Supply Chain Visibility Software Review: Panjiva



This article from early 2010 provides a review of a (then newly emergent) supply chain visibility solution, called Panjiva. This is a software-as-a-Service (SaaS) solution which focuses on global trade visibility for channels into the United States of America. I summarized what the solution is about, gave a user impression, and then used an early version of the visibility framework to examine where value-add was occurring. The article ends with a summary of where Panjiva was strong, where it was weak, and towards where Panjiva might likely grow in the future.


Panjiva is a Software-as-a-Service (SaaS) solution for extracting and using intelligence about global imports to the USA. The application is hosted on a publicly accessible website, and isn’t blocked by the Chinese great firewall. Users navigate to the site and then use their membership credentials to log in. Memberships vary in cost, but as of this article’s date there were three plans. The first plan is basically a trial or sample, it is limited enough to not be useful for daily operations. The second and third plans allow one or unlimited number of users for $400 and $1,000 USD per month accordingly. I’m not user of the single-user plan has a “named-user” license arrangement, but it’s likely.



Once a user logs into Panjiva, they are now using a tool which does business intelligence tasks on data concerning global imports to the USA. For those who are new to USA import standards, there is a rich data set available from several US government agencies (now unified under Department of Homeland Security). Obtaining the data directly from the government is free and straight forward, but a bit laborious. Before a user logs into Panjiva, that US government data has been imported and is the bulk of the data the user is about to mine for intelligence. Beyond just loading the data, which would just be text and dates, Panjiva has identified a set of information objects and loaded the data accordingly. For example, company names aren’t just text; they are represented as an informational object (a company) which has various attributes such as country of residence, aggregate volume imported, etc.


Panjiva uses a very well designed user interface, something similar to Google’s various sites. In fact, I wouldn’t be surprised if there was some shared underlying technology, such as in line-graphs. What a user first does, when entering Panjiva, is to “search”. It’s like opening a new Google search window, only now instead of searching for movie reviews or tomorrow’s weather we are only searching through a data set concerning global imports to the USA. At my current job, I don’t manage international imports to the USA. But, formerly I did manage international imports for the company Build-a-Bear Workshop. So, for the next few examples I’ve taken on the perspective of my former role.


To begin with, I searched for information about my own company (Build-a-bear Workshop) since most first-time users will be most familiar with that data and it offers a chance to start evaluating accuracy by comparing my insider knowledge to what Panjiva will say. What I get as results are a summary of the company (or a list of companies that meet my search criteria), and from there I can drill in further. What I drill into is, again, a dataset concerning international imports to the USA. So, for example, I can see the top suppliers of Build-a-Bear Workshop, but only among their international imports. If a supplier is domestic (within the USA), or if the supplier imports to the USA before exchanging ownership, there will be no records here. I should point out that there are also no records of air freight shipments or trucking from Mexico or Canada, since that data is not captured by the US government agencies in the same way. All we see are ocean freight imports.


From the beginning of the search & results experience I’ve noticed a very fine attention to detail in the user interface experience. Panjiva’s done a great job in making its web experience rich, consistent, fast, and intuitive. I did not have to undergo application training of any kind to get started. Once I start looking through the search results a natural pace of “surfing” behavior kicks in. From my initial search, where I looked up my former company, I am now clicking through various aggregates or drilling into specific shipment details.

For example, in the image above I have drilled into a specific Build-a-Bear Workshop supplier and am editing how the supplier’s shipments should be summarized onto the web screen. This kind of flexibility is a great feature in Panjiva and will drive up adoption and trust rates. Beyond Surfing The Panjiva application goes beyond the information surfing or scanning I discussed above. Users can switch from looking at companies to looking at products or product classifications. Since Build-a-Bear Workshop is an experiential retailer of children’s toys, I decided to lookup the “toy” imports. What does this do? It allows the user to see the same level of detail on competitors or wholesalers as what they can see about themselves.


For example, Toys R Us and Hasbro both play key roles in the US in this industry. By starting with the product, rather than the company, I can find who else may be a player and their related activity. I can also see who their international suppliers are. In industries where supplier source is a major part of the go-to-market competitiveness, this can be critical. I can group companies into aggregates, either for myself or for a peer user. I can flag companies, or add notes for later use. The flagging may also be done by Panjiva itself. For example, one company which acts a supplier to Build-a-Bear Workshop (Arbor Toys) was red-flagged automatically because a sudden drop in volume being imported to the USA. This kind of intelligence creation is good, especially since Panjiva’s user interface shows how the alert was calculated (a 50% or more drop in volume based on 3 month moving average compared to the same period last year). The image below shows the red flag.


Finally, the Panjiva application will produce email alerts based on certain events. The events are fairly generic, such as updates to a company being followed, changes in an industry or product, etc. At this time I would not rate the alerting functionality as very good, since it should incorporate extensible logical operations using the existing informational objects.


I’ll recap here the Panjiva application’s effectiveness using the visibility framework.


Panjiva’s data sources include the following:

  • CBT Filings (Homeland Security Registration)
  • US Customs Data for Ocean-Vessel Arrivals
  • US Credit Ratings (extra cost)
  • Sinosure Credit Rating (extra cost)
  • Notes from the user, or user’s colleagues
  • Documents attached by the user or user colleagues

Overall, the Panjiva visibility process is sensitive to US ocean imports, and the companies who declare themselves the sellers and buyers of these shipments. Data reliability is fairly high, especially given the data volume being discussed. Data timeliness is okay, perhaps with a one month to two month delay. Data accuracy is okay, but can be misleading. For example, a declared importer may be acting as an agent, a sourcing business unit, a parent company, etc and this effectively blocks aggregating on importer’s name alone. Integrating other data is necessary, but not being done at present.


Panjiva provides good accessibility and interconnectivity between data points. In particular, the accessibility level is driven higher by:

  • data quality check prior to loading
  • existence of informational objects, such as companies, countries of origin, ports, units of measure, etc
  • Connection of free-text comments, government data, and exterior credit ratings at appropriate levels (shipment, product, company, etc)


Panjiva’s main value-add is around intelligence. The application provides value by focusing attention on data relationships of value and bypassing those which are of little potential value. One interesting point is that the licensing structure of SaaS is well aligned to Panjiva’s value-add. Each month (or day, perhaps) the Panjiva team is augmenting application’s delivery of intelligence by behind-the-scenes data transformations. The work of sourcing data and converting to intelligence is ongoing work, which probably involves human expertise. Switching to a Software-as-an-Asset licensed model would not make sense in this context.

Key intelligence features in Panjiva include:

  • Rating system (0-100) based on red-flagging of KPIs. These include order volume compared to moving average, credit rating, peer feedback,
  • Grouping of individual companies into blocks for analytical purposes
  • Comparison of companies in a side-by-side lineup • Aggregation of data based on industry, country of origin, etc.
  • Geospatial mash up of a few volume metrics (shipment count or weight) by country


Panjiva’s decision interruption capacity is good, but not great. Whereas the intelligence cr

eation is substantial, especially considering the limited dataset breadth, I think the ability of the Panjiva application to interrupt decisions for different and better outcomes is not yet well positioned. It’s also not terrible. But the list of decision interruption features below is characterized by re-activeness, depending on an engaged decision maker and swivel-chair integration into other systems or processes. If the user is not pro-active, the Panjiva intelligence will not enter into key business decisions.

  • Alerting based on:
    o “updates” on companies being followed
    o competitor activity
    o product-type activity
  • Very good user interface, meaning…
    o Intuitive usage
    o transparent data source, such as a “graph methodology” link to show the underlying representation method
    o fast response time
    o rich graphical options, both for dynamic screen manipulation and for saving graphics for presentations
  • Potential for peers within a company or BU to collaborate, by sharing notes or documents.


The Panjiva application is currently very strong in its creation of intelligence (given a limited data scope) and its user interface. I’d suggest two key areas for potential improvement:

First, Panjiva should bridge outward in its data capture. They currently rely mostly on US government agency data sets captured during importation. They could expand outwards in data capture while maintaining strategic data alignment. For example, Panjiva could incorporate publicly listed company financial data, such as what is seen on Yahoo finance or Google finance. These data sets are already well mapped by other companies, so integration should be easier. And that data is a important contributors to the kind of intelligence Panjiva is already providing. Inventory levels, asset and return on assets, increase or decrease in store count, etc. These are all data pieces that are both available to capture for free, government-mandated reporting, and very closely aligned to the current Panjiva intelligence targets. Other data capture opportunities also come to mind, such as incorporating CIA world fact book or security alerts into country-of-origin analysis. Again, this is a US government-agency data set that is already digitized and available.

Perhaps the most obvious expansion of data capture would be for Panjiva to get access to activity that isn’t USA importing. But, I believe this is not a good idea. I’d suggest that Panjiva stick to its current (somewhat niche) area and reinforce the data capture as discussed above. Once they are the acknowledged masters of US-bound international trade, moving to another region would be more likely to succeed.

Besides, there are considerable areas left for improvement in the current scope. For example, “volume” of trade is only captures as weight of product or number of shipments. If your industry is concerned with very bulky items, such as mattresses, the “volume” should be captured as cubic meters (space, not weight). Likewise, for small items like semiconductors, diamonds, or watches, the use of weight or shipment count is not viable for most business insights. Actually all three of these items would probably have a large amount of air freight for their USA imports, which would further degrade the accuracy of Panjiva’s intelligence given that they do not have data capture on air freight as of yet.

The second area Panjiva could improve on is the ability of its intelligence to get tied into the business decision moment. For example, could Panjiva introduce workflow modeling so that a buyer has to complete some kind of process in Panjiva and the buying manager or director has to approve it before new orders are created? Could data on orders be loaded into Panjiva somehow, so that there is a mash up of data and approval of a season or product line can be made in-program? Could web service (SOAP) calls be used to have a Panjiva pop-up display during a crucial step in entering orders, such as displaying the latest credit report to an inventory planner? These are examples, among many possibilities, of how Panjiva can take its intelligence and get it closer to the decision process or moment.


As with most of my articles, I’m closing this with a summary of how Panjiva can help you supply chain managers come their next workday. All-told, I’d say that Panjiva is an interesting participant in the supply chain software space as of 2010 and that their limitations are likely to be overcome as the product matures. So far, what they are targeting within the functional scope is being delivered quite well. Here are the Panjiva strengths and limitations, from my perspective:

• Well designed for users: intuitive, consistent, fast, and transparent
• Good creation of intelligence from the supply chain visibility data
• Very reasonable pricing… I’d be surprised if any sourcing office considers this to be over priced

• The scope of coverage is somewhat niche: insights into ocean-freight imports to the USA.
• Requires proactive users to integrate with business decisions. No process or system-level hooks to ensure Panjiva’s intelligence is interrupting decisions to result in different, and better, outcomes.

Leave a Reply