Organizing goals into hierarchies provides precise control over progress and help you to understand possible risks and reasons of failure.

An intuitive and usual way to split a goal is based on your organization structure. If, for example, your goal is to produce a revenue of $ 1.2 M next quarter you can split that goal by assigning specific sub-goals by product, market region or both. Let's call such hierarchy "organizational" since subgoals will be probably assigned to similar descendants of a business unit, for example sales reps.

The second way to split a goal is through its functional components. Let's again consider our revenue goal. To make 1.2 M revenue we can set two sub-goals: $ 6 millions sale (1st goal) and at least 20% profitability  - revenue on sales - (second goal). In this cases sub-goals will be assigned to different figures of your organization, for example $ 6 millions to sales and 20% profitability to finance.

Tools like cause-effect graphs and strategy maps can help you identify functional goal hierarchies.

Interpreting variations (target to actual) in organizational hierarchies is quite intuitive, all what you need is a spreadsheet (or some kind of OLAP engine) to accrue variations by summing them up.

Things get more complex and more interesting in functional goals hierarchies. Say that we only made 1.05 M sales instead of 1.2 (thus 87.5% performance) how we justify missing in 12.5% our target? Looking at sales figures we see that we sold $ 7 millions (116% performance) but financial reports indicate only (15% profitability). You can visualize what happened in the following graph:

Yellow box area represents target value while blue box represents actual. You can see two variation rectangles (non-overlapped areas) dues to less profitability and more sales, and a third white rectangle (composite variation) resulting from combined effect of two variations.

How can we perform such computations? Let's suppose our goal formula is something like this:

goal = function (subgoal[1], subgoal[2]... subgoal[n])

Our case is simple:

revenue = sales × profitability

Let's compute actual and target goal values:

target[goal] = function(target[1], target[2], ... target[n])

actual[goal] = function(actual[1], actual[2], ... actual[n])

total-variation = actual[goal] - target[goal]

In our case:

target[revenue] = target[sales] × target[profitability] = 6M × 20% = 1.2 M

actual[revenue] = actual[sales] × actual[profitability] = 7M × 15% = 1.05 M

total-variation = actual[revenue] - target[revenue] = - 0.15 M.

Such total variation means we are 150 K$ under our target in revenues. Now we compute single variations for each specific subgoal k:

variation[k] = function (target[1], target[2], ... actual[k], ... target[n]) - target[goal]

In our case:

variation[sales] = actual[sales] × target[profitability] - target[revenue] 
    = 7M × 20% - 1.2M = 1.4 - 1.2 = + 0.2 M. 
    (sales increase responded for 200 K$ extra revenue.)

variation[profitability] = target[sales] × actual[profitability] -  target[revenue] 
    = 6M × 15% - 1.2M = 0.9 - 1.2 M = - 0.3 M. 
    (profitability decrease responded for 300 K$ less in revenue.)

And finally we compute composite variation:

composite-variation = total-variation - (variation[1] + variation[2] + ... variation[n])

In our case:

composite-variation = total-variation - (variation[sales] + variation[profitability]) 
    = -150 - (200 - 300) K$ = - 150 + 100 K$ = -50 K$.

Thus composite variation is responsible for 50 K$ less in revenue.

And now we have a more detailed explanation to our poor performance (87.5%) by computing impact as revenue variation relative to target revenue: