LDRDocs/Processes/Outbound Methods and Flows.md

# Intended Methods for Outbounds

#intendedProcesses #opinionated

## Foreward
This document is highly #opinionated and makes some assumptions. It might be a little ranty.

This article applies mostly towards [[Dock Team]].


# Overall Goals
The overall goal of this process is to ensure the following things:

- We have fewer than 1 complaint about a shipping error per 1000 orders we send out.
- We have fewer than 1 complaint about a shipping delay per 100 orders we send out.
- Customers are proactively notified about problems before they complain.

## Shipping Error Prevention
A shipping error is defined as a shipment never reached it's destination, arrived at the wrong place, arrived damaged due to improper packaging, or included the incorrect contents.

Ideally, the following five quality checks are in place around preventing shipping errors.

- [[#Pre-Pack Quality Check|Pre-Pack Quality Check]] - Make sure each item is the correct item before packing.
- [[#Post-Pack Quality Check|Post-Pack Quality Check]] - Make sure each box was packed and labeled correctly.
- [[#Pre-Load Quality Check|Pre-Load Quality Check]] - Make sure the previous step was done and all packages are accounted for before loading any trucks.
- [[#Post-Load Quality Check|Post-Load Quality Check]] - Make sure everything was loaded that should have been, and nothing we weren't supposed to load.
- [[#Next-day Followup|Next-day Followup]] - Check to see if any packages we sent out still don't have carrier updates and contact the carrier about it.

### Existing methods and tools (carrier pickup, log truck, etc)
At the time of writing, the reader should suspend their understanding of what these different tools are called or used for. This section will instead focus on explaining what they're supposed to accomplish.

### Pre-Pack Quality Check
We need to make sure the right inventory is picked and delivered to the packline. This is the reason inventory is supposed to be handled strictly by barcode. As the inventory moves to the packing location it should be scanned to update where it's at and who put it there. The electronic pick tool is one way to do this, but it's not the only method.

#### Failure Modes
This section will outline the things most likely to go wrong during the Pre-Pack Quality Check and either what to do if they happen or how to prevent them.

##### Missing Item
The most common failure mode is the item can't be found. However, that isn't a failure of this process.

When this occurs the picker should not spend extra time searching for the item because doing so will delay other work. Searching for missing items is like gambling: we might find it, but there's a high chance we won't. Meanwhile, continuing to do other work instead is guaranteed money.



##### Long picks and unattended carts
Somebody is picking to a cart or bin, but they don't do it continously. The cart/bin gets left alone for periods of time, ranging from 5 minutes to several days. In the worst cases, the stuff being picked isn't being updated to the cart or bin it's actually on.

There are many reasons this can be a problem.

Empty carts and empty space are a contested resource. Having carts around, especially blocking aisles, frustrates other people who might be tempted to move them around, move stuff off them, put extra things on them, etc. Whoever does that is probably in a hurry, so there's a good chance they're not updating locations or being very careful. On top of that, the carts can get bumped into and stuff falls off... then people see it on the floor and put it back in the wrong place without updating it.

There are various ways to avoid or mitigate this problem. The simplest method is using a wave approach to picking.

Using a wave approach, the picker doesn't try to pick everything or fill the cart. Instead, they pick just enough that can be processed in a short time frame, like 15 or 20 minutes. Then they deliver everything to the packline, clearing off their cart in the process.

### Post-Pack Quality Check
During this step, a person is checking that each package was packed correctly, and then sorting them by which carrier is picking them up. They are supposed to be checking for the following types of problems:

- Package not secured sealed closed;
- Holes or tears in the package;
- Multiple conflicting shipping labels, spals, or other stickers that could confuse other people;
- Package is appropriately labeled with any hazard placards like "lithium-ion batteries", "flammable", "team lift", "fragile", etc.

As they do this, they should be updating the location of where the package was put. That step acts as a confirmation/record this process was done, when, by who, and where the packages are when they finish.

This step is intended to be done more or less simultaneously with packing, where a different person is performing this check within moments after each order was packed.

At some points in the past, we tried doing this where the packer doesn't actually seal the package closed. Instead, the person doing this step does that. That's so this person can see what is inside, choose appropriate dunnage or bracing, and apply the right placards to the outside of each box.

There is no single clear reason we stopped doing this. Mainly, this was something the CEO came up with and did with one other person during 2019. That person ended up quitting, and the other people at the time refused to do it this way because they said it takes too much time. #opinionated We had only 1 complaint about outbound shipments prior to that, and it was a day the CEO was packing orders by himself and accidentally shipped an ammo canister along with a motherboard for an RMA.

### Pre-Load Quality Check
This step is to make sure everything we're going to hand the carriers is accounted for and the other quality checks were done. If any package isn't accounted for, we have to account for it before loading anything with the carrier.

Here's one way to do this:

If the "Carrier Pickup" tool is used to perform the "Post-Pack Quality Check", every package that passed that step will be sorted by carrier and location (cart). There's also a list of any packages that didn't pass that step.

To use this, someone would perform the following actions:

1. A deadline is chosen for each carrier. At that deadline every day, all orders for that carrier stop being processed.
2. Check the list for any packages that haven't cleared QC. If there are any, try to account for them before the carrier arrives.
3. If they can be accounted for, great. If not, none of the packages can leave the building until they're all accounted for.

As this is done, each bin, cart, or bag for the carrier is moved away from the packing area and put in a designated spot that doesn't change. Each one is visibly marked to indicate what carrier it's for and who completed this step. After that, nobody is allowed to put anything new on those carts or in that area.

### Post-Load Quality Check
During this step, the people loading each carrier are making sure they only load the right things for the carrier. They are documenting what they loaded, who did it, and when. They are also supposed to document anything about the pickup that may indicate a problem. For example, weather conditions, unusual drivers or vehicles, and that sort of thing.

They are supposed to compare a record of what was intended to be shipped out vs. what was actually shipped out.

### Next-day Followup
We've only recently gained the capability to do this, so it hasn't been tried before.

Basically, any package we give to carriers should have an update to it's tracking status within 24 hours. We're now able to check that. This probably looks like somebody checking that list the following morning and taking action if they see any packages that indicate we shipped them but the carrier has no updates.

The appropriate action to take is unclear. If the rest of this process was followed it's highly unlikely we didn't give the package to the carrier. Therefore, taking action probably looks like emailing the carrier a list of the packages and asking if they can search for them. It also may include emailing affected customers (before they complain) to let them know what's happening.

## Shipping Delay Prevention
Our customers tell us that shipping delays are a major problem for them, especially when they aren't notified about the delays ahead of time. This section will cover how to deal with this.

For purposes of measuring this process, a shipping delay counts as a complaint when a customer contacts us to ask why an order hasn't been shipped out yet, and the characterists of the order fall within an agreed SLA. It doesn't count as a complaint if the order doesn't fall into an agreed [[SLA]].

### Default SLA for Outbound Orders
This section talks about the default level of service for outbound orders.

In general, we can think of outbound orders in two broad categories as [[Fulfillment]] or [[Definitions/Prep]]. Please read carefully, because it's easy to misunderstand these terms. These terms are covered in more detail within their definitions pages.

Broadly speaking, fulfillment means collecting, packing, and shipping items without changing anything about the product. Prep refers to making changes to the product before it can be fulfilled.

The confusion here is that sometimes customers place orders that need to be [[Definitions/Prep|prepped]] before they can be fulfilled.

### Fulfillment Timeframes

|   Order Type and Size                   | Timeframe                                          |
|:----------------------------------------|:---------------------------------------------------|
| Single item or single package           | 1 business day; same-day if placed before noon     |
| Single-SKU order; 100 units or less     | 1 business day                                     |
| Multi-SKU order, 100 units or less      | 2 business days                                    |
| 100+ items in same order                | 5 business days                                    |

#### Expected Rates of Work
Each of the following steps of the work should take a maximum time of 60 seconds per item involved in the shipment and they should be done simultaneously.

|  Step            |  Max Time (@$15/hour) | Max Time (@$18/hour) |      |
|:-----------------|:----------------------|:---------------------|:-----|
|   Pick           | 30 - 60 seconds       | 15 - 45 seconds      |      |
|   Pack           | 30 - 60 seconds       | 15 - 45 seconds      |      |
|   Quality Check  | 30 - 60 seconds       | 15 - 45 seconds      |      |

This table assumes employees working 8 hour shifts with breaks, and an average pick/pack fee of $2.50 plus an additional $0.50 for each item in a shipment after the first.

##### Explanation of Simultaneous Flow
The intended flow means the picker, packer, and QC person are each different people working simultaneously so within the span of a single hour on the clock, a total of at least 60+ outbound items have been processed.

#### Capacity
When the work is being done at the expected rates the following table represents the number of items a single packline should be able to process during an 8 hour shift with employees who make an average of $18/hr.

|   Min Items   | Max Items     | Mean Items      |
|:-----|:-----|:-----|
|  158    |  294    | 236     |

### Prep before Fulfillment
When a customer places outbound orders that require [[Definitions/Prep]] before they can be fulfilled, they are supposed to be picked according to the same timeframe as fulfillment.

Either once picking has begun or the order can't be picked within the expected timeframe, the customer is to be contacted to let them know when they should expect the prep to start. Once items have been picked, the customer should be updated about this to let them know when they should expect the prep to be completed.

### SLA Failure Modes
This section talks about what can go wrong during pick that can lead to a complaint.

#### Not enough people
When there are not enough people to properly run a packline it is known to cause delays as well as lead to shipping errors since it's more likely steps will be skipped.

With fewer people the line can't run simultaneously so the total clock time it takes to complete outbounds correctly is longer unless people skip critical steps designed to prevent errors.

#### Missed orders
Sometimes the team picking/packing orders will overlook an order. They don't realize it exists, so they don't work on it. The most common contributing factors to this failure mode are explained below.

First and foremost, many employees check the list of outbound orders by searching for particular customers. That causes an issue where it's easy to skip over or forget, and means orders from new or 
infrequent customers will probably be skipped.

A second failure happens when employees forget to check the list of orders or they make assumptions about who is working on them.

#### Stagnant bins and material handling equipment
Sometimes we can end up a shortage of bins, rolling racks, and other methods of moving things efficiently. When that happens it makes any form of picking slower and more error prone.

There are two common reasons this happens.

One is a failure of the [[STOW]] process, resulting in inefficient use of storage space. The second is Long Picks with unattended carts, as touched on [[#Long picks and unattended carts|here]]. That's where someone starts a pick or other collection of inventory that can't be finished in a reasonable period of time, so multiple carts or bins are occupied until the long pick is finished.

#### Damaged / unmaintained carts
Carts of various kinds can become damaged from misues or lack of maintenance. When that happens they are harder to stear or move around, making movement inefficient.

#### Blocked aisles and debris on the floor
Aisles blocked by static projects and inventory, power cords, and other junk on the floor can cause delays in this process.
# Accountability & Observability
Here are some #opinionated ways to add acountability to this process.

The general problem we've had is that most people involved in this process aren't effected by failures. Therefore, they are less motivated to prevent them. For instance, when the CEO was packing outbounds himself he's the one who loses money if we make a mistake. It effects everybody else too in the form of lost profits, but not in a direct way most employees can identify. The other thing about this is that it's very hard to pinpoint exactly who messed up when an error happens.

Therefore, to put accountability into this process we probably need to do two different things.

First, if a step was skipped or modified without approval, whoever is responsible for that step should be held accountable. This is regardless whether or not the skipped step definitely caused a problem -- since skipping steps without approval is known to *cause* problems in general.

Second, each step needs to be made [[Observability|observable]]. That means it should be possible for just about anyone to quickly observe whether or not the step is being done without needing to ask anybody.

## Acountability for Outbounds
At minimum, there should be some record of who was in charge of each of the 5 quality steps each day. That record should probably be kept for 30-45 days, since complaints about shipping errors are sometimes received up to a month later.

### Pre-Pack Accountability
Perhaps the simplest way to handle this would be assigned picklists, since that will keep some record of who did them. However, that's not very observable because checking that means special training or skills. It also opens the door to claims about software bugs and other excuses.

A more reliable way to handle this might be making some pre-made KanBan cards for picking. A limited number of them exist, and they are easy to identify from a distance. Here's how they would be used:

- Each time someone wants to pick, they go to a person who distributes these cards.
- That person helps them fill out the card, and also keeps a record of who has the card.
- The card has information on it like:
	- Who it was assigned to
	- What they are picking
	- The time they took the card

The card is hung on whatever they're using to pick. That way, it's super obvious what that cart/bin/whatever is for. If a person hasn't returned their card(s) by the end of a shift, they're held accountable for it in some way.

### Post-Pack Accountability
This could be handled by a pre-made sticker or something. Basically, as each outbound cart is loaded and staged for pickup, some kind of report is attached to the cart that says who did the QC, when, how many packages, etc. We could probably make this something that can be easily printed from NMI.

The QC person in this step is probably the one who attaches this report. Then, the other person who stages the racks in the outbound area checks each one for this sticker/paper before moving it there.

If it has the sticker/report, they sign off their initials and a time, indicating they actually checked. If it doesn't have the sticker or there is some problem, they probably push the cart back to the packline it came from and make a QC Report about it or something.

### Pre-Load Accountability
This step overlaps a little with the previous area in terms of attaching some kind of report to each rack.

A reasonable thing to do here might be creating some kind of reporting mechanism to log that someone actually did this step. Like a list of what locations/racks were checked, by who, and at what time. The report would also ask how many packages were missing, if any.

### Post-Load Accountability
This probably looks like collecting all the reports from each rack that gets loaded on a truck... and maybe even asking the driver to initial them. The reports then get filed somewhere by date and retained for 30-45 days. That way, if there is a complaint from a customer we can easily find out details regarding what happened that day and who was involved. It shouldn't require a programmer to write any database queries. It's just checking a filing cabinet.