Monthly Archives: February 2011

Trade Extensions Trades Up to a Fact Sheet User Interface

Trade Extensions is another company that has refused to sit still during the downturn. Even though it did a complete overhaul of its UI in late 2009 (when it traded up its UI and e-Negotiation Management Capabilities), it realized that was just the first step of many that would be required to make its application more useful and more useable across the Supply Management Organization. Trade Extensions has been working hard to break out of the “complex transportation optimization” niche that they started in and has developed a fully functional Sourcing and Supply Optimization solution because they see the value that an organization can receive when optimization-based sourcing events are undertaken across the board.

While Trade Extensions started as a niche Strategic Sourcing Decision Optimization platform in the early naughts, it has expanded into a full suite solution over the years and now includes one of the most powerful RFX modules on the market, real-time optimization-powered reverse auctions, contract management, and OLAP spend analysis (which rivals many of the other suite spend analysis solutions on the market and a few of the pure-play solutions as well). And while they don’t yet have much of a SIM/SRM/SPM solution, they recognize that as a weak point and it is one of their top priorities for the coming year.

In addition, while they don’t yet have category RFX solutions or a centralized data management facility (or “Business Centre”), they are planning to introduce new project types to simplify RFX and Auction setup and their new support for fact-sheet based RFX and reverse auctions makes data management significantly simpler and takes them further down the data management path than many of their e-Sourcing Suite provider brethren who have been standing still for the last few years.

Internally, Trade Extensions uses OLAP cubes to store its data. Thus, the user can provide data in d-dimensional fact sheets, which include 2-dimensional spreadsheets and 3-dimensional workbooks, that define data in familiar matrix notion. This is a very powerful method of data input. A user can define a fact at any level that makes sense. If the cost of interest is a single freight cost from a ship from to a ship to, the user can provide costs in a 2*2 matrix of ship from locations and ship to locations. Users can also define costs at the region or country level if they like, and if a user defines a fact sheet that relates lanes to regions, the user only needs to specify costs at the region level. Moreover, if the user so desires, once costs are specified at the region level, they can be overridden at the ship from – ship to level. The system will do this automatically if the fact sheets are appropriately defined and included in the project as it knows d+1-dimensional costs always override d-dimensional costs. In other words, if you define continents as collections of territories, territories as collections of regions, and regions as collections of lanes, the system knows that territory data overrides continent data and region data overrides territory data, if defined. Otherwise, default values from a higher level in the hierarchy are to be used if no suitable cost exists at a lower level of the hierarchy. This is important as this allows a user to only enter detailed cost data where it is required (and where it differs from default pricing for the category).

As a result, data collection in Trade Extensions’ e-Sourcing platform is quite easy. A buyer only has to define what the organization wants to collect, and a supplier only has to bid on what it can provide. Furthermore, a supplier only has to bid to a level of detail that makes sense. Plus, all bids can be formula based. So, if a supplier supplies a product in different sizes, and the price is dependent on weight or volume, it can define a formula once and re-use it on each size. Furthermore, a user can re-use any and all fact sheets from the event in other events, minimizing data collection and data entry. It’s definitely another step along the path to ultimate usability.

Don’t Forget The Most Important Principle of Performance Measurement

SIG just ran a great article on the “Three Guiding Principles of Performance Measurement” that hit upon a key point that is regularly and repeatedly overlooked. Simply put,

Your data is good enough. Really, it is.

As the author says, your data will always be murky. No matter how many times you clean it, massage it, enrich it, transform it, and polish it up, there will always be errors. There will always be omissions. There will always be inaccuracies.

But that’s okay. Remember the 80/20 rule, which is just as appropriate here as it is everywhere else. (In fact, more appropriate.) The law of the vital few states that roughly 80% of the effects come from 20% of the causes. This says that only 20% of your data is vital anyway. And since it is expected that at least 80% of that data will be clean, it’s going to be quite straight forward to spot important trends. (That go just beyond who your top suppliers are, but which items are ordered the most. What types of purchases are regularly bought off contract. What products are really selling best. Etc.)

And if the decision being made is critical or risky, the data will always come up short. There will always be some flaw, some missing piece of information. Because if there wasn’t, the proverbial pointy-haired boss could be an effective manager. Executive decisions will always require a bit of a risk, but they will be much better when there are data to base them on then when there aren’t. So measure, manage better, and manage some more.

“Trendspotting”

Editor’s Note: Today’s post is from Dick Locke, Sourcing Innovation’s resident expert on International Sourcing and Procurement. (His previous guest posts are still archived.)

the doctor requested that I look at Panjiva’s new product, Trendspotting, which they are advertising on their site and blog. It has some good points, but it also has many not-so-good points that will ultimately render it unusable for many purchasing people.

Let’s look at the three main features:

“Crack the HTS code.”

The world really does need an HTS-to-your language translator. I don’t think there’s one into English, and this definitely isn’t one. From their demo page, put in “computers” and see what happens. You won’t find anything. OK, that’s a trick because I know that the entire global customs community calls them “automatic data processing machines.” So, type that in the search box.

You get a list of potential HTS codes. Here the system shows its US-centricity. I’m in Mexico, the internet knows I’m in Mexico (when I go to Google.com I get their Mexican home page) but Panjiva gives me the special 10-digit US codes. That’s still helpful in the goal of finding the right countries if you are in North America. It’s not so helpful if you are somewhere else.

“Find countries”

Pick the first HTS code (8471.30.01.00) and click “Trends.” You get some really helpful data on laptop imports. You can see that the two primary sources are China and Malaysia. You can see their trends. This is really a handy sourcing tool. However you can get the same data free from the US International Trade Commission‘s web site. I’ve been advocating doing that for about 14 years in my seminars. The USITC site is less graphically pleasing and slightly harder to use, however. Panjiva did a good programming job.

“Find suppliers”

Here it really falls apart. You would expect to find Lenovo and Foxconn as suppliers of laptops. They’re not listed. However, you do find Autoliv China Inflator Company Ltd, who makes airbag inflators. What’s going on here?

It’s a data source problem. They list companies for one of two reasons. One is that a company shows up as a sea freight shipper of automatic data processing machines on a public data base. There’s no such data base for air shipments. Very few laptops travel by ocean.
I’ve been using this kind of data for more than a decade. It has two more big problems. First if the words “automatic data processing machines” (ADP machines) show up on import documentation, the exporter gets listed. That gets items that connect to, contain, or are parts of ADP machines. Second, many companies have set up legal entities in China that purchase for them. The Chinese entity (i.e. HP China) buys the goods, and ships them to HP US. HP would show up as both the exporter and importer and you would never find the name of the manufacturer.

Of course, your results could be different than mine. Try it on a product where you know what the countries are and where the suppliers are and see what happens. Maybe it will work better for you than for me.

Thanks, Dick.

Does Verizon Have Its Head in the Clouds?

As we all know, Verizon recently acquired the right to offer iPhones on its network, a status symbol that mobile providers the world over have been vying for as they wait for the initial exclusivity agreements to expire. As the first provider given permission by the world’s fifth largest mobile phone manufacturer (Globe and Mail) to challenge AT&T in the United States, it is probably a little bit smug right now as it waits for its February 10th launch date. But is that any reason to leave its head in the clouds?

Needless to say I was a little shocked to hear that Verizon is paying $1.4 Billion for Terremark Worldwide (New York Times, January 29, 2011), a provider of cloud computing services. While it’s true that Terremark runs 13 global data centres, and data centres are valuable, 13 data centres are not $1.4 Billion valuable. Trust me. I can build you ten modular and scalable Green Data Centers for 1/10th of that with a lot more computing power than you might think you’d get for that price by going with high-density low-power 64-core servers from IBM or SUN which, by the way, cost less than 100K each, even maxed out on memory. You can easily fit 8 of those on a 20u rack. That’s the equivalent of 256 dual core servers on a single rack. Fill 8 racks with servers and 4 racks with SANs (Storage Area Networks) and you have a data centre — that fits in a small room — with the equivalent of 2,000 machines of processing power, that only costs you about 10 Million. I know Terremark also offers services and has a market cap of about 945M, but a 35% premium for something you can build yourself in a few months? I just don’t get it!