The pilot nobody wants to keep
You manage a 14-person customer service team for a housing association. Complaints have doubled in six months, mostly about repairs not being done when promised. You ran a six-week pilot: tenants get a text when a repair is booked and another when the operative is 30 minutes away. During the pilot, complaints about operatives not turning up fell by 40%, but operatives disliked it - the scheduling app added about ten minutes to each job, and two of the three repair supervisors have told your manager the pilot was 'a gimmick that slows the lads down'. Your manager wants a decision by the end of the month. The pilot also failed to send 78 of 1,100 texts because of out-of-date phone numbers, and one tenant complained that a text went to her ex-partner.
1. What is the evidence actually telling you?
Read the worked answer
The headline is a fall in one complaint type, not a fall in complaints. I need to know what share of total complaints 'didn't turn up' was: if it was 20%, a 40% fall takes total complaints down about 8%. I also need to know whether anything else moved in those six weeks - a new repairs contractor, a holiday period - because an uncontrolled before-and-after can't separate the texts from the season. Comparing matched properties over the same period would settle it.
The ten minutes is real and has to be priced. An operative doing five jobs a day loses around 50 minutes, which is roughly a job per operative per day. Does that get repaid by fewer no-shows, fewer repeat calls and fewer rebooked visits? If it does, the supervisors' objection is about implementation rather than value, and implementation is something I can work with them on.
78 missed texts out of 1,100 is a 7% data-quality problem with an obvious fix - confirm the number at booking. The text to an ex-partner is different. That is a privacy defect, not a data-quality one, and I would suspend the arrival texts until numbers are confirmed and be able to say exactly what changed.
2. Two of the three supervisors are resisting. How do you handle them?
Read the worked answer
They have already gone to my manager, so going over them now guarantees they dig in. I don't.
What I do instead: give them the numbers I have just built, including the cost of the ten minutes, because if I am asking them to sell this to their teams they need to believe it. Then ask a question I genuinely don't know the answer to: 'what would make this take no extra time on the job?' Their answer will probably be better than mine - pre-populating the arrival time, having admin send the booking text, removing a duplicate phone call.
Then hand them ownership. Each supervisor runs the fix in their own area for four weeks, with the measure agreed up front and a review they chair. Objection becomes accountability, and the thing they were resisting becomes the thing they own.
If they still refuse after that, it goes back to my manager - but as a documented disagreement about a decision, with evidence, not as me reporting that they are being difficult. That is both fairer to them and a far stronger position for me.
3. Your manager wants a yes-or-no recommendation by the end of the month. Draft the substance of it.
Read the worked answer
Recommendation: extend for three months, conditionally, with a pre-agreed stop rule. Not a permanent roll-out - the evidence is one uncontrolled pilot with a privacy defect, and I would say that in the paper rather than dressing it up.
Conditions:
- Phone numbers confirmed with the tenant at booking; arrival texts suspended until then. The privacy incident gets reported to my manager in writing this week, with the fix - not because it is a crisis, but because they should not hear about it later from someone else.
- Arrival texts only, redesigned with the supervisors to cut the ten minutes.
- A comparison group of matched properties, and monthly reporting on no-show complaints, total complaints and jobs completed per operative.
Costs: the licence plus the admin time to re-confirm numbers, set out with the capacity cost if the ten minutes cannot be recovered.
Success measure and stop rule, stated in advance: if no-show complaints are not down by 25% at month three, or jobs per operative fall by more than 3%, we stop - and I will be the one saying so. That is what turns this from a preference into a decision.