Yes, push is a very scary place, but what is a push system and what does it mean to go a pull system?

One simple summary of a push situation is that we plan our actions in detail for a longer term based on what we think, but not what we know. This means that we plan our actions based on forecast of demand. In development, this is called waterfall and in production forecast-based planning, usually in an MRP system. In a push system where we then work based on the forecast, problems arise when the planned development is finished or the product has been produced. Most likely, what we’ve done will not be what the market wants now, i.e. it will not be according to forecast. The end result will be less sales and waste of work. That is why push is so scary.

A pull system is instead trying to produce or develop according to the actual market need. This doesn’t mean that the forecast is NOT needed. In a perfect world, where an organization has extremely short lead-times and unlimited resource flexibility, one could do without forecast. In the real world you need one, but the forecast is used differently.

In a push system the forecast breaks down what is needed for the long-term. In a development environment it would manifest itself in detailed long-term project plans, and in production the MRP system would trigger production and purchasing based on the forecast.

In a pull system we instead break up the planning in different time-buckets and then use the forecast for long-term decisions like manning, sizing of buffers from suppliers or setting high-level deliverables for product development.

On medium-term in a pull system, we set a more detailed plan focused on actual demand. In production this is either producing or replenishing according to real demand, and in development it is developing according to demand and then deploying development all the way to the customer.

On daily basis in a pull system, we make small reactive and proactive decisions to keep on track on demand. This can be compared to daily management in push which is mostly firefighting.

This model seems quite simple but it takes time to understand. I have applied and adjusted it into different settings when working in healthcare, but also in pro bono for a church and a microbrewery.

But the challenge is not understanding the model, it is the actual change. In a manufacturing company there will be a challenge in your planning system. In my first pull project in the 90s for a floor manufacturing company, we actually had to reprogram the IT system. Thanks to an agile IT department we were able to create a pull system based on re-ordering point, we used Little’s law for calculating the demand based on max order or forecast in near time and through an exponential smoothed forecast, we were able to also split the global forecast and also forecast variants. Then we created a priority index to easily sort all articles which we also grouped in product families that had to be produced in the same run. As you understand, it was not a straight road as we didn’t have this solution from the start. But the model of creating a pull system in three levels was our guiding compass.

So, what are your thoughts on push versus pull? Do you have any good examples of how to think when going from push to pull?

>> More Lean Learnings