What shared systems change about measurement
When each business unit keeps its own records, comparing their performance means collecting figures by hand, asking how each was calculated and hoping the answers are compatible. Shared systems change this: if all units record sales, costs, orders and service requests in the same system, the same measures can be produced for every unit from the same data.
That is the real contribution of centralized systems to performance. They do not improve performance by themselves. They make it easier to see performance, compare it fairly and spot problems sooner. Whether anything improves depends on what people do with what they see.
Start with a few measures
The temptation, once data is available, is to measure everything. Dashboards fill up with dozens of figures and nobody can tell which matter. A more workable approach is to choose a small set of key performance indicators (KPIs), often somewhere between five and ten for a unit or for the whole organization, linked to what the organization is trying to achieve.
Reasonable questions when choosing performance metrics include:
- What are the organization’s main goals this year?
- Which few figures would tell us whether we are moving toward them?
- Can each be produced reliably from systems we have?
- Who is responsible for each, and can they influence it?
- What decision would change if the figure moved?
If nobody can name a decision a measure supports, it may not need to be on the main dashboard. The wider question of choosing metrics that matter for centralized operations is covered elsewhere; this article concentrates on the performance side.
Common categories
Most organizations draw KPIs from a handful of areas:
- Financial: revenue, margin, cost against budget, cash position
- Customer: satisfaction, retention, complaint rates, response times
- Operational: order cycle time, error rates, on-time delivery, utilization
- People: turnover, absence, engagement results
- Quality or risk: defect rates, audit findings, incidents
A balanced set usually includes more than one category, so that improving one number does not quietly damage another. It also helps to combine leading and lagging indicators: lagging measures confirm what already happened, while leading ones offer earlier signals.
Consistent definitions across units
Comparing units only works if they measure the same thing. “Delivery time” might run from order to dispatch in one unit and order to customer receipt in another. Shared systems help because the calculation can be defined once, but definitions still need to be written down, owned and changed through a controlled process. Where units genuinely differ, for example in product mix or market conditions, comparisons need context, otherwise a ranking of units can be unfair and demotivating.
The danger of measuring what is easy
Data from systems is not automatically the right data. Some of the most important aspects of performance, such as customer trust, quality of advice, or the long-term health of a team, are hard to capture. Easy measures such as number of tickets closed, calls handled or reports produced are readily available, so they tend to dominate.
Problems then follow a familiar pattern:
- People work to improve the number rather than the outcome. A help desk rewarded for closing tickets quickly may close them before the problem is solved.
- Parts of the job that are not measured get less attention.
- Units look better or worse because of how they are counted, not what they do.
- Targets set on one measure push costs or problems into another.
These unintended effects of targets are well known, and no measurement set is immune. One safeguard is pairing a quantity measure with a quality measure, such as closures with reopen rates. Another is reviewing figures alongside human judgement and comments from the people doing the work.
A short illustration
Suppose a shared service handles requests for three units, and the only KPI is average handling time. The numbers below are invented. Unit A averages 10 minutes, Unit B 20, Unit C 30. A manager might conclude that Unit A is best. But if Unit A’s requests are simple password resets and Unit C’s involve complex configuration, the comparison tells little. Adding a measure of request type, and a measure of whether the problem stayed fixed, gives a fairer picture.
What centralization can add and what it can take away
Potential benefits
- Comparable figures across units without manual collection
- Faster, more regular reporting
- A common language for performance discussions
- Earlier detection of problems that span units
Potential costs and risks
- Central choice of KPIs may not reflect local priorities, which can reduce buy-in.
- One standard set can hide what matters locally.
- A single reporting system becomes a single point of failure for management information.
- Heavy reporting requirements can burden local teams with effort that produces little value.
Many organizations use a hybrid: a short group-wide set decided centrally, plus a small number of local measures chosen by each unit.
Making the numbers useful
Collecting figures achieves nothing unless they are discussed. Regular performance review meetings, held at a steady rhythm, with clear owners and follow-up actions, usually matter more than the dashboard itself. A performance baseline before any change makes it possible to judge whether things have improved. It also helps to review the KPI set itself at intervals, retiring measures that no longer inform decisions.
A practical checklist
- Write down the few goals the measures should serve.
- Choose a small number of KPIs and name an owner for each.
- Define each precisely and record where the data comes from.
- Pair volume measures with quality measures.
- Note the known limits of each figure on the dashboard.
- Review the set at least yearly.
Shared systems make good measurement easier but do not guarantee better performance. Treat metrics as aids to judgement rather than substitutes for it.