The last week the blog post turned long. Sorry. This meant to just be short updates. Here’s another one.
The first thing I did this week was to create a diagram over our throughput; the number of items we’re getting done per week. And it looked pretty bleak. And was about to get worse… Here’s how it looked at the beginning of the week.
| |
|:——————————————————————————————————-:|
| 
|
| Throughput trend at the start of the week |
That’s not looking good, right. I think I can explain it… and I think that I shouldn’t care too much about data with these few data points.
A lot of the things that we are doing require third part input. Someone else needs to approve, read or work with the item before we can finish it. I’ve been working with a LOT of teams that have this situation. It’s a drag…
If you remember in the second post we created a special section of our Doing column called Tunggu (waiting). This is how it looks most days:
We have put a dot on our Reason-sticky to indicate how many days we have
waited. Yeah, that's 13 dots for the one on the left there. And we have
worked maybe 2 hours on that one....
Also, this week we started 4-5 new items that was Large or Medium.
Meaning that they will take some time before they are finished. And
accordingly, after this week our throughput looked like this:
| |
|:-------------------------------------------------------------------------------------------------------:|
| 
|
| Throughput at the end of the week |
Bleaker still... But this can be explained and teach us all something
about the relationship between lead time and throughput. And the danger
of just focusing on one metric.
We have just complete three things this week. Not surprising since we
started the board over more or less with 4-5 M to L items. There will be
another "ketchup effect" in a week or two. You should try to mix big and
small things to avoid this. Maybe we need a policy about how many L, M
that can be on the board at the same time.
But is this necessarily bad? Well, it doesn't say much about much each
item takes to complete. The lead time can be pretty good and still no
throughput. Since they all will arrive at the same time. Especially if
you do L items only... Another reason to strive for smaller things.
> Small things, arriving frequently and often is better than large
> things arriving seldom.
> Old Marcus proverb (since this morning)
Today on our "retrospective"... Actually it just a 10 minute "Ok,
people. How can we be better next week?"-kind of meeting. But today we
decided to add the date we started to work on the item to each card.
This is a very easy way to calculate the
Lead
time for each item. When the item reach Done (Selesai on our board)
we just subtract the start date from the current date and we get the
lead time.
And since we're tracking how much we're waiting for each item (in days)
we can also see how much time we have spent waiting.
> "It took 22 days, but we waited for 20... maybe we should talk about
> this"
Also now we have two metrics to try to balance against each other; we
want short lead time as well as good throughput. (And great quality, but
that is still to be measured) .
This very slow week thaught us a great deal.
>