Who carries the forecast error
A store roster can appear to be a technical output of the sales forecast. In practice, it is also an allocation of uncertainty. When a manager changes a shift at short notice because trading is weaker than expected, the uncertainty moves from the business plan into a colleague's day. Transport, care arrangements and other work may have been organised around the original hours. A chief executive who sees only the revised labour percentage can miss the consequence of the adjustment while still approving the system that makes it routine.
I learnt this in store management, when I was often the person trying to reconcile a labour budget with a week that had changed. Some changes were unavoidable. People became ill, deliveries arrived late and equipment failed. Other changes began much earlier in the organisation: a promotion was confirmed after the roster, a range change created more work than expected or a forecast assumed that a department could absorb another task. Treating every late amendment as local flexibility prevented us from distinguishing those causes.
The distinction matters because the remedies differ. An unexpected absence may need a reserve arrangement or a voluntary swap. A recurring workload surprise may need better planning between the commercial and operating teams. A labour budget that can only be met by repeatedly revising published hours may be understating the cost of the promised service. Asking a store manager to be more considerate does not fix an enterprise plan built on that assumption. The executive team has to decide which commitment it is willing to fund.
There is a further measurement problem. Not all changes are experienced in the same way. A colleague may request a swap that improves their week, while a management-initiated change may make it harder. Combining both into one instability measure produces a poor account of the issue. I want to know who initiated the change, how much notice was given and whether the colleague could decline without an informal penalty. The record should be designed with the people who use the roster, not only those who report the labour cost.
Budgeting for a steadier week
Our eighteen-store trial began with a costed commitment. We changed the publication routine, reviewed late management amendments and provided additional cover where the operating plan required it. Over six months, the proportion of scheduled shifts changed with less than seven days remaining fell from 23% to 9%. Labour cost per trading hour increased by 2.4%. I regard both figures as part of the result. Reporting only the first would conceal the resource choice; reporting only the second would fail to describe what the expenditure bought.
The difficult weeks were those in which demand disappointed. Managers could see colleagues working hours that the updated sales forecast no longer appeared to justify. I understood the discomfort because I had managed the same comparison. However, the trial would have meant little if its promise applied only when the forecast was accurate. We maintained the published commitment and used the difference to improve planning for subsequent weeks. That made the cost of forecasting error visible inside the business instead of automatically passing it to the people with the least influence over the forecast.
I do not claim that every retailer can adopt identical arrangements. Store formats, opening patterns and employment structures differ. A small branch may have fewer options for covering an unexpected absence than a large one. That is a reason to design a credible model, not to make a universal promise and leave managers to improvise. A useful proposal identifies the additional capacity, the range of exceptions and the point at which the service offer itself must change. Opening hours and departmental coverage may need discussion alongside the roster.
It is tempting to finance the argument through assumed improvements in retention or productivity. Those benefits are plausible, but plausibility is not a measured return. During our trial, some branches appeared to retain colleagues more successfully. The period was short and the stores differed in other ways. I did not use that observation to cancel the recorded cost increase. We could continue the approach because we judged the employment and operating commitment worthwhile, while continuing to examine longer-term effects. The choice did not need a fictional saving to become legitimate.
Exceptions should teach the organisation
An exception register can become another administrative burden unless someone uses it. I want recurring causes discussed at the level where they can be changed. If late promotional decisions repeatedly disturb rosters, the commercial calendar needs attention. If a delivery window creates work at a time the staffing model cannot cover, the operating assumptions need review. A manager should not have to solve the same structural conflict every Friday and then explain it as a personal failure to plan. Recording the cause is useful only when the organisation is willing to act on the pattern.
The process must also leave room for colleagues' preferences. Predictability does not mean that every person wants the same fixed pattern or that voluntary changes should become difficult. Some value extra hours; others need a stable limit. Store management still requires conversation and judgement. I resist designing the measure so narrowly that managers improve it by refusing changes colleagues actually want. The aim is a more dependable working arrangement, not a perfect score achieved by making the roster less responsive to the people it is meant to serve.
There are limits to what a public account can establish. A lower share of late changes does not prove that every colleague has enough hours, an acceptable income or a good manager. Those questions require their own evidence. I prefer a specific claim about the scheduling practice we changed and the cost we accepted. That precision makes the discussion more useful to another operator. They can examine whether the mechanism applies to their circumstances rather than being asked to admire a broad claim about culture.
A board can help by asking where uncertainty is being carried. Is the store absorbing it through unused capacity? Is a supplier absorbing it through erratic orders? Is a colleague absorbing it through a changed shift? There is no cost-free place to put variation, but some allocations are more consistent with the enterprise's commitments than others. My responsibility is to make that allocation explicit. A published roster should represent a plan the business intends to support, and the trading plan should acknowledge what keeping that intention will require.