What it calculates

An EPS bridge decomposes the change in earnings per share between two periods into the drivers that caused it - operating performance, margin, interest, tax, share count - so you can say why EPS moved rather than only that it did.

How it is calculated

Each driver is isolated by changing one variable at a time from the base period while holding the others at their base values, and the resulting EPS delta is that driver's contribution. Share count is treated separately because it affects the denominator: a buyback raises EPS with no change in earnings at all. The individual contributions should sum back to the total change, and any gap is the interaction effect from drivers moving together.

How to read the result

The important split is between earnings growth and share count reduction. EPS growth driven by buybacks is real for the holder but tells you nothing about whether the business improved. Look at how much of the bridge comes from below the operating line - tax and interest contributions are usually one-off or unrepeatable, and a bridge dominated by them is a warning that the underlying trend is weaker than the headline.

Worked example

EPS goes from 2.00 to 2.40. Revenue growth contributes +0.22, margin expansion +0.11, higher interest expense -0.06, a lower tax rate +0.05, and a buyback that cut the share count by 3% adds +0.08. The contributions total +0.40, matching the change - and a fifth of it came from the buyback rather than the business.

Common questions

Should the bridge use basic or diluted shares?

Diluted, since that is the figure the market quotes and it reflects the claims that actually exist. Using basic shares flatters EPS by ignoring options and convertibles that will eventually participate in earnings.

Why do the individual contributions not always sum exactly to the total?

Because the drivers interact. Higher revenue at a higher margin produces more than the sum of the two effects measured separately. That residual is the cross term, and it should be shown as its own bar rather than quietly allocated away.

Does order matter when building the bridge?

Yes, if each step is measured from the previous one rather than from the base. Sequential decomposition assigns interaction effects to whichever driver comes later. Measuring every driver against the same base period is more neutral and makes the residual explicit.