Visioning: Customs OS for Trade
“Any good broker would verify with the data. I wouldn’t just assume.” - Flexport Broker
Customs is the most deep and edge-case critical vertical at Flexport. It poses the greatest compliance risk to Flexport, and is the most profitable revenue generator for the business.
We must file customs documents accurately and on-time to avoid penalties, and provide the best customer experience.
At it’s core the work of customs is knowing what data is required by which government agency and providing it in the right format. This is the expertise of the customs broker, and their work is supported by the customs product team with the goal of increasing operational efficiency, accuracy and customer experience through building the proper tools.
The Result: An Envisioned Workflow
Self-service Classification Prototype
Our User: Customs Brokers
“Quality control is everyone’s job.”
Matthew is a certified Customs Broker based in Flexport’s headquarters. He worked in the logistics industry for a few years before starting at Flexport. Matthew is one of the more experienced Custom Brokers so other brokers at Flexport often turn to him for advice. Providing guidance and mentorship to his colleagues is one of his favorite aspects of the role.
1. Be compliant with U.S. Customs.
2. Clear Flexport shipments in a timely manner.
3. Help with anti-dumping or other special Customs cases.
4. Onboard new Flexport clients (only done by certain Customs Brokers).
1. Often has to fix information incorrectly entered upstream.
2. HTS research and client communication time consuming & can delay shipment releases.
3. NetCHB is slow & difficult to use yet the Custom Brokers must use it for many key workflows.
Onboarding, Research and Alignment
We were a scrappy four member crew (soon to be 11) focussed on maintaining the system and protecting Flexport’s business by supporting compliance. Customs tools were four years old and had received no significant upgrades since then.
Doing some preliminary broker shadowing revealed that brokers had a broken workflow across multiple screens, tabs and apps that decreased accuracy and filing efficiency. There was a lot of room to grow.
Before proposing an immediate solution to fixing the issues, I proposed we create a product vision. This would give us a future landmark for working towards and align product, design and engineering to the best possible user experience and business outcome.
The customs broker experience is fragmented
“It’s not a logical straight through process. So you almost are tricking core to perform with what you want to happen in netchb… if brokers aren’t careful…you’ve created an incorrect entry”
Our app wasn’t designed for customs brokers, there isn’t an end-to-end experience and so they’re forced to find workflow hacks to complete their work.
This lead to a big future project codenamed breakout, it turned into a company-wide initiative to leave the monolithic software of Flexport into a specialist experience. LINK
Classification takes a lot of time because clients need to provide more info
“You really have to know what you’re asking, if they dont give us information we have to go back to them [clients] probably more than once. Even with these questions [written up] I will need to go back to them depending on what their answers are.”
Brokers spend a lot of time with clients back-n-forth communicating with brokers about the intended use of the products, their material and where they came from. It’s one of the most costly processes to our business because it has dependencies on the client’s ability to source information and get back in time before the shipment arrives or else face penalties and even worse a customs hold.
Upstream data issues creates more work downstream
“Im going to go through all of these [docs]. I’m checking the MBL and HBL and make sure they’re keyed correctly and I’ll check AN… It’s our job to make sure that data is quality.”
Since the accuracy of our data is sometimes questionable, it requires a lot of double checking by brokers because they are the gate keepers to what gets transmitted to US Customs, and are held responsible with the business if things go wrong.
There are several points at which data enters, first with the client uploading docs and then data ops keying in the data. There’s also lots of communication overhead with brokers checking in with clients for other information.
Visioning of a framework for customs
Goals of Vision
A Product framework
Evolving the Flexport customs broker role through automation
Client experience and automation are a big part of Flexport’s value add. This is a proposal for what a Flexport broker’s role in the future might look like with right products in place to help.
Phases for Flexport customs self-service automation
After research identified the problem areas, I divided the workflow into it’s product components so we could envision how everything fits together like lego. This provided a birds eye view and a thorough roadmap on how we can get to an excellent customer experience that is automated.
Yeah, it’s a lot. The TLDR is this a lego piece roadmap of how all the parts come together for automation.
This accounts for clients, brokers, compliance and data ops.
We are entering phase 1 capturing the workflow.
Thinking Big with a Design Sprint, Prototyping and Testing
My recommendations served as a jump-off point for a design sprint to ideate, prototype and validate 10x improvements for the customs product future.
The challenge was to enable a client to complete a standard US customs entry with no human intervention.
Our secondary goal was to improve data ops workflow, because let’s be honest typing in documents all day is brutal.
The current data entry method also introduces a lot of data quality issues which lessened broker trust downstream. This led to a lot of re-checking data ops works by brokers. Ideally data would be entered and verified once and not re-entered or re-verified.
Building, testing and validating prototypes
Prototype #1 - Improving Data Ops Workflow
A key area for improving efficiency for customs brokers is improving data quality and efficiency upstream. Data ops manually keyed in commercial invoices which are essential documents from clients and manufacturers that are used as source material for classification of products for US customs and filing info. The trouble is that it takes a lot of time for data ops to enter data and there’s a 3% error rate because its manual. Data accuracy is critical to filing with US Customs.
Automation of data entry would improve their workflow, and shift their work from data entry to data accuracy. This would reduce time for brokers re-checking data ops work too.
Testing this product with data ops was a breeze, they really liked the new dashboard we provided and dragging and dropping the data into a spreadsheet was described “like a game” by one of the users.
This prototype helped us validate that ingesting data from the commercial invoice would be well received by data ops with caveat that they wanted to trust the accuracy of the data. This lead us to a future project where we introduced an OCR (optical character recognition) tool to their workflow.
Prototype #2A - Client self-service
Brokers spend a lot of time with clients back-n-forth communicating with brokers about the intended use of the products, their material and where they came from. This lets them assign the correct HTS code and tariff rate. It’s one of the most costly parts to the business because it has dependencies on the client’s ability to source information and get back in time before the shipment arrives or face penalties and even worse a customs hold.
In this prototype we tested customs as a self-service - the turbo tax for customs so to speak.
We wanted to know from prospective clients
1. Would they be willing to enter their product data before signing up?
2. Is offering classification as a self-service was appealing
3. Is learning the additional costs before signing up a value add?
4. Would they add freight to their customs-only shipment?
One of the great mysteries of customs when contracting with a freight forwarder is the customs costs of shipping the goods. Certain products have enormously high tariff rates such as pencils.
We wanted to provide a simple tariff calculator once the HTS code was discoverd to see if the value prop would be high enough.
We learned a lot from clients
There’s no user friendly way of searching tariffs, I suck it up and do a find through the long list. This [prototype] is a million time improvement compared to that. You get the answer of what you’re looking for, what is the HTS code and what you have to pay to bring those products in.” - Stephan - Client 1
“AHH EXCELLENT” - Client 2
The client-facing prototype allowed us to test and validate our big ideas. Some of our theories that they might reject for too many questions or not be able to complete the information was invalidated by how impressed clients were with this new product offering.
They were very excited if they could classify on their own. Along with validating that self-service as a future for customs is the way to go, we also learned a lot more about clients.
1. Would they be willing to enter their product data before signing up? Yes!
2. Is offering classification as a self-service was appealing Yes!!!
3. Is learning the additional costs before signing up a value add? Yes!
4. Would they add freight to their customs-only shipment? Yes!
Full Visioning Deck
Planning and conclusions
This was my ramp up project to joining the new customs team. I’m thankful to my project manager Justin. He gave me the breathing room to research and learn about customs first-hand so I could address root issues rather than start chipping away at superficial problems.
I went from customs newbie, to becoming an expert about the customs experience and talking to clients. I helped the customs product team craft a vision for a future through collaboration, ideation, and validation with research, and prototyping.
This process gave us a common language and context to align and start planning building out our future customs app experience.