Reduce WIP (Work In Progress). The only condition under which further limiting WIP will not increase throughput is where WIP is currently below the optimal capacity of the team. I dont think I’ve ever seen that in the real world.
Little’s Law allows us to calculate throughput, lead time and WIP. If you reduce WIP to zero, response time goes to zero, right? Of course not. It’s a useful relationship to calculate one variable from the other two, but less useful for understanding the dynamic relatlship between them.
More useful is the relationship of throughout to utilisation, which is WIP as a percentage of capacity.
Throughout peaks between 0% and 100%. At both those points it is 0 throughput, thanks to Littles Law. There is an optimal ratio of WIP to capacity. Below that ratio, increasing WIP increases throughout. Above that optimum, increasing WIP decreases throughout. It’s typically around 80%.
Capacity does *not* equal the WIP limit. If it does, you are setting your WIP limit at 100% which is a mistake. WIP limit should be set at optimum throughout, which is less than 100% of capacity, as I explained.