Zvi Schreiber was like most other people: Interested in getting the occasional bargain or a gift for their loved ones, but he wasn't really knowledgeable about what it takes for goods to get to their doors. Like most people, Schreiber did not bother to learn more. He assumed there is some sophisticated mechanism taking care of business behind the scenes.
Like most people, Schreiber was wrong.
But unlike most people, in 2010, Schreiber was appointed as the CEO of Lightech, a company involved in electronics. As Lightech manufactured its products in China and shipped them to the US and Europe, Schreiber was up for a rude awakening. He soon realized there was no sophisticated IT system running the freight business.
Where others see problems, people like Schreiber see opportunities. Schreiber, a seasoned IT expert and entrepreneur, saw an opportunity. Fast forward to 2017: Schreiber is the CEO of Freightos, a 160-people strong global company that's raised about $50 million and is working on disrupting the freight industry.
Freightos has more data on global shipping than anyone else, and can offer some insights on this industry and the interplay between retail, data, and events such as Amazon Prime Day (APD).
Also read: Amazon effect: QVC buys HSN for $2.6 billion | Samsung is building its own Amazon Echo but faces a Bixby data shortage | Amazon just hit me with the worst tech support ever | Amazon adds five budget-friendly devices to its Prime smartphone lineup
By all measures, freight is big business. It's total worth is estimated at $30 billion per year. It employs countless people and affects practically everyone. The fact that getting a quote for a shipment would typically take over three days in this day and age is mind boggling.
If you want to fly for example, you just go online and you can compare prices and alternatives and book within minutes. It's been there for 20 years and it's a commodity by now. So, while it's clear that the opportunity for first mover to become the Expedia of freight is there, why nobody else did this before is not.
Schreiber points out that there were a couple of early attempts during the dot-com era, but like many at that time, they went bust when the bubble burst. Still, it's been a while -- how come there were no Freightos since? Schreiber believes it has to do with technological capacity, since the advent of cloud infrastructure makes collecting and processing data and running complex algorithms feasible.
Schreiber says that in order to become the go-to marketplace for freight, it has to collect lots of data from lots of different sources: "There are no standards. What we do is we ingest and process spreadsheets from all over the world. Tens of thousands of spreadsheets with thousands of rows and millions of data points each."
Where are these spreadsheets collected from? Two different types of sources. First, from carriers. These are companies that send containers overseas, mostly by sea actually. Schreiber says this market is dominated by a few large players, and it's traditional and secretive.
"It's very different from the tech world -- you will mostly see people in suits who have been with their companies for 20 or 30 years," says Schreiber. This may explain why the lag in technology as well. This is business that is capital intensive and often passed down to family, as not everyone has the capital, know-how, or connections to run it.
Then there are also the freight forwarders, quite a different beast. Freight forwarders (FFs) are the ones that take freight to the last mile -- from the port of delivery to the final destination -- using all combinations of transport modalities.
It's a hugely fragmented space, with about 100,000 companies operating, according to Schreiber. With such fragmentation and competition, perhaps nobody could amass the courage and resources or gain the trust required to act as the middleman.
Not much changes on Amazon Prime Day
Essentially, that's what Freightos is: A sophisticated middleman, a marketplace serving FFs and carriers. Freightos collects data from them, processes and stores it, and uses it to power its patent-pending routing algorithm. The end result is that users can now go online and see their options for sending cargo anywhere in the world in seconds, not days.
To get there, Schreiber says Freightos has been working with FFs for five years now, and in the last year, it have also been running a public marketplace. Getting hold of data is part of the masterplan, so Freightos made sure it framed its SaaS offering accordingly.
"In order to generate our instant pricing we needed to collect data from FFs and carriers and combine it," says Schreiber. "Nobody had ever done this before, and we realized that data is valuable in its own right." Case in point: Analyzing the effect of APD on global shipping.
"Most of the goods on sale on APD arrive on a ship from China 30 to 40 days ago," says Schreiber. Ninety percent of all products are shipped by sea, mostly from China, and Schreiber says Amazon sellers are Freightos' most significant customer, as more than 20 percent of shipments end up in one of Amazon's warehouses. So what can Freightos data tell us about APD?
According to Eytan Buchman, Freightos' marketing director, "Freight shipping is inextricably linked to holiday and retail cadence, we're always on the lookout for these types of patterns. On a global level, this is one of the largest drivers of changes in international freight pricing. We're definitely seeing an increase in importing leading up to Prime Day on July 11th.
We've seen a full 35-percent increase in shipments to Amazon warehouses on the Freightos Marketplace platform over the past month. As a result, Amazon shipments, which usually represent about 10 percent of the shipments on the Freightos Marketplace, grew into 15 percent of our volume over the past two months." It sounds like APD is kind of a big deal. However, you would not know just by looking at this chart.
First of all, Freightos has data about prices, not volume. According to Buchman, "Given how commoditized the market is, prices are a great proxy for supply and demand. Barring major external factors, the driver of supply/demand shifts is indeed volume. One great example is Apple famously dominating air freight volume when importing iPhones or Macs, which, of course, leads to a spike in prices."
And then, this chart seems to be at an all-time low before APD. If APD is so big, how can this be? Buchman says that, "Amazon's individual shipments, and especially those of FBA vendors, are still not a significant enough factor to impact freight capacity availability.
"That said, the current magnitude of importers is one of the reasons that Amazon is increasingly making forays into international logistics -- from registering as a licensed freight forwarder to scaling, it's Prime Air airplane fleet.
"Amazon's Prime Day placement is likely due to it being a lull in the retail cadence, like Alibaba's Single's Day. However, I still don't believe that Amazon and its third-party vendors are moving enough freight independently in order to significantly impact general US shipping rates. Given Amazon's growth trajectory, however, it's easy to envision a future in which it does."
Guerilla data marketing and the future of shipping data
APD may be big as far as Amazon, and the ecosystem connected with it is concerned, but it's a drop in the ocean in the bigger picture. So, what is the main driver in global shipping? Buchman emphasizes that "seasonality is one of the most powerful aspects, frequently overpowering other factors that you would think might have more influence, like sanctions on Qatar or shifts in oil prices.
"Christmas, Chinese New Year, and the three months leading up to them are the peak but larger retailers go into action 4 to 5 months before in order to begin the gradual buildup toward them. Black Friday typically is more about inventory liquidation than new importing, although they do likely have a more graded impact on import volume.
"Bank holidays typically have lower impact on rate changes because ports will continue to operate in China. The real blows come from longer holidays, like the Chinese New Year, in which all truck drivers head home to their families, leaving an extended period of time where manufacturing and shipments to ports/airports slow down."
This is quite a data-driven analysis there. You may not be surprised to hear it's not the first time Freightos is in the news using a similar tactic. It's part of the strategy, what you might call guerilla data marketing.
It's all about having the data and the knowledge to make sense of them, and acting on the right opportunity to bring your brand to the fore. As one of the venture capitals backing Freightos points out, it is prepared to get lucky.
So, what's in store for the future? All in all, the technical challenges Freightos has to deal with have more to do with data integration than anything else at this point. Freightos is an end-to-end Google shop, running on Google cloud and using infrastructure such as MemCache, Google Data Store, and Task Queues.
Schreiber says Freightos is looking into applying machine learning to be able to predict things such as price surges. At this time it would not be feasible to do this, as the data would not be representative. Freightos marketplace has only been running for a year, and it has been constantly growing, so it needs to amass more data to be able to use it for this purpose.
Freightos is also looking to integrate IoT data as it becomes available. When mentioning IoT, people typically think of millions of sensors everywhere sending loads of data. In the real world, or least in the one Freightos has to deal with, things are less dramatic, according to Schreiber. For most cargo, you don't need a sensor in every single package, and you don't need a whole lot of data either, as you can deduce information based on the carrier.
What could be a game changer for Freightos, however, is graph databases, as its routing algorithm heavily relies on graph. Schreiber explains that it could not use a relational database for this (in line with Emil Eifrem's tale), so it is processing the graphs in memory with a NoSQL database.
This is an interesting approach, which we will return to in the near future.
More on Digital Transformation: