NHD Partners All articles
Performance Assurance

Green All Over: Why Service Dashboards Flatter and What Boards Should Ask Instead

NHD Partners
Green All Over: Why Service Dashboards Flatter and What Boards Should Ask Instead

Photo: DHSgov, Public domain, via Wikimedia Commons

There is a particular kind of confidence that settles over a boardroom when the service report arrives with every indicator in green. It is not the confidence of evidence. It is the confidence of a full page.

Dashboards are built to be read quickly, and speed favours measures that are easy to produce. Availability, ticket volumes, response times, satisfaction scores — these can be generated from the provider's own systems at the press of a button. They are also, almost by construction, measures of activity rather than measures of outcome. A board that reads them attentively may still be no better informed than one that reads nothing at all.

Easy to measure, hard to matter

Consider what a typical indicator actually tells a non-executive director. That ninety-eight per cent of incidents were resolved within the agreed window is a statement about the classification of incidents and the definition of the window. Neither is visible on the page.

Many service indicators improve most reliably when the service is deteriorating. Resolution times fall when simpler issues are logged and harder ones are absorbed into informal channels. Satisfaction scores rise when the survey is sent to a narrower group. Volumes drop when users give up. A metric that can move in the right direction while the underlying service moves in the wrong one is not an assurance tool; it is a comfort object.

The deeper problem is one of authorship. Most of what arrives at a board on service performance is produced by the party whose performance is being assessed. That does not make it dishonest. It makes it partial, and partial information presented without its provenance invites the board to treat a summary as a finding.

Three tests before a metric earns its place

A short set of tests is usually enough to thin a dashboard considerably.

Would a change in this number change a decision? If the answer is no, the indicator is decorative. Useful indicators are those where a movement in either direction triggers a conversation about resource, scope or risk.

Could this number improve while the service deteriorates? If so, it needs a companion measure or a stated caveat, or it should go.

Who produces it, and can we see the raw data? An indicator that cannot be traced back to a source the client can inspect is a claim rather than a measurement.

A shorter set, chosen deliberately

A board does not need twenty indicators. It needs a handful that answer distinct questions, and it needs to know why each one is present.

A workable structure is threefold. First, one or two outcome measures chosen because the organisation genuinely cares about them — the things that would be discussed if no dashboard existed. Second, one or two risk measures that capture the ways the service could fail suddenly rather than gradually. Third, at least one measure whose definition the provider does not control, such as an independently collected user survey or a client-side operational count.

That last category matters more than it appears. A set of measures entirely defined by one party will, over time, describe that party's view of the world with increasing precision and decreasing usefulness.

A questioning routine for the boardroom

Metrics do not govern anything on their own. What governs is the conversation that follows, and that conversation is repeatable.

Ask what improved and what it cost. Improvements in service levels are frequently purchased with additional spend, relaxed scope or reclassified work, and the board should hear which.

Ask what is worse this quarter than last, and why it is not on the page. The silence around a deterioration is more informative than the indicators that surround it.

Ask which number the management team would remove if forced to choose. The answer tends to reveal which measure is least connected to how the service is actually run.

Ask what the organisation requested that it did not receive. Unmet requests are the most reliable early indicator of drift, and they never appear in a service level summary.

Ask where the provider's data is the only source. Concentration of evidence is a governance risk in its own right, quite separate from the performance it describes.

Ask which indicator would have warned the board if the service had failed last month. If no answer comes quickly, the dashboard is describing the past rather than protecting the future.

The page that is missing

One practice repays the effort more than any redesign of the reporting pack. Ask the provider and the client-side contract manager each to prepare, separately and without conferring, a single page titled what this dashboard does not tell you. Then compare the two.

The overlap is usually small. The differences are the agenda.

What the board should actually do

Board time is scarce, and reading a dashboard aloud is not a good use of it. Papers should arrive in advance, with exceptions highlighted and the narrative context supplied. The meeting should be spent on the two or three items where the numbers and the story disagree.

Fewer indicators, chosen for their connection to decisions, will produce sharper questions than a comprehensive page ever can. The purpose of assurance is not to feel informed. It is to be able to act, and to know when action is required before the provider's next quarterly report says so.

All Articles

Related Articles

Counting the Days: Decision Latency as a Hidden Cost in UK Service Relationships

Counting the Days: Decision Latency as a Hidden Cost in UK Service Relationships

Inherited Obligations: A Contract Walkthrough for New UK Board Members

Inherited Obligations: A Contract Walkthrough for New UK Board Members

The Unbilled Hour: Counting Management Time in UK Service Contracts

The Unbilled Hour: Counting Management Time in UK Service Contracts