All Aboard! Bringing partners onto the network.

In order to have an effective data network, you need to have a well thought out approach to bringing new partners onto the network, both in terms of providers of data and consumers of data.

One  of the most important things to consider when adding participants to your network is standards. However, standards are a double-edged sword. You need to have agreed upon standards, but you also need to recognize that not everyone can adhere to the standards.

For instance, in our world, we often deal with UN Location codes, standard codes for many cities in the world. This is a well known standard, which we support. For partners coming on to the system, if they can understand UN Location codes, coming on-board is very easy. However, we need to recognize and handle exceptions: UN Location codes occasionally change, and some partners will still use the old codes; some cities do not yet have UN Location codes assigned; some partners use a competing standard (Schedule D/K); some partners use a completely proprietary standard. We need to exist in a world where things are often messier than we would prefer.

When partners can use standard codes, on-boarding is smooth and quick. The exceptions can cause some delay. However, we have designed our systems to handle the exceptions, and while handling partners who cannot use the standards will never be as quick as a “vanilla” on-boarding, we have tools to map customer specific codes into standard codes, etc., that help expedite these exceptions.

Locations are just one example of the messiness we need to handle every day. While we have learned how valuable standards are, we also know that deviations from standards are a way of life and we need to accept these. However, at the end of the day, we must ensure that when the data goes out to our customers, we have untangled this messiness and present the data to our customers in a normalized format they can digest. Therefore, though we build our systems around standards, we also have learned that we must build systems that can smoothly accept deviations from standards.

The Value of Quality Data – Benchmarking

Sometimes, when you get too deep into the nitty gritty of improving data quality, you can lose sight of why getting good quality data is important in the first place. After all, quality data is not an end unto itself.

One clear reason you want to have high quality data is to power your visibility solution. My colleague John has a blog  devoted to supply chain visibility and the value derived from it.

I’d like to focus a little bit on another source of value that good quality data enables — benchmarking. Once you achieve a level of reliable and complete data, you can establish benchmarks by which to measure your supply chain partners.

I’ll give just a few examples. When you are negotiating rates with your carriers, you agree on certain travel times from origin to destination. Are your carriers really honoring these travel time commitments? If you’re like most people, you probably don’t know.

However, if you’ve taken the time to collect good quality data, you can measure the actual travel times and use them to compare against the benchmarks you established during your negotiations with the carriers. Which carriers are honoring their commitments? Which ones are not? Wouldn’t this be useful information when negotiating your next round of contracts?

If you are using a service provider to load your containers, one metric you may want to benchmark is the container utilization. In other words, how full is your service provider packing your containers? You are paying freight charges per container, so you want your containers to be as full as possible. If you have good data on the volume of the contents of your containers, you can measure container utilization and see if the utilization is meeting your expectation. The “least expensive” service may well not be the one that is costing you the least in the end.

This type of benchmarking and comparison of one service provider to another is only possible once you have a systematic way to collect complete and accurate data.

Other Network Effects

So far, when I’ve been talking about Supply Chain Networks, I’ve been mostly talking about networks of data providers sending data to us, so we can normalize the data and send it on to our customers. This is a vital network for managing your supply chain, but I wanted to talk about some other network effects that are in play at GT Nexus.

Another type of network that has become very powerful is our network for freight procurement. We have an application that allows shippers to submit requests to ocean carriers and other providers for freight transportation and allows the providers to respond. Unlike  traditional license and install software, the carriers log in to the application and respond directly to the customer’s bid online. The carriers respond to the bid using the exact same screens, regardless of the customer. Once the carrier has been trained for one customer, they will know what to do from the next. Over time, this has built up a powerful network of shippers and carriers who interact over the platform for freight procurement. As more carriers participate, the more streamlined the process comes for everyone new who comes along. More customers bring on heavier carrier participation, which, in turn, makes things easier and easier as more customers come on board. Just like in the data quality network, this network demonstrates how network effects increase the value for everyone.

Another network well worth mentioning is the GT Nexus Shipper’s Council. This is a group of our customers who have grouped together and meet every quarter to discuss best practices and to push for industry change to help meet their mutual needs. The Shipper’s Council has helped pull some of the “levers and knobs” of the data quality network to drive improvement. There is a real human component to the network here, beyond the network of bits of data traveling over wires. Like the data quality network, this network of major shippers pooling their resources to drive change is a powerful manifestation of the network effect. As the council grows, it’s influence expands and the benefit to each member increases.

In the online world, there are many opportunities to take advantage of powerful network effects that would not be available in the license and install world. When network effects are in play, there is value creation that would not be otherwise attainable.

Joseph M Juran

In my work on data quality, I’ve learned a bit about Joseph M. Juran, the founder of the modern science of data quality. Joseph Juran did a lot of amazing things during his lifetime, including creating the “80-20 rule” that 80% of the problems stem from 20% of the cause. This interesting individual passed away only last year and lived to the ripe age of 103, staying active well into his 90s. Among the many things Juran is credited with is adding the human component into the study and management of data quality.  I think we all need to keep in mind that this is about managing behaviors and attitudes as much as it is about managing data.

As a final note of trivia, Joseph Juran’s brother is an Academy Award winner, and the director of “Attack of the 50 Foot Woman”.

I recommend taking the time to learn more about this interesting fellow.

Managing Data Quality in an On-demand Network

As we’ve been working, along with our customers, on working with partners on improving data quality across the network, one issue that has come up is, who should be the party to work with the provider of the data?

Some would say that it should be our customers – the end consumer of the data. Why? Because they’re the ones who understand their data quality needs the best and because they’re the ones with the most leverage over the partner. If the partner cannot meet the needs of the end customer, they might lose their business.

Some would say it should be us, GT Nexus, to work directly with the data provider. Why? Because we’re the ones who can see all the data, regardless of the customer. We understand how the data maps to our systems. We have the contacts with the data providers and we know how to present the problems to the providers in ways they are famiar with and can understand.

So, how do we reconcile these two approaches? Over the years, we’ve leaned heavily towards one side or the other. We’ve found, unsurprisingly, that the middle path works best. We have established regular programs working with the providers on high level issues that impact many customers. But we still encourage the customers to be involved, as there may be some issues that are specific to them, and the partner needs to understand the business impact to their customer in order to be most motivated to fix it.

So, it’s not really a question of whether we or our customers do the data quality work. It’s more a question of what are the best tools we can provide to foster the atmosphere of collaboration to tie this together and make it all work.

Approaching your partners

When building a data quality program, you need to be mindful about how you approach your partners and how you provide them the information they will need to make improvements.  I learned this when I worked with a group of our customers to establish a data quality program to improve the data received from the major ocean carriers.

Our customers rely on data from the ocean carriers they use to ship containers around the world. This data is crucial for managing complex global supply chains.

When we started this program, we needed to be careful about how we positioned it. We did not want the program to seem like something that reflected at all  negatively on the carriers. We were simply helping to identify gaps in the data so the carriers could fix them and provide better service to our mutual customers. There may have been a little resistance at first. But, as the program progressed, the results became dramatic. In fact, several carriers have come back to us to let us know how much this program has helped them.

Why has this program been successful? I think there are a few keys points.

- The information needed to be specific. What are the problems, exactly?  Give specific examples and metrics. You need both roll-up metrics and detailed examples.

- Choose your “battles.”  Go for the easy wins first to show some dramatic early gains to bring everyone on board. Refine later.

- Establish a collaborative atmosphere. This is not about knocking anyone down. Identifying gaps is done to point out areas for improvement. Treat everyone involved with the respect a true partner deserves. Make clear that when the program succeeds, everyone wins.

The Value of Measuring

I ride my bike to work almost every day. It’s almost 8 miles each way (7.9, to be precise), but I didn’t know that until recently. Before anyone who doesn’t know me thinks I’m some kind of a jock, this is far from the truth. I just love to ride my bike – I think it’s a holdover from a summer I spent in Beijing in 1993.

Anyway, I recently installed a little bike computer on my bicycle. This is a small device that reads from a sensor on the front wheel which calculates how fast I’m riding and how far I’ve gone. That’s how I know it’s 7.9 miles to work.

I’ve found that, since I’ve installed the computer, I’ve been starting to ride a little faster, without even consciously trying to. The simple fact that I’m measuring my speed and paying attention to it makes me go faster.

I think this is similar to data quality. I’m not suggesting that just by running some metrics, your quality is suddenly going to get better on it’s own. I’m just saying that putting in a program to measure your data quality is the most important first step. Just by measuring your quality, you’ll start paying attention to things you probably never thought about before. You’ll likely identify some “easy wins” which will improve your data quality with little incremental effort on your part. So, go ahead and install that “bike computer” on your data network!

Measuring Data Quality

In order to improve the data quality of your network, you need a way to measure data quality and find the gaps. In measuring data quality across a set of partners, you need to insure the following:

• Find something that is easy to measure
• Measure the aspects of the data that are important to your business
• Have metrics that are applied consistently across all partners

Once you have found the correct things to measure, you can how your partners compare and find data quality gaps.

If you are just beginning a data quality program, I believe it is appropriate to use more “carrot” than “stick”. Your partner may never have been engaged in a data quality program before. Most of the partners we’ve worked with have been eager to help, once the correct framework has been established. Simply finding an objective way to accurately measure data quality and give specific feedback to your partners may go a long way in improving your data quality.

Next time, I will go into more detail about getting the feedback to your partners that they need to help improve data quality.

Network Effects and Data Quality

As I mentioned in my last post, there are a few clear benefits from tapping into an existing data network, rather than building your own. However, I haven’t yet talked about what I think is the most important benefit, though, perhaps, this is somewhat more subtle.

I’m talking about the Network Effects that come into play when many customers are receiving data on the same network. The Network Effect refers to what happens when more and more users join the same network. As more users join, the value goes up to everybody on the network.

Let me explain what I’m talking about. Suppose you tap into a data network that is being shared by 100 other users. Now suppose you uncover a data problem with one of the partners providing data to the network. You may work with the data provider to correct the problem. Now the 100 other users of the network get the benefit of this data quality fix at no incremental cost to them. Similarly, the 100 other users may be working on issues of their own. When these issues get corrected, you too get a free ride.

With Network Effects, you may be able to achieve levels of data quality you would never be able to attain on your own. Here you are truely leveraging the power of the network. Yes, it’s great to save the costs of building and maintaining the connections, as I talked about last week, but the benefits go far beyond this.

Getting the data: Build or Rent?

The way I look at it, the most important decision you can make when deciding how to get data into your applications is whether you will “go it alone” or share the costs of getting data to your network with other folks in the same shoes as you.

Of course, I feel that sharing makes more sense for many reasons, some more obvious than the others. The obvious reason is that when you share the costs of building and maintaining the connections with other people, your own costs are minimized. You can think of this as similar to an infrastructure project. You would never build your own bridge or highway. But, when you pay a toll to cross the bridge, you are effectively sharing the cost of building the bridge with many users. Makes sense, huh? See the recent entry on my colleague’s blog.

However, the efficiencies gained when sharing data networks go beyond these somewhat straightforward ones. I will be blogging more about this in the near future.