I've been advocating the use of this technique for some years now. We used
it first two a couple of years back. It's quite simple, very powerful
because it emphasizes the relative estimation much more than basic planning
Here's how I've facilitated it to several teams:
A) Before you start using the technique, you first create a the buckets
with examples (a scale).
For example just start with two items and ask which one of them is bigger,
then add a third item and ask whether that is the same size, smaller, bigger
or in between of those two items in size. We've used buckets in sizes of the
Fibonacci sequence (just like planning poker cards). BTW when you do this
exercise it is better if you have the items written on index cards and
create the initial scale on a table - this way the items are easy to move
and the scale is easy to create.
When you have the buckets ready (usually 2-3 examples per bucket). It's
time to create the physical buckets. I usually have printed out papers with
a box within them which we can use in the backlog estimation meetings (part
B). I have attached the template that I use to create the buckets.
B) When estimating backlog items, bring the scale with you. Also print or
write the items to be estimated on index cards. I really prefer to do this
physically, because a very nice characteristic of index cards is that they
are easy to move, sort and read.
Then, one by one, assign each new item to a bucket based on the items and
examples that already have been assigned into the same bucket. It's very
easy and fast to do the comparison. ('Yea, I think it's about the same size
as these...' or 'It's somewhat bigger then these, so let's put in the next
Assigning to the buckets can be done quickly by just comparing and
discussing OR you can do it with planning poker to facilitate also
discussion about the size. Use the latter especially with new teams.
After the estimation session (or couple of times during the session if it
is a long one - say half-a-day or more) go through each of the buckets and
see whether the items really are of the same size. If not, move the items
that you think are bigger or smaller to the corresponding other bucket.
Anyway the advantages of this in my opinion are:
- it facilitates relative estimation better than planning poker (where you
have to 'remember' what estimates you have been given)
- it prevents the estimation inflation or deflation (because the examples
don't move from bucket to bucket - with planning poker estimates sometimes
inflate or deflate with time, because they are based more on your personal
feel of the numbers and their sizes)
- it's very fast
- items are sometimes not discussed as thoroughly. Especially if you do not
use planning poker with the bucket estimation, might not discuss the
differences in opinions about the item.