1 When the Northbridge library service published its annual performance table, the branch on Alder Street appeared to have had an unremarkable year. Loans were down, organised events attracted fewer people than expected, and the building remained expensive to heat. Its manager, Lena Ortiz, recognised the figures. What she did not recognise was the institution they seemed to describe. The table had no place for the man who spent three mornings learning how to open an attachment, or the teenager who discovered that she could stay until closing without being asked to buy anything. Ortiz was not claiming that these encounters cancelled out a heating bill. She was asking what kind of account could make both visible.
2 The council's analyst, Arun Shah, had heard objections like this before. Anecdotes, he pointed out, tended to arrive already selected for their emotional force. A moving story from one branch could not tell him whether another, less articulate manager was serving a greater need. Comparable figures offered a defence against decisions made by whoever could tell the best story. They also gave residents a way to challenge a service that was expensive but ineffective. Shah regarded the familiar opposition between numbers and human experience as unhelpful: without some common measures, the people most easily overlooked could become harder, rather than easier, to defend.
3 The difficulty emerged when a useful indication of activity became a definition of success. Alder Street recorded each visit, but its busiest staff member often helped the same few people with problems that took an hour to resolve. Across town, a branch beside the station registered far more visitors, many of whom passed through to use a printer. Neither pattern was inherently preferable. Yet a target based on visits gave one branch an advantage before anyone had asked what the visits accomplished. Ortiz called this a problem of compression: the measure was accurate, but several unlike activities had been squeezed into the same unit.
4 The limits of counting became clearer in a small experiment. The service introduced a target for completed digital-help sessions, hoping to demonstrate how much assistance staff provided. Within weeks, staff began recording a long appointment as several short sessions. No extra help had been delivered, and the staff had not invented appointments; the recording rules permitted their interpretation. The rising total nevertheless looked like an improvement. Shah withdrew the target. The episode did not show that staff were dishonest or that measurement was pointless. It showed how quickly a count could become a task in its own right, diverting attention from the purpose for which it had been created.
5 Ortiz's preferred alternative was a brief account of work that the standard categories missed. Shah agreed to read these accounts but resisted treating them as a substitute for the table. Who, he asked, would decide which stories were typical? And would the time spent writing favour branches with enough staff to turn their daily work into persuasive prose? The service eventually asked each branch to describe the same ordinary week, chosen in advance, including unsuccessful attempts to help. The request was less photogenic than a collection of success stories. That was part of its value: it made visible the time spent on work whose outcome could not be presented as a tidy achievement.
6 The accounts changed the discussion, though not in the way Ortiz had expected. They showed that Alder Street's patient, individual assistance relied heavily on two experienced employees. When either was absent, other staff struggled to continue it. What had sounded like an argument for leaving the branch alone became a reason to train colleagues and reconsider the timetable. This was an awkward gain. Giving a richer account of a service did not guarantee a more flattering one, and asking to be understood did not confer immunity from change. Ortiz accepted the training proposal, while continuing to oppose the suggestion that every appointment should be shortened.
7 There was a less comfortable question still. Time given to one person is time unavailable to someone else, even in a service whose purpose includes making room for people who are routinely hurried along. A detailed account could reveal that trade-off; it could not settle it on the council's behalf. Shah therefore wanted the final decision to identify what the service was prepared to sacrifice, not merely what it hoped to achieve. Neither the reassuring neatness of a dashboard nor the warmth of an individual story could relieve elected members of that responsibility. The disagreement was about priorities as well as evidence, and improving the evidence would not make those priorities identical.
8 The following year's report retained the table. Alongside it were the accounts of ordinary weeks and a short explanation of decisions that the combined evidence had informed. Alder Street received additional training time but no extra posts. Ortiz was disappointed by that limit; Shah remained wary of a report that busy readers might skim for a favourable example. Neither presented the arrangement as a final answer. Its achievement was smaller and more demanding: the figures could now be questioned without being discarded, and the stories could be valued without being allowed to decide the case. A better report had not removed the need for judgement. It had made the exercise of judgement harder to conceal.