Posts

blog_0303

Local Intelligence in Global Address Verification: History Lessons

I recently published an open letter to the address verification industry, a group to which I’ve belonged professionally for over 20 years. In it, I critiqued our lack of innovation over the years and suggested another way: making global address verification better by using local intelligence.

As hoped, the letter spurred a lot of interesting dialogue. In hopes to keep the conversation going, I’m posting a series of blog entries that dig deeper into the themes of technology, address data and how we can meet the needs of our most demanding customers.

History Lessons

Let’s start with some of the history that made me aware of the problems we have in the address verification industry.

In 2008 I received a call from an executive with a prominent eCommerce company. He had been piloting a cross-border initiative in which customers from other countries could order from his website and he would ship their products directly from a U.S. distribution facility.

The experiment was not going well, he told me. Deliveries were getting lost, shipped back, or just taking longer to arrive than he (or the customer) expected. The problems stemmed from addresses. “Could you help me?” he asked.

I had worked in the address data industry for 15 years at that point and was currently Chief Operating Officer for a leading provider of generic global address verification services. We had accumulated a massive database of addresses which we licensed mostly from national postal operators. It covered 240 countries and territories. We solved the very problems the eCommerce executive was describing. Or at least we thought we did.

I asked him to send me some address files, and I’d have my team run them against our database.

When the results came back, I was pleased. We had cleaned up his files nicely and our analysis said he could confidently ship his packages to nearly every address. But when I called him back to share what I thought was good news, he just sighed.

“No, this won’t do at all,” he told me. “You’re quick to tell me to ship, but I happen to know these addresses are incomplete. You’re asking me to put my packages into the postal stream and trust that the local infrastructure to do the dirty work to get it delivered.”

He had conducted the same test with other “generic” global address services, and they all came back the same. As hard as I tried, I didn’t have a good response to his points.

“When you figure out how to solve that last mile problem before I ship,” he concluded, “give me a call. But for now what you have just isn’t good enough.”

His criticism stung, but it wasn’t entirely unfair or inaccurate. Since the dawn of our industry in the 1960’s, we global generic address companies had struck a Faustian bargain with our primary customers – direct mail businesses and the vendors that provided them support. If we could supply them a good enough service for a cheap enough price, the bargain went, they would license our address data. But if we raised the price for any reason, they would just stop using us.

It sounded brutal (and it was), but it made sense for the problem they wanted us to help solve. They sent mailers en masse. These were inexpensive cards and envelopes that often went out in batches of many thousands at a time. It often didn’t matter if some percent of them didn’t get delivered. It was a volume game with everyone operating on tiny margins. Our job was not to get every envelope delivered. It was more of an actuarial calculation: if they could use us to cull out some small percentage of the bad addresses, the pennies they spent on us would save them at least that much in postage costs.

And then we had our dirty secret: our False-Positive Bias. If we could verify an address to a locality level, confirming, for example, that the street existed even if the number was wrong – we would call the result reliable and advise the customer to send it.

Why?

Because we knew most countries have impressive last mile delivery infrastructures. If you can get mail to the locality level, the people in the post office were really good at fixing your mistakes and getting the letter where it needed to go. It would take longer to get there because it had to go through exception processing, but the direct mail customers were generally okay with that. As long as the right percentage got through and our prices stayed low enough, the Faustian bargain held.

At the time the eCommerce executive contacted me, over 70 percent of our revenue came from customers related to direct mail. We were beholden to their requirements, and to be candid, we weren’t making very high margins ourselves. This was not exactly an environment conducive to innovation.

Without investing in a lot of improvements, we were never going to be good enough to satisfy the eCommerce industry’s needs. Their requirements were based on precision and spend – they expected every package to be delivered…and quickly! Mis-delivery cost them shipping and product losses and slow delivery cost them customers. The “good enough” approach that defined the generic global address industry – that False-Positive Bias – wouldn’t cut it.

For the next couple years I watched eCommerce trends closely. I saw cross-border shipping accelerate, growing rapidly quarter over quarter even while the address problem persisted. They hadn’t yet found good alternatives to meet their needs.

There is an opportunity here to do something useful, I remember thinking to myself, if only we can find a way to take the local intelligence that’s so good in delivery and apply it earlier in the chain of events for shipping packages. We needed to find a way to apply local intelligence in the address verification process. Before a package ever leaves the warehouse.

Two years later, during a trip to Brazil, I finally saw a way forward. More on that in my next post.

blog_icon_0224

The Importance of User Feedback in Address Verification Accuracy

We tend to judge the quality of international address verification software by the size of the reference database and the return values it generates. But these metrics are incomplete and have the potential to mislead.

What are better gauges of quality? Local user transactions and local user feedback.

A local address solution, built and maintained in its own country, will have a base of users from that same country. These local users are familiar with the address nuances of their markets so they know what constitutes a good address (and a bad one) based on what people like them expect to see.

Better yet, when they see something that doesn’t look right, they report it. In the nearly five years we’ve seen this dynamic at work, we’ve noticed that providers of local address solutions react quickly to input from their users by incorporating corrections into their address data. This creates a real-time feedback loop for tuning the local engine, ensuring its information is constantly improved and done so quickly.

Local address solutions consumed by local users creates a virtuous cycle where both sides work together to produce the most localized (and reliable) results.

Consider the following hypothetical: Two vendors have provided address solutions for a country for over ten years each. One vendor has a local staff of developers and has processed over 750,000,000 address transactions from a customer base of more than 10,000 users in its country. As a result it has developed a deep competency of experience in the area.

The other vendor has no local staff in the country even though it has a handful of customers there. This generates a significantly lower number of addresses being run and reviewed by users with a local mindset. The vendor’s knowledge of the market’s local nuances is developed from a distance and without the benefit of significant local feedback to help improve its results.

If you’re an eCommerce business shipping goods into this country, which vendor would you have more confidence in for verifying your customer addresses? Which one would you trust to make sure your parcel is delivered to your customer on time?

If you are vetting international address verification solutions, consider a different set of metrics. Local user transactions and feedback should play as much of (or perhaps more important role) than the simple measures of size of the reference dataset or the outputs of an engine.

Let local customers and local experience be your guide.

blog_icon_glocalization

Glocalization as an Edge in Strengthening Retailer eCommerce Efforts

Reading the December 4th, 2014 edition of the Wall Street Journal I spotted two side by side articles about retailers growth and challenges (see below). Fundamentally U.S. Retailers are trying to determine what their ecommerce strategy is and whether to go stay local or to glocalize.

Yes I said glocalize which is a derivation of glocalization. Glocalization is the art of globalizing your web commerce presence while at the same time localizing it for the respect countries and regions in which you are expanding. A classic example of where this is important is noted in the WSJ article about Nordstrom’s intent to expand into Canada and how the competition in Canada in the form of Harry Rosen Inc expects to maintain an edge.

Expanding your retail footprint into a new country is not as simple as copying your web site and adding some additional payment options. Even if there is a shared language there is not a shared cultural and experiential baseline. In the case of Nordstrom’s versus Harry Rosen Inc, the Canadian retailer’s chief executive states “We’re experts in Canada”.

Further to that point glocalizing the online order entry process will help enhance the sales experience and create a smoother sales process for customers. Also on the backend making sure your local data capture of Name/Address/Phone and other details are mapped to local standards will ensure that you are leveraging customer data in a way that correlates to a local customer’s expectations and experience. A great example of this is using address correction technology that doesn’t account for localized data and rules. A customer enters Hollywood Hills, CA into an order along with a proper zip code. However the address correction software changes the information to Los Angeles to correlate to the zip code provided. While this may be correct in terms of the address it is not correct for local residents of the region who expect that their “vanity name” is acceptable.

Hopefully Nordys (The local user term for Nordstroms) will apply a glocalized strategy to their Canadian growth efforts both in the brick and mortar experience as well as the online ecommerce experience. Until then I suspect Harry Rosen Inc. has little to be worried about.

U.S. Retailers Learn to Speak Canadian

http://online.wsj.com/articles/u-s-retailers-learn-to-speak-canadian-1417660005

 

A&F and Aeropostale Can’t Shake Teen Blue

http://online.wsj.com/articles/SB121810860555720233