No organisation I have joined had a productivity problem. They had a waiting problem. Those are different things, and confusing them costs you six months.
The symptom I see in every audit
The team is not working little. They work a lot, arrive late, and the result does not depend on the hours they put in but on how many times the work stopped to wait.
When you ask "how long does it take?", the answer is always a number from memory, always lower than reality. Nobody tracks how long an item sits still moving between three colleagues before someone signs it.
The first job is not to improve anything. It is to measure.
The four causes, in order
I have solved this in retail, in B2B e-commerce and in consulting. Waiting time almost always reduces to four causes, and they are always the same ones.
1. Waiting on a person. One approver for a process twenty people use. Approvals that bounce back with nobody having written why. Whoever decides is not in the same working hours as whoever works.
2. Waiting on information. The data exists, but it sits somewhere nobody looks. Two versions of the same file in circulation. Asking a person every time when you could ask the system.
3. Waiting on a decision. Whoever has to choose has no time to choose. The options were never narrowed to two. The decision criteria were never written down, so every decision is redone from scratch.
4. Waiting on an outside answer. Suppliers and customers with no expected response time. Requests that arrive without context and therefore bounce back. Nothing nudges when the deadline passes in silence.
The trap in almost every company
The first board that fills up is "add a person".
It fixes nothing. It adds a handoff, and that handoff comes back as coordination cost. I have watched whole quarters go this way: the bottleneck moves two metres and total time does not drop.
The sequence that works is always the same: remove a step first, then move a decision further up the flow, and only then consider adding people.
Where to start, concretely
Pick one process. Not the easiest one to measure: that is almost always the least problematic, so you will learn very little. Pick the one that makes you miss promises most often.
Write it in one sentence: when X happens, someone does Y, and the result is Z. If you cannot, the process is not defined, and that is already your first finding.
Then count only three things: how many times the work changes hands, how many tools touch it, how many approvals it must clear. If you find four or more handoffs on a process that should be simple, you have found the problem and you need nothing else for the first week.
One rule that saves time
Measure the same processes again after a month. If waiting time has not dropped, the problem was not the process you changed: it was one of its neighbours.
That is the mistake I see most often. You optimise a process, declare victory, and total company time does not move. Because the bottleneck was elsewhere, and what you touched was simply where the pain was loudest.