# The missing middle *Someone has to work in months* --- Being able to zoom in and out of a business without getting whiplash is really hard to do, and rare. The fewer the layers, the further each person has to zoom, so if you don't have people who can do it, you might have too few layers rather than too many, or you might have cut the wrong one. Certain parts of an organisation move on a daily or weekly cadence, and that's the resolution you want a manager or a team leader thinking at. Somewhere above them there'll be a C-level "strategising", setting annual budgets and three-year plans. What connects the two? You need a level, or levels, of people who actually work at the grain that all the OKRs and the other planning fads tell you to work at, which is months and quarters. I think it's at this level that change and transformation lives or dies. Change looks easy on an annual plan, where there's all the time in the world. Change in any given week looks cheap to defer to the problem of the day. The actual transformation happens through lumpy weeks and effort sustained over months, and if nobody's job is measured in months, nobody is doing the sustaining. That takes someone, even if it doesn't take a whole layer. ## Time-span Elliott Jaques spent fifty years measuring what makes one job bigger than another, and landed on time: the longest task a role is trusted to see through to a finished result without someone checking along the way. His data had those time-spans clustering into bands, roughly three months, a year, two years and five, with a healthy hierarchy keeping one layer to each. His critics called the system rigid and mechanistic. The part worth keeping is that horizons differ by layer, and that a structure can get that wrong. > [!note]- The working > Jaques' bands: up to three months of discretion; three months to a year; one to two years; two to five; five to ten, and up. Each corresponded, in his studies, to a distinct mode of thinking about the future, and each is the natural home of one organisational layer. Strata are not pay grades, and both the boundaries and the measurements are argued over, but the shape held up across the firms, armies and hospitals [Elliott Jaques](https://en.wikipedia.org/wiki/Elliott_Jaques) measured. The useful residue is a one-page audit: for each layer, the longest piece of work it genuinely owns, asked of the layer above, because in Jaques' method it's the manager who sets the task and its deadline. Two layers sharing one band are easy to spot. The manager and the report own work of the same length, so the manager can't add anything except delay. A band with nobody in it is harder to see, because every role is filled and everyone is busy, and the work measured in months, the change that has to survive a run of lumpy weeks, belongs to no one. From outside it tends to look like the top job sagging, with the chief executive doing next quarter's work because it feels responsible. ## How it gets cut wrong In businesses that are acquisitive you can end up with too many parallel lower layers: three service-desk team leaders for three support teams, still operating as separate units, all reporting into a single chief operating officer. The executive either gets spread thin trying to manage down into the organisation, or promotes someone up to fill the span, without thinking about how long the new seat's work runs. I've made that mistake myself, over-promoting a sales leader because I felt exposed by a commercial gap beneath me. Their strong suit was the daily cadence of a sales machine, and the new role ran on the longer burn of building partnerships and adjusting pricing and packaging that would only land a year later.