How to Measure Engagement Health Before It Goes Wrong
By the time an engagement turns red, the decisions that could have saved it were taken six weeks ago.
Every firm reviews its engagement portfolio, and most reviews use the status the engagement leader assigned. That figure is a judgment made by the person with the strongest incentive for it to be green, reported at the point when changing it requires admitting something. It is not a leading indicator; it is a summary of a conclusion already reached.
A health view worth having is built from measures the engagement produces as a by-product of running, which nobody sets deliberately and which move before the status does.
The leading indicators
- Effort consumed against progress. The earliest and most reliable signal, available weekly, and independent of anyone's opinion.
- Outstanding client dependencies and their age. A growing list of overdue items is a schedule problem that has not yet been recognised as one.
- Review rounds per deliverable against the number assumed. The second round is normal; the fourth is a definition problem or a relationship problem, and both need intervention.
- Unpriced approved changes. Any number above zero is a commercial exposure that will be argued about at the final invoice.
- Age of unbilled work in progress. Old work in progress is usually work somebody is reluctant to bill, which means somebody already believes there is a dispute.
- Team stability. Two changes of key personnel in a quarter predicts trouble on almost any engagement, and it is knowable in advance.
- Sponsor contact frequency. A sponsor who has not been spoken to in three weeks is a relationship risk regardless of how the delivery is going.
Why a single score is usually a mistake
Combining these into one number is attractive and it destroys the information. An engagement with excellent commercials and a deteriorating relationship averages to healthy, and the average is exactly wrong: the relationship problem is the one that ends the account.
Report the dimensions separately, with a small number of them, and let the reader see the shape. Three or four dimensions with a clear worst case is more useful than a composite that nobody can decompose.
Thresholds that trigger action rather than discussion
A health measure without a threshold produces conversation. A threshold with a defined consequence produces action. Set the consequence in advance and make it automatic, so nobody has to decide whether the situation is bad enough to raise.
For example: effort exceeding progress by ten points requires a written recovery action within a week. Three dependencies overdue by more than ten working days requires a sponsor conversation. Any deliverable in its fourth review round requires the engagement leader to re-agree the acceptance criteria in writing. None of these is punitive, and that is the point: the trigger has to be cheap enough that nobody avoids it.
The judgment that still matters
Measures do not replace the question a good practice leader asks, which is whether the client would engage this team again. That is not derivable from data, and the people who know the answer are usually the most junior members of the team, who see the client's reaction without the filter of a management meeting.
A short, direct question to them once a month, asked separately from the status process, is the single highest-return addition to any health framework. The metrics say whether the engagement is running to plan. The team says whether it is going well.