not drift. A trip you closed in March shows the same figure next year that it showed the week you got home. That property is worth more than getting the rate exactly right, and it is the reason the number is stored at all: a record that quietly re-values itself is not a record.

Your home currency is a single account-level setting, and it drives every converted figure — the trip total, the card and cash split, the average per day so far, and the "daily excluding flights and stay" line that strips the two big one-off lumps out of your burn rate so the remainder resembles what a day on the ground actually costs.

Where this is wrong, and by how much

Four honest gaps, in rough order of how much they can cost you.

Currencies outside the reference set fall back to today's rate. The app offers forty currencies. The ECB reference set covers twenty-eight of them. The other twelve — among them the UAE dirham, Saudi riyal, Vietnamese dong, Argentine, Chilean and Colombian pesos, Peruvian sol, Egyptian pound, Moroccan dirham, Nigerian naira, Kenyan shilling and Russian rouble — have no historical reference rate published at all. When the lookup fails, conversion quietly falls back to the current rate and stores it as though it were the rate on the day you spent. For a stable currency over a fortnight, that is noise. For a currency in a steep slide, it is not noise; it can be tens of percent, and nothing on screen tells you the substitution happened.

Weekends and public holidays are silently back-dated. Reference rates exist only on working days. Ask the API for a Sunday and it does not return an error — it hands back the previous working day's rate. Checked while writing this: a request for 6 September 2026 came back with the rate published on Friday 4 September. Your Sunday dinner in Rome is converted at Friday's number. In a calm week that is a fraction of a percent. Across a weekend when something moved, it is more.

The last-resort fallback is a rate of 1. If both rate sources fail, conversion returns 1 and the home-currency amount simply equals the local amount. A 5,000 yen dinner recorded as 5,000 pounds is obvious enough to catch by eye, but nothing flags it for you.

Converted amounts round to two decimals. Rates keep six decimal places, so the arithmetic is always recoverable from the row. The converted figure is not: you cannot reconstruct an unrounded total from a long list of small purchases in a currency that has no minor unit.

Two further limits are worth knowing before you lean on any of this. The per-trip PDF export lists each expense in the currency you paid in, while the trip total is given in your home currency — so the line items will not visibly add up to the total printed above them. And neither bulk export carries expense rows at all: the CSV export covers places only, eight columns from name through notes, and the JSON export covers places, trips and summary statistics. Your individual expense lines live in the app, and the PDF is currently the only way they leave it.

A rule that survives contact with a real trip

Record what you paid, in the currency you paid it, on the day you paid it. Treat the conversion as an annotation rather than as the record itself.

This is the opposite of the common habit, which is to convert roughly in your head at the rate you half-remember from the airport and log the estimate. The estimate is unrecoverable. Six months later you cannot tell whether "£12" came from a rate or from a shrug. "480 THB on 14 March" is a fact, and any conversion of a fact can be redone later.

Then, once, at the end of the trip, do a reconciliation. Open your card statement, find the three or four largest transactions, and compare them against what the app shows. You are looking for a consistent gap, not an exact match. A steady two or three percent above the reference conversion is your issuer's foreign transaction fee behaving exactly as advertised. A single wild outlier is usually a dynamic conversion you accepted without noticing, or a currency from the fallback list above. Cash withdrawals will be their own small disaster — an ATM operator fee, plus your bank's fee, plus sometimes a conversion mark-up — and none of those three are visible to the app.

If the gap is consistent, leave the record alone and simply know that your true cost ran, say, two and a half percent above the figure on screen. If you want the record to be exact, edit the affected rows: changing an amount re-runs the conversion for that expense's own date.

Where this fits

Choosing your rate rule is one small piece of a larger habit: recording spending as it happens instead of reconstructing it weeks later, and logging a trip while you are still on it. It is the same discipline as picking a rule for what counts as visiting a country and then writing that rule down. The value of a record comes from consistency far more than from precision.

If you want to test the habit, expenses live inside a trip, and the free plan includes one trip — enough to run a whole trip end to end and see whether the numbers hold up against your statement before deciding whether $7.99 a month is worth it.