Long-Term KPI Dashboard: Know Months Later Whether the Gain Is Still There

The Long-Term KPI Dashboard does one job. It defines the long-term KPI dashboard that tracks sustained performance well beyond project closure. Not during the project. After it.

Most improvement work has a good final week. The team is attentive, the sponsor is watching, the numbers look right. Six months later nobody is watching, and the question nobody can answer is whether the gain is still there. This tool is the answer to that question, agreed in advance and written down.

Use it in Sustain, before handover, while the project team still exists. You pick a small set of measures. You record where each one started, where it has to stay, and who owns it. From that point the dashboard is a standing check rather than a one-off report.

The design choice that matters is the one about direction. Each KPI is read against its own target and its own direction of good. Some targets are ceilings. Some are floors. The tool does not mix units or flatten everything into a single score.

Everything that follows uses the sample data built into the tool. The case is fictional. It describes an accounts payable process that was improved and then handed back to the business.

Five KPIs carry the result. Error rate in per cent, with a baseline of 35 and a target of 8 or lower. Cycle time in hours, baseline 18.5, target 4 or lower. On-time payment in per cent, baseline 82, target 97 or higher. DPMO, baseline 350000, target 80000 or lower. Cost per invoice in dollars, baseline 12.40, target 5 or lower.

Four owners sit behind those five measures. The AP Manager owns error rate and cycle time. Finance owns on-time payment. The Black Belt owns DPMO. The Sponsor owns cost per invoice. That spread is deliberate. The measures closest to daily work sit with the person who runs the daily work, and the commercial measure sits with the person who funded the project.

The inputs are short on purpose. For each KPI you supply a name, a baseline, a target with its direction, a current value, a trend label and an owner. Three of those fields are required. KPI name, target and current value.

Write the target with its comparison symbol. In the sample, error rate is recorded as a target of 8 or lower and on-time payment as 97 or higher. That symbol is how the tool knows whether a higher current value is good news or bad news. Leave it out and the reading becomes guesswork.

Keep the units in the name. Error rate is labelled as a percentage. Cycle time is labelled in hours. Cost per invoice is labelled in dollars. DPMO needs no unit because the acronym carries it. This habit costs nothing and prevents the most common argument in a review meeting, which is about what the number actually means.

Set the trend label from evidence, not from hope. In the sample, all five KPIs are marked as sustained. That label is a statement about the direction of travel, separate from whether the current value passes the target. The two together are what make the dashboard useful.

The calculated output starts with five headline counts. Five KPIs. Five of five on target. Zero off target. Five of five with a trend sustained or improving. Four owners.

Then each KPI gets its own card. Error rate sits at 7.1 against a target of 8 or lower, on target, owned by the AP Manager. Cycle time sits at 3.8 against 4 or lower, on target, same owner. On-time payment sits at 97.5 against 97 or higher, on target, owned by Finance. DPMO sits at 69000 against 80000 or lower, on target, owned by the Black Belt. Cost per invoice sits at 4.70 against 5 or lower, on target, owned by the Sponsor.

The tool also reports how much of the gap from baseline to target has closed. The lowest is 101.4 per cent for cycle time. The highest is 104.1 per cent for DPMO. Every measure has closed more than the full distance it was asked to close, which is what the on target verdict reflects.

An AI review is available alongside the figures if you want the result explained in plain language before you take it to a meeting.

A clean board like this one supports a reduction in oversight. Five of five on target and five of five sustained is the evidence you need to move from weekly review to monthly, or to close out the heavier control activity the project left behind.

It also supports a change in who looks. With four owners named, the dashboard can be distributed rather than centralised. Each owner reads their own line. The Black Belt does not have to chase five numbers across four people.

Look at the margins before you relax too far. Cycle time sits at 3.8 against a target of 4 or lower. On-time payment sits at 97.5 against 97 or higher. Those are the two tightest lines on the board. They are the first two that would cross if anything changed in the process, and they are the sensible place to keep attention when everything else moves to exception reporting.

Slippage rarely announces itself. Watch for the combinations rather than a single bad number.

A KPI that is still on target but whose trend is no longer sustained is the earliest warning the dashboard can give you. Cycle time at 3.8 with a declining trend is a different situation to cycle time at 3.8 with a sustained trend, even though the current value is identical. The verdict column alone will not tell you that. The trend column will.

The second sign is owner drift. The sample has four owners for five KPIs. If a role becomes vacant and nobody takes the line, that measure stops being read even though it keeps appearing on the board. An unowned KPI is a decorative KPI.

The third sign is the current value going stale. The dashboard shows whatever you last entered. If the value has not been refreshed since handover, the board will keep reporting five of five on target long after that has stopped being true.

The most frequent error is too many KPIs. The sample uses five. That is a readable board. Twenty measures produce a board that nobody reviews and a result that nobody can summarise.

The second error is omitting the baseline. Baseline is not a required field, but without it the tool cannot report how much of the gap has closed. The sample case tells you cycle time has closed 101.4 per cent of its gap and DPMO 104.1 per cent. Drop the baselines and all you have left is pass or fail.

The third error is giving every measure to the same owner. Spreading five KPIs across four owners, as the sample does, means no single person can quietly stop reporting without someone noticing.

The fourth error is treating the trend field as a formality. Marking everything sustained because last month was sustained removes the only forward-looking signal on the board.

It will not collect the data. The current values are entered by you or your owners. The tool reads what it is given and does not check it against a source system.

It will not tell you why a number moved. A KPI that falls off target shows as off target. The cause sits outside this tool, in the control plan, the process map and the conversation with the owner.

It will not detect a stale entry. Nothing in the five sample KPIs tells the tool when those values were measured. If the dashboard is not refreshed on a schedule you set yourself, it will keep showing the last good result.

It will not forecast. The share of gap closed, from 101.4 per cent at the low end to 104.1 per cent at the high end, describes where the process has reached. It says nothing about where the process is heading next quarter. And it will not make anybody look. A dashboard that is never opened reports nothing, no matter how well it is set up.