Cascading a KPI looks simple: split the target among the units below. For some KPIs it works perfectly. For others it produces targets that don’t add up, can’t be controlled, or reward the wrong behavior. The type of KPI decides which.
In this guide
Key Takeaways
- →Additive KPIs split cleanly: revenue, units sold, sites certified, tonnes reduced.
- →Ratio KPIs don’t: margin, services share, on-time delivery, satisfaction.
- →Enterprise-only KPIs stay at the top: share price, brand reputation, credit rating.
- →A division at 25% could fall to 20% and still raise the group average while “missing” its share.
- →Cascade the drivers of a ratio instead: price realization, cost per unit, mix.
- →One definition everywhere; a unit can add local KPIs but not redefine a shared one.
Three kinds of KPI
Which kind a KPI is decides how it should travel down the organization, and getting that wrong is the commonest cause of targets nobody can own.
| Kind | Examples | How to deploy |
|---|---|---|
| Additive | Revenue, units sold, sites certified, tonnes reduced | Split the target; check unit targets add up |
| Ratio or rate | Margin, services share, on-time delivery, satisfaction | Don’t split; cascade the drivers each unit controls |
| Enterprise-only | Share price, brand reputation, group credit rating | Keep at the top; deploy the objectives that move it |
If you average a ratio across units you get a number. You don’t get a target anyone can own.
Why ratios don’t split
A group operating margin target of 15% can’t simply become 15% for each division. Divisions have different structures, and a division at 25% could fall to 20% and still raise the group average while “missing” its share. Cascading the ratio creates targets that are both unfair and uninformative.
Instead, ask what moves the ratio, and cascade those drivers: price realization in one division, cost per unit in another, mix in a third. Each unit owns a driver it controls, and the ratio stays a group result. The same logic applies to KPI design on an X-matrix; see Writing KPIs for the Hoshin Kanri X-Matrix: Definitions, Baselines, Sources.
Rules for cascading
- One definition everywhere. A unit can add local KPIs but can’t redefine a shared one.
- Additive targets must add up before they’re agreed.
- Contribution weights are explicit, so each unit’s share of a result is visible.
- Ratios cascade as drivers, never as copies of the ratio.
- Some KPIs stay at the top. Not every result needs a target at every level.
The second rule is the one that needs enforcing rather than remembering, and how it works in a negotiation is in Deploying Down: Turning Strategic Choices Into Targets Each Level Owns.
A worked example: five KPIs, five treatments
Here are five hypothetical KPIs from one strategy, each classified and treated accordingly:
Five KPIs from one strategy, each classified and treated accordingly.
| KPI | Kind | Treatment |
|---|---|---|
| Service revenue | Additive | Split by region; regional targets checked to add up to group |
| Services share of sales | Ratio | Not split; regions own contract attach rate and calls served locally |
| Operating margin | Ratio | Not split; divisions own price realization or cost per unit |
| Sites at maturity level 2 | Additive | Split by site; each site commits to a level and date |
| Group credit rating | Enterprise-only | Not cascaded; supported by margin and cash objectives |
What a badly cascaded KPI costs
A ratio split evenly across divisions costs a year of arguments. The division that was already strong carries a target it can miss while improving the group; the weak one carries a target it was never going to reach. Both spend the year explaining rather than acting, and the year-end review cannot separate a bad target from bad delivery.
Among published alternatives for enforcing the sum check, OnStrategy lists $200 per month for unlimited users. i-nexus publishes a free Starter tier for up to five users and $28 per user per month at its Standard tier. Ask either one what happens when the unit targets do not add up to the parent: whether the tool blocks the agreement or simply stores the numbers.
Profit.co publishes two plans, Essentials and Enterprise, quoting against modules and scale rather than a public seat rate. Explicit contribution weights and the balance constraint sit in Essentials with OKRs and Goals, Balanced Scorecard, Hoshin Kanri and Strategy Roadmap. What Drives the Cost of Hoshin Kanri Software lists the rest to total.
Cascades in Profit.co
In Profit.co, objectives show their contributing key results with explicit weights, catchball sessions apply a balance constraint so additive targets must add up before agreement, and KPI correlations show which drivers each ratio depends on. See the strategy deployment software page.
Those three map onto the three kinds of KPI above:
- Additive KPIs use the balance constraint, so the sum check happens at agreement rather than at year end.
- Ratio KPIs use correlations, so the drivers each unit owns are linked to the group ratio they are meant to move.
- Enterprise-only KPIs stay at the top with contributing objectives beneath them, rather than being copied down with nobody able to act on them.
How the same problem looks across whole companies rather than divisions is in Strategy Deployment for Multi-Entity Groups.
Free Product Demo
Check how your KPIs should cascade
A 30-minute walkthrough, tailored to your team.
Book a free demo
What happens in the call
How your teams set goals, track progress and review results.
Strategy, performance, projects and engagement on one platform, with AI.
A practical path to align teams, execute faster and drive results.
Questions this article answers
Additive KPIs such as revenue, units or sites can cascade by splitting the target, as long as unit targets add up to the total.
Not by splitting the ratio. Cascade the drivers that move it, such as price, cost or mix, so each unit owns something it controls.
No. Some enterprise results should stay at the top, supported by the objectives that move them.
No. One definition everywhere. A unit can add local KPIs for its own objectives, but a shared KPI has to mean the same thing in every unit that reports it.
The targets are unfair and uninformative. A division already above the group ratio can fall and still raise the group average while appearing to miss, and a weaker one carries a number it was never going to reach.
