Frequently-Asked Questions

Below we have summarized common questions and attempted to give guidance based on our experience and that reported by others. This FAQ concentrates on order-picking in a distribution center because we believe this to be the area of widest application.

Bucket brigades in the distribution center

Applicability

How can I know whether bucket brigades will work in my warehouse?

A nice thing about bucket brigades is that you can try it out on a slow day. You need no special equipment. We have found workers pick up the idea quickly.

In general, if you are currently using zone-picking to progressively assemble orders, you should consider changing to bucket brigades. If you are currently doing broken-case picking from flow rack with a pick-to-light system you are almost certain to benefit from conversion to bucket brigades.

Bucket brigades work best when pick-density is high and the items to be picked are small. Also, the waves of order-releases should be of sufficient duration to allow self-balancing.

Finally, to make bucket brigades most productive you should commit to building and sustaining a team-oriented work environment.

Why should I change from zone-picking?

Because it fails to balance the work as well as bucket brigades and therefore is less productive. Furthermore, because it is based on only a static idea of balance, it requires constant management attention to re-balance work during the day.

It is impossible to balance the work by a work-content model because work-content is only a guess based on time-motion studies and the assumption of some standard worker. It is always wrong because workers are not standard and the model cannot foresee the inevitable disruptions of the day. Furthermore it solves the wrong problem: Zones try to give each picker the same amount of work in total, over the day; but this does not mean that the work is balanced from order to order.

In contrast, bucket brigades spontaneously and dynamically reallocate work based on how long it actually took specific workers to pick specific orders. The means that bucket brigades get better balance, and therefore better productivity, and with less management.

What is likely to be the biggest difficulty in implementing bucket brigades?

Accepting the fact that you may not need the elaborate work-content and load-balancing model you built or bought.

Can bucket brigades work with paper pick-sheets? With pick-to-light? With RF devices? With in-the-aisle/pick-to-cart? ...

There are only a few essentials to support the effective operation of bucket brigades. One is that the handoff must be quick, which requires that a worker know exactly where she is and what needs to be done next when she takes over the work of another. We have found paper pick-sheets work fine as long as each picker clearly indicates her progress, such as by putting a check mark beside completed pick-lines.

Pick-to-light supports bucket brigades especially well because it is easy to see what needs to be picked next: Just look at the lights! This makes hand-offs very efficient. It also reduces any lost time due to one worker blocking another: The blocked worker can see what needs to be picked and help the blocking worker finish picking so she can move on, out of the way.

When picking with RF devices, pickers can simply pass the RF device with the work and scan their identify tags so the system knows who has which device. Alternatively, depending on how the system is programmed, each worker can simply scan their new work.

We have found that improvements due to bucket brigades have been less dramatic when the team picks in multiple aisles. Part of the problem is that pickers can lose team focus and intensity when they cannot see their team mates. Another problem is that, if you are doing in-the-aisle picking, you might not have sufficient pick density to support order-picking by bucket brigades.

Bucket brigades have been used successfully in case-pick-to-pallet operations (25% increase in pick rate); but there are some additional challenges. For example, some workers felt frustrated at having to take over someone else's partially-completed pallet because they had a different style of building pallets.

My customers average only (for example) 2.3 lines per order; can I use bucket brigades?

Possibly, but you must make some adjustments to increase the pick density so that workers do not catch up with one another often ("pick-density" = picks-per-linear-foot of aisle). The easiest way to do this is to batch orders. Under bucket brigades you can ask the first worker to start two orders (instead of one) or three or... . This is easy to adjust.

I tried bucket brigades and they did not work: pickers kept catching up with each other and getting in each other's way.

You must ensure that there is always work to be done between any two workers. There are two main ways to accomplish this: Batch orders (see above); or use fewer workers on the line (which could be compensated for by higher rate of production).

Management

Do the workers walk more in a bucket brigade?

The amount of total walking under a bucket brigade is no more than that under a zone pick operation, although the walking will not be split equally among the workers. To understand this first note that with a bucket brigade each batch is walked forward the length of the line, and this forward walking is split among the workers. And, under a bucket brigade the total length of the line is walked again after each batch is completed --- this walk back is also split among the workers.

For a zone operation the same holds. Each batch is walked forward the length of the line as each worker picks the order within her respective zones. Also, after each worker completes the picks of a batch within her zone she walks back the length of her zone; and thus, in total, for each batch completed the entire line is walked back.

The main difference between the two is that under a bucket brigade the walking is not divided equally among the workers, nor will the amount of walking for each worker remain the same from batch to batch. This leads to a misperception of the workers and management that there is more walking in a bucket brigade operation. Perhaps this is because workers remember more vividly those occasions at which they had to walk farther than usual; or perhaps it is because the faster workers will indeed walk farther in a bucket brigade. But this is exactly what should happen: The faster workers will take on more of the load.

Won't the workers complain about the walking?

Possibly. We have found that, after an initial period of adjustment (about a week), the workers prefer it. They like the rhythm and predictability of it: After picking intensely down the aisle, a worker gets a "break" by walking briskly back upstream, changing field of vision, and greeting his team mate as they hand-off work.

In any event, you can reduce walking by increasing the amount of work that each worker carries. For example, if each carries two orders then workers will walk half as far as if each had carried only one order at a time.

How do I handle underachievers?

We have found that peer pressure by team mates generally handles such problems; but you need to give your teams some authority to manage themselves.

In one implementation, the first, slowest worker never moved beyond the first section of shelf. (A nice thing about bucket brigades is that everyone's productivity tends to be reflected in the distance they cover.) Everyone could see this; there could be no dispute. The team captain asked whether that slowest worker could be removed from the team because the remaining workers could produce as well without her (and split the team incentive n-1 ways).

How can I track the picks of individual workers?

We recommend focussing on team productivity, for which a better measure is orders completed. Focus on individual pick rates risks rewarding the accumulation of work-in-process. Focus on order completions encourages teamwork.

Nevertheless, it is easy to track individual productivity. If picking from a paper list, just have each worker initial where she begins picking from each list. King-Way Order Selection Systems, IPTI and, probably, others have configured pick-to-light systems to track individual productivity under bucket brigade. IPTI offers a white paper describing their product.

How large should a team be?

In our experience 3-8 works best. As teams grow larger than 15 it becomes harder to maintain "team spirit" and so they lose that productive boost.

How should I pay team members?

This seems to be very site-specific and we have no authoritative answers. One scheme that has worked for others is to provide some incentive based on team productivity, to be shared equally among team members. This is in addition to components of pay based, for example, on seniority or skill level or quality or attendance. Another good idea is to pay the fastest worker a bonus to act as team leader.

How should I form teams?

Again, this seems to be site-specific. We have found that teams are more likely to be successful if they are allowed to form on their own. Unfortunately the best workers may tend to group together, leaving underachievers leaderless. In this case you may need to step in. In any event you must make it clear that each team is expected to help train new hires and temps (and you may need to protect any performance-based pay during such times).

Some sites have allowed teams to interview prospective new hires. This encourages them to take a stake in getting new workers quickly up to speed.

For what it is worth, in our experience females seem to perform better under bucket brigades than males.

Operations

How do I start up the line in the morning? How do I close it down?

There are many ways. One that has worked well for us is the "shotgun start", which we used at a site where each picker carried four orders: Initially each picker carries one order; at the next walk-back the first worker introduces two orders, until everyone has two orders; then three orders; and finally four orders.

You can close the line down in a similar way. For example, when the first worker introduces the last orders she can signal to the last worker (team captain); then at the next walk-back the team captain can take only, say, two of her predecessors four orders.

What happens if some worker other than the last first finishes an order?

The worker who has finished an order can push it off and start a walk-back. If it is required that orders be completed in the order in which they were released, the picker can push it down to her next team mate to carry along.

Sometimes a worker catches up with the next worker and then is blocked; isn't this wasteful?

Yes, but this should be only temporarily, while the line balances itself. If it persists, then either the workers are not slowest-to-fastest; or else you may have too much variability in work. Reduce variability by asking each worker to carry an additional order or more.

Incidentally, if you are using a pick-to-light system then the blocked worker can help the busy worker finish picking that section of shelf so she (the busy worker) can move on out of the way.

Will I have to re-warehouse my skus?

Probably not, unless your pick density is very low and batching orders is not an option. In this case, you could move your most popular sku's to the start of the line.

Miscellaneous

Can I see bucket brigades in action?

We have videotapes of bucket brigades picking from flow rack, guided by a pick-to-light system. Contact us to borrow a copy. You can also watch a video of order-pickers operating as a bucket brigade. (Note that you will see the workers constantly pressing buttons: This is to communicate with the pick-to-light system; it has nothing to do with bucket brigades.) Finally, you can also run a simulated bucket brigade with real people. Our clients and students have found this exercise very instructive and are generally quite surprised by the results. Here are instructions and materials.

Can I use the Java applet from the Bucket Brigade Home Page to illustrate bucket brigades in my presentation or report?

Sure. Here is a version that you can clip, modify if you wish, and run within a web browser. It shows the distinctive pattern of movement of workers in a bucket brigade.

Bucket brigades are not suited to situations like this, where there are few picks and multiple possible paths.

When do bucket brigades not work?

There must be sufficient pick density to ensure that workers generally follow the same path through the warehouse. They are not well suited to situations in which sparsity of work and warehouse layout allow multiple possible order-picking paths through the warehouse. In such case, it is not clear which worker is ahead and which behind and so bucket brigades are not well-defined.

What is the difference between bucket brigades and the Toyota Sewing System (TSS), also known as "bump-back sewing"?

Bucket brigades are similar to TSS in that the system pulls work at a pace set by the last worker. However, bucket brigades improve on TSS in two important ways. The first difference is that bucket brigades abolish zones altogether, while TSS workers are restricted to static zones with a little overlap, as you can see in the following snippet of advertising video for TSS. Abolishing zones gives the bucket brigade maximum capacity to compensate for disruptions, variability in work content, etc.

A second improvement is that bucket brigades sequence the workers from slowest to fastest, which accelerates each product through the assembly line. In contrast, none of the TSS sites that we have visited nor any TSS consultants that we interviewed acknowledged sequencing workers from slowest to fastest. (We did find one who sequenced workers shortest to tallest!)

These two improvements are the key that makes bucket brigades self-balancing.

Anything else?

To repeat an important point: We have generally found that single most determining factor in the performance of bucket brigades is the pick-density. If it is too low, bucket brigades will not be as effective as they could be (though they may nevertheless be more effective than your current operation).