Dirty Data

The recent pronouncement from the research firm Gartner that "dirty data is a business problem, not an IT problem," puts a spotlight on an important issue: automating your business processes won't help -- and might even hurt -- if the underlying data is old, inaccurate, poorly fielded or inconsistent.

Data publishers fully appreciate that their value is based on well-managed data. But businesses -- our customers -- continue to avoid the issue, which most of them find confusing if not overwhelming. What we consistently hear from executives at end-user companies is that because their data is "in the computer," keeping it clean is an IT problem. Those of us who have worked with corporate IT departments know that IT folks typically go to absurd lengths to avoid directly touching data, ever.

To their credit, IT departments are increasingly investing in data hygiene software to try to clean up dirty databases, and there seems to be increasing understanding that the only long-term solution is to catch bad data at input, before it gets into the system. But initiatives on both these fronts have been limited and slow.

This has created a buregeoning opportunity for data publishers because of a growing need for clean look- up databases, matching services to help separate the good data from the bad, and even manual and automated data scrubbing services. Once these companies get their databases in shape, there are then great opportunities to sell data augmentation services, or even to provide databases on a turn- key basis to companies that don’t have the interest or resources to maintain good databases themselves.

As an industry, there are a lot of ways we can help tackle the dirty data problem at its roots and help make the world of data a lot cleaner, while cleaning up in the process.

Labels: data hygiene, data scrubbing, dirty data, gartner, Infocommerce, lookup files

Is Hyper-Local Just More Hype About Local?

Hyper-local is the latest buzzword making the rounds. Reduced to its essence it describes an intensely detailed focus on specific, small, local market areas. The term may have originated with newspapers, who have apparently decided that their new niche should be reporting news of their local communities rather than reporting the international news that everyone else is reporting. The logic is simple and sound: readers value coverage of their local communities, and in many cases, nobody else is providing that coverage. Eureka!

Not surprisingly, a number of entrepreneurs have rushed in with websites to exploit the hyper-local opportunity as well. These online, hyper-local publishing ventures draw on every trendy new concept there is: community, blogs, user-contributed content, tags, the list goes on and on. The potential revenue streams are just as varied. Two companies getting a lot of attention right now are backfence.com and outside.in.

There is merit to the hyper-local concept. But it can't succeed without significant investment and a lot of hard work, and that's where a lot of these online ventures come up short. They've designed themselves to take the path of least investment and energy because they want their businesses to be intensely local yet scalable, something they can replicate nationwide. And that's the rub.

The more ambitious the business plan, perhaps ironically, the more compromised the offering. Operators of hyper-local sites, by choice or necessity, end up supplying little more than a platform that they expect people to engage with and pour content into. Some have tried to short-circuit the "chicken and egg" aspect of user-contributed content by supplying aggregated local news and business listings and classified ads. It all looks neat and cool, but at the end of the day, it's content readily available elsewhere with little value-add (do I really need to see Google maps of my own town? I already know how to find Main Street, thank you very much).

Many of these sites represent virtuoso programming efforts, but they are just tools, platforms devoid of personality and soul. Nobody is organizing or focusing the conversation. Imagine a local newspaper entirely composed of whatever people sent into it that week. That's about what you get with these online sites, and the result is anything but compelling.

Success in hyper-local really depends on "becoming one with your market," and that's hard to do when your real goal is to be in 100 markets as quickly as possible. This is as true for local consumer markets as it is for vertical B2B markets, and should you doubt this, just think back to a little start-up called VerticalNet. You can't build a sturdy national empire off a shaky local base.

Labels: data, hyper-local, local search, marketing, newspapers, verticalnet

Finding Value in Data

Orbitz, an online travel booking site, has just launched a new specialized site for those who book their own travel - http://roadwarrior.orbtiz.com.

How does Orbitz plan to win over this small but lucrative segment of the hotly competitive business travel market? Apparently, not by having the lowest price. Indeed, in an article in the New York Times describing the new site, an Orbitz spokesperson described their fares as "roughly the same" as can be found on the airline's own sites. So why bother going to Orbitz?

Rather than focusing on deals and bargains, this new Orbitz site seeks to build a loyal following by indulging users with valuable, relevant information and useful productivity tools.

It starts with lots of notification options. In addition to the email or phone call a lot of us now already get about flight status, Orbitz offers to send the same status alerts to up to six people -- the person picking you up at your destination, for example. But then Orbitz goes much further, proactively monitoring weather, airport conditions and closures, gate changes, and even problems at your hotel that might impact your stay, and contacting you while there is still time to avoid the problem.

And Orbitz offers much more than just travel alerts. It provides detailed city guides (licensed from wcities.com), along with restaurant and event listings. Hotel listings contain commercial reviews (licensed from Frommer's) as well as reviews from verified Orbtiz business travelers. There's also a searchable database of wireless hotspots (powered by jiwire.com), and a hotel database searchable by business amenities such as Internet connectivity.

Overall, Orbitz has done a commendable job aggregating and integrating a variety of data sources to create a self-service travel agency for the business traveler. It's a great example of how information can power commerce by adding value and differentiation. It's also a great example of how there is increasingly a commerce component (in this case a ticketing business) that is providing the necessary revenue to justify aggregating and integrating all this content in the first place.

This vibrant intersection between information and commerce can be a good thing for data publishers (look at the number who have done deals with Orbitz!), but it can also be an area of concern. Companies selling products and services can justify building or licensing competitive databases and giving them away in the hope they will spur a sale. But when publishers find their subscription data product is someone else's free product catalog, it might not put them out of business, but it could make selling that content just that much harder.

Labels: data publishing, frommer, Infocommerce, jwire, orbitz, wcities

That's Entertainment!

I came away from the SIIA Content Summit in New York City this week enthused by the upbeat tone, but with a few alarm bells ringing in my head.

It started with the keynote presentation by Ann Moore of Time Inc. After taking us for a ride through her celebrity-studded product portfolio, she casually mentioned that while Time has created 133 viable, targeted communities in print, they will henceforth be lumped together into a handful of "content cluster" sites online. Rationale? Advertisers like big traffic numbers and operating 133 sites is too much work.

Maybe that's an intelligent approach in the consumer realm, and maybe more celebrity dross in one place makes for a more compelling site, but note to business information providers: don't try this at home. The targeted audiences you have built are, and will continue to be, one of your most important assets.

In a similar vein, references to "Second Life" managed to creep into virtually every session. The buzz around Second Life is extraordinary, though few in the room (including myself) seemed to have more than a vague understanding of this online simulated world. But I get nervous when I hear the discussion turn to how we need to "adopt Second Life concepts" in our own online sites, and possibly how we should even be looking to Second Life as a marketing channel. Though we learned that Reebok sold 20,000 pairs of sneakers through Second Life, I still suspect it will be a long while before there is a meaningful market on Second Life for industrial buying guides and business credit reports. Indeed, I'd want to run a credit check on anyone who said he found my business while pretending to be someone else in an imaginary online world! Second Life is fascinating, but that doesn't mean it has any relevance to business marketers.

Finally, there continues to be intense interest in user- generated content and the somewhat related concept of online communities. While I remain bullish on the power of user-generated content generally, I continue to shudder every time I hear community used in a discussion of a business site. I have seen far too few successful online business communities, and most of them seem to cluster in the IT/engineering world, where they exist as giant, collaborative help desks. It's not a concept easily portable to any business vertical, though it remains enticing because of its success in the consumer world where exchanging online opinions passes for entertainment. When it comes to role models and best practices be careful to stay on the proper side of the consumer- business divide because being on the wrong side creates a giant mess, and the middle is nothing but a deep hole.

Wake Up and Smell the Data

I ran across several particularly interesting articles by Mike Orren, president and founder of an online local newspaper venture in Dallas called Pegasus News, in which he reviewed some of the lessons learned in launching this new site.

One of his most intriguing findings was that the primary traffic draw to his site was not the highly localized news content that has become the new sine qua non of so many newspaper publishers these days, but rather data. Indeed, he reports that a full 75% of the traffic to his site was from users looking for specific pieces of data as opposed to news. Whether it was restaurant reviews, movie guides or local event calendars, the big benefit to users of online newspapers appears to be the compilation and aggregation of locally relevant facts, not local news.

This may be a somewhat chilling finding from the perspective of the newspaper industry, but it does tend to support what we're seeing in the world of online databases: there seems to be continuing and potentially endless opportunities in both developing and compiling and aggregating highly specialized datasets for both business and consumer use. Users are drowning in information, creating a growing need to organize, normalize and summarize this information to make it more useful and easier to act upon. If you build a quality, useful dataset, there is ample evidence that you'll have an audience for it, provided users know you exist -- and that's increasingly where the business challenge and expense appear to reside.

In another fascinating online posting, Orren muses intelligently on the difference between news stories and data, a favorite hobby horse of mine. Writing for journalists, he explains in a clear and concise way that news content stored in a database is not really databased content. It's only when you break out key aspects of the news story into separate fields, done in a consistent manner that you are building a database. This is what Orrens refers to as storing data "atomically," and only after this is done can you extract maximum value from news content.

What Orrens is really examining is the limitations of full text content, and full text search. While both are useful, convenient and have a clear and needed role, neither can do the full job for the user, because they both limit and obscure information in the process of finding and delivering it. In a world increasingly driven by the automated discovery and processing of information, the most useful, discoverable and valuable information will be that which is optimized for these automated systems (read: computers).

As regular Perspective readers know, we have been underwhelmed by the newspaper industry’s digital initiatives, and this industry insider nails it when he says, "The news business as we know it is only going to continue to contract and weaken unless and until news organizations start treating everything as data rather than stories." We couldn’t make a better case for infocommerce ourselves.

Labels: data, databases, journalism, newspaper, technology, text