Reading an EBITDA bridge
·3 min read
An EBITDA bridge answers one question: the number moved — which lines moved it?
It is not a chart of performance. It is an arithmetic identity, drawn. Every bar is a line item's contribution, and the bars must sum to the difference between the two endpoints. If they do not, the bridge is wrong, and the honest response is to say so rather than to add a bar called other that makes the picture close.
A worked example
These figures come from the worked example built into the EBITDA Bridge tool. Load it yourself and you will get the same numbers, because the tool calculates them in your browser from the lines below.
| Line | Movement | Effect on EBITDA |
|---|---|---|
| Prior year EBITDA | $8.000m | |
| Revenue | rose | +$8.000m |
| COGS | rose | −$6.000m |
| Payroll | rose | −$1.500m |
| Marketing | fell | +$1.500m |
| Other operating costs | rose | −$0.200m |
| Current year EBITDA | $9.800m |
Check it: 8.000 + 8.000 − 6.000 − 1.500 + 1.500 − 0.200 = 9.800. The bridge closes exactly, which is what reconciled means on that page.
The three things this bridge is claiming
One: revenue did not carry it. Revenue rose by $8.000m and EBITDA rose by $1.800m. Costs absorbed $6.200m of the increase. A reader looking only at the revenue line would draw the wrong conclusion about the year.
Two: marketing is the only favourable cost line. It fell, which adds $1.500m. Whether that is good news is a question the bridge cannot answer — a marketing cut improves this year's EBITDA and may cost next year's revenue. The bridge shows the arithmetic; the judgement is yours.
Three: the direction of a cost is not the sign of its number. COGS rose by $6.000m and its bar points down. This is why the tool colours by the nature of the line rather than by whether the figure carries a minus sign — an expense that rose is bad news even though the movement is a positive number.
What sits outside the bridge
The same example carries depreciation moving from $5.000m to $5.400m — an increase of $0.400m. It appears in no bar and in neither EBITDA figure, because depreciation is below the line the metric is defined by.
It is still shown. A pasted profit and loss carries those lines, and silently dropping them is worse than displaying them under a heading that says they are excluded. A reader who cannot see what was left out cannot tell whether the exclusion was right.
Where bridges go wrong
The residual gets plugged. A bridge that does not close is telling you something — a line is missing, a sign is inverted, or the endpoints come from different scopes. Spreading the difference across the other bars destroys that signal and produces a chart that looks correct and is not.
The endpoints are not comparable. Twelve months against eleven, or a restated prior year against an unrestated current one, produces a bridge whose every bar is arithmetically right and whose conclusion is wrong.
The tail is not aggregated. Forty bars is not a chart, it is a table drawn badly. The tool gives the eight largest movements their own bar and groups the rest as Other, while the driver table beneath keeps every line at full detail — so the chart stays readable without the numbers being hidden.
Try it
The EBITDA Bridge takes a pasted profit and loss for two periods and produces the walk, the driver table and the reconciliation check. It runs entirely in your browser; the figures you paste are never uploaded.