How it works

2. Start with the numbers

One month, five products, two currencies. Everything is reported in US dollars (USD). 1 EUR is worth 1.10 USD in the Budget and 1.20 USD in the Actual.

Revenue in USD = local revenue × the rate:

3. Sort every line

The question: Which products can be compared like for like?

A line is one product in one currency and one unit of measure, in one month. The same product in two currencies or two units is two lines. A launch never shows up as a price rise, and a delisting never shows up as a volume drop, because no price is invented for a period where the product did not sell.

4. Price

The question: Did we charge more or less for the same product?

Take the units we actually sold and ask what the price change was worth. Price is measured at the Budget exchange rate so currency moves do not leak in.

5. Volume and mix

The question: Did we sell more units, and did we sell a different blend of products?

Products in the same currency and unit are looked at together (A and B). Volume is the total number of units. Mix is which products those units were. Both use Budget prices.

Check: A and B revenue moved from 3,000 to 2,940, which is -60. Price +140, Volume +150 and Mix -350 add to -60. Product C is alone in its group, so its Volume and Mix are 0. Another valid method could split Volume and Mix differently while the total stays the same, which is why the method is documented and subject to Finance approval.

6. Foreign exchange

The question: Did exchange rates change what our euro sales are worth in dollars?

The euro sales did not change. Only their dollar value did, by +165. Try it: move the rate and watch only this effect change.

This is translation only: the value of foreign sales when converted. It is not a gain or loss on settling invoices and not a hedge result. The rate is quoted as USD per 1 EUR.

7. New and discontinued, then the whole bridge

The question: Where do products that appeared or disappeared go?

On their own bars, so they never distort Price, Volume or Mix. New product N adds +80 (10 × 8). Discontinued product D removes 100 (20 × 5). Both are valued at the Budget rate, like everything else except FX. If a new product sells in euros, the exchange-rate movement on its sales is in the FX bar, not in the New bar. A discontinued product has no Actual sales, so it has no FX. Here both are in USD.

The axis starts at 4,300, not zero, so small bars stay visible. The exact values are on the bars and in the table below.

The effects add to +250, exactly the variance. If anything were left over it would show as the residual, never be hidden inside another bar. Click a bar to see what it means.

8. Check it independently, and know the limits

The question: How do we know the bridge is right, and what does it not tell us?

The tool does not trust its own bridge. Each control recalculates the answer straight from the raw rows, without using the sorting above, and compares.

Tolerance = the larger of 0.01 and one part per million of Budget revenue (here 0.01, because 1 ppm of 4,750 is only 0.005). Within tolerance is Passed. Up to 10 times it is a Warning. Beyond that is Failed, and every page and export says the figures are not validated.

Limits to keep in mind

• The bridge explains the variance under this method. It does not say why customers behaved as they did.
• Results depend on how lines are grouped (product, currency, unit), the exchange-rate convention and the allocation method. Other valid methods can split Volume and Mix differently.
• Year to date and quarter views are the sum of the monthly bridges. Shares and mix are recalculated each month, never averaged.
• The gross profit bridge uses unit margin for Volume and Mix, plus a separate cost-rate effect. A higher unit cost raises cost and lowers profit.
• Mix by category, channel, region or pack is shown separately. Those views overlap, so they are not added together.
• FX is translation only: no transaction gains or losses and no hedging.
• Missing rates, duplicate rows and zero quantities are flagged on import. A missing rate stops the calculation instead of being guessed.