Preparation Scaled to Reversibility, Not Full Review for Every Change or Usual Approval: Library Branch Managers' Service Changes and What Happened When Problems Appeared

Abstract

We randomly assigned 212 branch managers in 10 public library systems to one of three ways of preparing service changes for 12 months. "Scaled" managers (n=71) first judged whether a change could be undone within a month at little cost. For changes that could, they wrote a 15-minute risk sketch: three to five plausible problems, the sign that each was happening, and a stop line set in advance at which they would revert. For changes that could not, they held a full review with branch staff and a regional coordinator. "Full-review" managers (n=70) held the full review for every change, and "usual-approval" managers (n=71) prepared as before. Results are for the 196 managers still in post at month 12. Scaled managers trialed a mean 6.8 changes, against 4.9 for full-review managers (28% fewer) and 6.1 for usual-approval managers. Among changes that researchers, blind to condition, classified as hard to reverse, 14.6% of scaled managers' changes ended in a serious problem, against 14.1% for full review and 18.1% for usual approval. When a problem appeared in any change, scaled managers adjusted the change and kept it going 61.3% of the time, against 58.1% and 49.2%, and rated themselves more certain of what to do next (5.2/7, against 5.0 and 4.1). By month 12, scaled managers' branches had kept a mean 4.1 changes, against 3.4 and 3.3. Branch staff in scaled branches rated the pace of change as heavier (3.8/7 for "too many changes at once," against 3.1 and 3.3).

1. Introduction

Public library branches change their services often and mostly in small ways: an extra evening opening, a homework club moved to a different room, a new rule for booking computers. A few changes are harder to take back. Removing shelving, ending a long-running program, or signing a contract for a new service commits the branch for months or years.

Many library systems handle both kinds of change in the same way, either with a uniform review process or with none. A uniform full review treats an extra evening opening as seriously as a contract, and managers facing that process may simply try fewer things. No preparation at all leaves a manager without a plan when a change goes wrong, which, for the hard-to-reverse ones, can be expensive for the branch and its patrons.

We tested a third arrangement, in which the effort of preparation follows how hard a change would be to undo. For an easily reversed change, the preparation was a short list of what might go wrong and a line, set in advance, at which the manager would reverse it. For a hard-to-reverse change, it was the full review. The question was whether the brief version was enough for reversible changes, and whether it left managers better placed when a problem did appear.

2. Method (Summary)

Ten library systems took part. All 212 branch managers were randomly assigned, stratified by system, to one of three preparation practices for 12 months. A service change was any change to opening hours, programs, space, lending or booking rules, or contracted services that the branch logged in its system's change register, which all ten systems already kept.

Scaled managers (n=71) attended a two-hour session. Before each change, they rated it as reversible (it could be undone within a month at little cost to the branch or its patrons) or hard to reverse. For a reversible change, the manager wrote a risk sketch on a one-page form in about 15 minutes, listing three to five plausible problems, the sign that would show each was happening, and a stop line: the point, decided before the change began, at which they would revert it. They shared the form with branch staff at the next team meeting. For a hard-to-reverse change, they held a full review: a two-hour meeting with branch staff and a regional coordinator, followed by a written plan. Full-review managers (n=70) attended the same session and held the full review for every change. Usual-approval managers (n=71) worked as before, which in most systems meant a line manager's sign-off and no written preparation.

All outcomes came from the change registers, supplemented by short surveys. Two researchers, blind to condition, classified every logged change as reversible or hard to reverse from its register entry, using the definition given to scaled managers, and resolved disagreements by discussion. Outcomes were compared using this classification, not the managers' own. A change ended in a serious problem if it was reversed at a cost recorded in the register above the system's minor-works limit, or drew three or more formal patron complaints attributed to it. When a manager logged a problem with any change, the register recorded whether the change was adjusted and continued, reverted, or left as it was, and the manager received a one-item survey asking how certain they were of what to do next (1 = not at all, 7 = completely), scored as a mean per manager. A change was counted as kept if it was still in place at month 12. Managers logged their preparation hours. At month 12, branch staff (1,027 respondents) rated whether their branch had made "too many changes at once" (1 = strongly disagree, 7 = strongly agree).

3. Findings

Of the 212 managers, 196 were still in post at month 12 (66 scaled, 65 full-review, and 65 usual-approval), and the results below are for them.

Scaled managers logged 449 changes, a mean 6.8 per manager. Full-review managers logged 318 (4.9), 28% fewer than scaled managers, and usual-approval managers 397 (6.1). The researchers classified 96, 92, and 94 of these changes, respectively, as hard to reverse. Scaled managers themselves rated 118 of their 449 changes as hard to reverse and sent them to full review. Of the 96 changes the researchers classified as hard to reverse, 84 had been given a full review by scaled managers and 12 only a risk sketch.

Of hard-to-reverse changes, 14 of 96 by scaled managers (14.6%) ended in a serious problem, against 13 of 92 by full-review managers (14.1%) and 17 of 94 by usual-approval managers (18.1%). The usual-approval rate was 24% higher than the scaled rate and 28% higher than the full-review rate.

Problems were logged for 142 changes by scaled managers, 93 by full-review managers, and 132 by usual-approval managers. Scaled managers adjusted the change and kept it going in 87 of 142 cases (61.3%), full-review managers in 54 of 93 (58.1%), and usual-approval managers in 65 of 132 (49.2%). The remaining changes were reverted or left as they were. For problems in reversible changes only, the rates were 72 of 115 (62.6%) for scaled managers, 38 of 68 (55.9%) for full-review managers, and 48 of 99 (48.5%) for usual-approval managers. Among the 167 managers who logged at least one problem (58 scaled, 52 full-review, 57 usual-approval), mean certainty about what to do next was 5.2/7, 5.0/7, and 4.1/7. Scaled ratings were 27% higher than usual-approval ratings.

At month 12, scaled managers' branches had kept a mean 4.1 changes (271 of 449), against 3.4 for full-review managers (221 of 318), 21% more, and 3.3 for usual-approval managers (215 of 397), 24% more. Full-review managers kept a larger share of what they tried (69.5%, against 60.4% for scaled managers) but had tried fewer changes.

Scaled managers logged a mean 13 hours of preparation over the year, including the initial session, full-review managers 29, and usual-approval managers 7.

Staff in scaled branches rated "too many changes at once" at 3.8/7, against 3.1/7 in full-review branches (scaled ratings 23% higher) and 3.3/7 in usual-approval branches (15% higher).

Fourteen scaled managers rated more than half of their changes as hard to reverse, whereas the researchers classified only 13 of these managers' 66 changes, about one in five, as hard to reverse. These managers trialed a mean 4.7 changes, close to the full-review mean.

4. Discussion

The headline result is that scaled managers matched the full review on hard-to-reverse changes while trying more changes overall. The design cannot fully separate two things here: whether the brief sketch prepared managers for reversible changes, or whether being released from the full review for those changes simply gave them more time and appetite to try things. The certainty ratings are consistent with the first, though they do not settle it. Usual-approval managers, who had no paperwork at all, were the least certain what to do when a problem appeared, but full-review managers, who wrote no stop lines, were nearly as certain as scaled managers (5.0 against 5.2).

One figure in the findings is worth dwelling on: for problems in reversible changes, 62.6% of scaled managers' changes were adjusted and kept going, against 55.9% for full-review managers and 48.5% for usual approval. Written preparation of either kind seems to have helped, and the stop line may have added something for reversible changes. A manager who had written down, before the change began, the sign of a problem and the point at which to reverse could judge a problem against a line already set. Without any written preparation, a manager had to decide under pressure whether the problem was serious, and usual-approval managers more often reverted the change or left it alone.

The arrangement was not free. Scaled managers spent nearly twice as many hours on preparation as usual-approval managers, though less than half as many as full-review managers. Branch staff carried a different cost: more changes meant more adjustments to routines, and staff in scaled branches agreed more strongly that the branch had made too many changes at once (3.8/7, against 3.1 and 3.3). The branches kept more changes, which may suggest that more of them were worth keeping, though full-review branches kept a larger share of what they tried and we did not survey patrons.

The fourteen managers who rated most changes as hard to reverse are the warning in these results. They trialed about as few changes as full-review managers did. A scaled approach depends on a manager being willing to call a change reversible, and where that judgment drifts toward caution the brief preparation gives way to the full review in practice.

5. Limitations and Future Directions

Managers knew their condition, and scaled managers may have logged problems more readily because their form asked them to watch for signs, which would inflate their problem counts and, if they logged smaller problems, their adjustment rate. The serious-problem threshold depended on each system's minor-works limit, which varied. Hard-to-reverse changes were relatively few, so the serious-problem rates rest on 13 to 17 events per group, and neither the near-equality of scaled and full-review rates nor the higher usual-approval rate is precisely estimated. Twelve changes the researchers classified as hard to reverse received only a risk sketch from scaled managers, so the scaled rate reflects managers' own judgments of reversibility, errors included. The researchers' classification of reversibility was made from register entries after the fact and may differ from what a manager could reasonably judge beforehand. Certainty was a single item. We did not measure patron outcomes directly, and changes kept at month 12 are an indirect measure of whether a change was worth making. A follow-up will test whether a short calibration exercise, comparing managers' reversibility ratings with colleagues' ratings of the same changes, reduces the cautious drift seen in the fourteen managers.

Editorial Note

This report underwent CfAHC's Rapid Review Protocol, including AI-assisted literature cross-checking and a two-reviewer human sign-off.