The Goldilocks problem of packing
On a WhatsApp chat with my girlfriends today, one of them lamented about a T-shirt she ordered from Target that was shipped in a too large size box. Since I work for Target I jumped in.
You see those pictures and news articles all the time, people talking about how they ordered this single item and how it came in a package big enough to accommodate their whole body!
It's horrible of course. Don't retail companies care about the environment? (we do!). Don't they want to reduce shipping costs? (oh, yes!) And yet this happens.
How does one pick the perfect package every time?
I've read a lot about this and the problem statement is actually quite a captivating one since it is deceptively simple but has many variables increasing the runtime of calculation. To begin with, the problem is NP Hard. So one can only find a heuristic solution. While not going into the exact implementation details, here are some variables that could be at play:
- All item dimensions need to be accurate for the correct carton recommendation. This in itself can be a problem for logistics companies.
- Once all item dimensions are usable and accurate, the system needs to make a carton recommendation - should it use a volume-based solution or a stack first solution? A volume-based solution is simpler whereas a stack first would be more accurate. The best algorithm is a function of what assortment of items are most popularly shipped and the package size of each shipment. This means, in general, the idea would be to prioritize speed of recommendation and "good enough" over perfect since this is a heuristic solution.
The best algorithm isn't necessarily what's the most accurate but what's the most effective.
3. Once we find the most appropriate algorithm what type of carton should one use? Corrugated, plastic bags, mailers, the choices are many and it depends on the type of item. So it's not enough to know the dimensions of items but may be necessary to know the properties of the items being packed. In the case of the t-shirt, the system needs to know that the item is indeed a t-shirt and pick the appropriate carton type. It's obviously easier if you are a Myntra and primarily stock one type of items (apparel) and harder if you are a general retailer.
4. Once we find the best algorithm and we recommend the best carton, that carton now needs to be available to the packer. The manufacturing cost of bulk vs custom carton sizes needs to be considered. Even if we had found a perfect carton match, a packer may still override if they run out of that particular size of carton. So it is also a supply chain/inventory management problem statement.
5. In some companies, packers could override the system even if the carton is available because they feel they know better than the system. That could be true, for sometimes they do, that's why we are so obsessed with Artificial Intelligence, aren't we?
6. When the carton the packer selected is too large then they put the dreaded fill pack (plastic or bubble wrap) that makes our environmental conscience cringe. They do it because they don't want the item to rattle or squish and get damaged - you do want the item in pristine condition when you receive it.
As you can see, the problem statement is not that straightforward and one would make approximate decisions along the way to find the most optimal one.
So if the package you received is too big or too small and not "just right" you now know that something didn't go perfectly between the series of approximations.
How are companies such as Target solving the packaging problem in other ways? By introducing a one box solution such as Target Restock, or removing packaging completely from the equation by introducing Shipt.
As consumers, we can make sure we recycle (don't soil them during storage, as damp or wet corrugated cardboard cannot be recycled) And yes, try to group your items when you shop rather than buy a single item every other day, be mindful of the packaging and delivery costs to the environment.
The next time you tweet about how bad the packaging is, give it a moment's thought - it's a hard problem to get perfectly right!