Conference giveaways: A mini experiment in pull based production

Posted on February 22nd, 2007 in Lean by siddharta || 1 Comment

I was a volunteer at the Proto.in conference last month. I was a part of the registration desk, and my job was to coordinate the packing of the giveaways for the registrants. There were two types of giveaways: one type of giveaways for the presenters and venture capitalists and another type for the audience. Each type required some packaging like putting on some stickers, bundling a couple of items together and tying them up with a conference badge. One item was common between both the sets of giveaways.

The challenge then, was twofold –

  • We needed to have some inventory of packed giveaways so that the giveaway could be given quickly
  • Since we didn’t know the exact number of registrants for each type of giveaway and the time when each registrant would arrive, we needed some way of handling the uncertainity

There are two methods of handling these challenges: Push based production and pull based production

Push based production

Push based production is when you try to predict the demand, plan your inventory accordingly and then ‘push’ the inventory onto the market. In this case, we had some pieces of information that we could use to predict demand

  • Although we didnt know the exact numbers, we knew that there would be many more audience than presenters
  • We had a ballpark estimate of the number of audience and presenters based on registration numbers
  • One huge variable was the number of on the spot registrations, of which we had no idea

In a push based strategy, we would pre-pack the two types of giveaways based on the registration numbers of audience and presenters. We had no idea about the number of on the spot registrations, so we would have to make a guess and pre-pack that as well. To top it off, we would have to pre-pack a few more to act as buffer. However, since one item was common between both giveaways, and it was in short supply, we could only have a very small buffer.

Once registration started, we would simply hand out whatever was pre-packed. In case the actual numbers were accurately predicted, the registration would be smooth as we just hand out the pre-packed giveaways. However, if the actual numbers were substantially different from the predicted ones, there would be a mess as we would have to pack giveaways on the spot, or worse, unpack one type of giveaway and pack the other type.

It is easy to see that the success of push based production depends upon the accuracy of the prediction.

Pull based production

In pull based production, the production is controlled by the actual demand. This was the method that I used.

To start with, we packed ten items of both types of giveaways. This was the inventory. As registration proceeded, the inventory would decrease. When the inventory came down to around five items of a giveaway, we would pack a few more items of that giveaway and replenish the inventory. Therefore, if there were a lot more presenters at the start, we would be packing more of that giveaway. If there was a surge in audience numbers, the inventory for that giveaway would start to reduce and we would pack more of that giveaway. In all, the production pattern would be correlated with the demand pattern.

For pull based production to work, a few conditions need to be satisfied

  • It should be easy to switch between the different types of production. In our case, that was easy because all the items to be packed for both types of giveaways were available right next to us
  • It should be possible to get real time information on demand and resupply. In our case, we were sitting right behind the registration desk, so that was easy.
  • It should be possible to frequently resupply small batches. Again, we were right behind the registration desk, so no problems there.

An interesting situation occured when it seemed like the inventory would run out, at which point we started supplying items one at a time as they were produced. This was possible because there was virually no time gap between production and supply. However, introduce delays in the system and it becomes more complicated. In such a case, setting the resupply level becomes crucial, as the inventory has to hold out until it can get replenished.

All in all, I would say that it was an interesting experiment.

Doing Distributed Agile?

Share and collaborate with distributed teams with our electronic agile board tools. Get all the benefits of electronic tools without sacrificing the benefits of physical boards. Supports Scrum taskboards, Kanban boards and user story maps. Check it out!

One Response to “Conference giveaways: A mini experiment in pull based production”

  1. Silver Stripe Blog » Blog Archive » Kanban for software maintenance Says:

    […] The system uses a pull-based system for moving change requests through the system. So when test completes a request, the request is taken out of its queue and placed into the release pool, leaving one space in the queue. At this point development finishes another CR to fill the gap, in the process creating a gap in their own queue, which in turn is filled up by Analysis. Once in a while (say every 2 weeks), a release is made with the fixes that are in the release pool. […]

Leave a Reply