The Delay You Cannot Fix, and the 12 Points You Can Recover
Delays cost this airline 47 NPS points and could not be engineered away. Telling passengers six hours earlier recovered 12 of them. Here is how.
At a glance
- Delayed flights scored 47 NPS points below on-time flights. That single gap explained most of the airline's overall score.
- Punctuality was not a customer experience problem and could not be solved by one. Fleet, ground operations and air traffic control set it.
- Advance notice, roughly six hours ahead, lifted NPS among passengers on delayed flights by 12 points over eight months.
- Projected across the network, that recovery lifts airline-wide NPS from 15 to roughly 21, without a single additional flight departing on time.
The situation
The airline was measuring NPS across its network and watching the number sit at 15 without understanding what was holding it there. Feedback volume was healthy. The programme runs on roughly 1.2 million passenger interviews a month, so sample size was never the constraint. The score was not moving.
Splitting the score by flight status made the problem immediate. Passengers on flights that departed on time returned an NPS of 40. Passengers on delayed flights returned negative 7. With 47 percent of flights departing on time, those two numbers blend to the 15 the airline had been staring at.
On-time flights scored 40. Delayed flights scored negative 7. A 47 point gap, sitting inside a single averaged number that told nobody anything.
The uncomfortable finding
The largest driver of dissatisfaction was the thing the customer experience team had no authority to change.
More than half of all flights were departing late. That is a scheduling, fleet utilisation and air traffic control problem. No amount of survey design, closed-loop follow-up or service training moves it. A CX programme that promises to improve punctuality is promising something it cannot deliver, and most of them quietly do.
So the question changed. Not how do we stop the delays, but what is the delay actually costing us, and which part of that cost is ours to recover.
What we changed
Analysis of open-ended feedback from delayed flights showed the anger was rarely about the lost time itself. It was about finding out too late. Passengers who learned of a delay on arrival at the airport, or worse at the gate, had already made decisions they could not unmake. Passengers told earlier could adjust.
The intervention was narrow and operational: identify likely delays earlier in the day and notify affected passengers roughly six hours ahead, with enough specificity to be useful rather than a generic apology.
Nothing about the flying schedule changed.
The result
Over eight months, NPS among passengers on delayed flights rose by 12 points, moving that group from negative 7 to positive 5. Crossing from a negative score to a positive one matters more than the arithmetic suggests. It is the difference between a passenger who actively warns other people off and one who does not.
Because delays affect 53 percent of departures, that measured improvement projects across most of the network. Applying the segment result to the full passenger base moves airline-wide NPS from 15 to approximately 21. That network figure is an extrapolation from the measured delayed-flight result, not a separately measured overall score.
Amitayu Basu, CEO and Co-founder, NumrWe could not make the aircraft leave on time. We could change what the delay cost them. Six hours of notice was worth 12 NPS points.
Why this matters beyond aviation
Every industry has a version of this. The largest driver of customer dissatisfaction is frequently owned by a team that does not report to CX and cannot be fixed on a CX budget. Insurance has claim adjudication times. Banking has credit decisioning. Telecoms have network coverage.
The usual response is to keep measuring the thing you cannot change and report on it quarterly. The more useful response is to quantify precisely what it costs, then find the adjacent behaviour that recovers part of that cost.
Here that was worth 12 points, and it required no capital expenditure, no new system and no change to operations. Only knowing early enough to tell someone.
