Here's what I would write.
A company must have access to catalogs of niches with enough accurate information and connectivity--both internal and external--so that a system can use filters, relationships and/or recommendations in a single interface to help a customer repeatedly find new and satisfying selections, regardless of the sales volume for those selections.
That's a mouthful. And it will definitely take more than one post for me to dissect that sentence. I'll start with niches for this post and then discuss external connectivity in a later post.
(+/-)Niches
Why focus on niches? As Chris Anderson points out, the items at the head of the distribution change when your viewpoint changes. You don't necessarily like to watch the same types of shows that I like to watch. And I definitely don't watch the shows my mother watches. But if you combine all TV viewing habits in the US, then you'll find that "CSI" and "American Idol" were in the mainstream head for the 2004-05 US primetime TV season since they were two of the most-watched shows.
Now we'll make the universe bigger. Imagine a mainstream head comprised of all TV series that aired at least 13 episodes on primetime network TV in US history. This is likely where tvtome.com started. But how do you add the rest of the data? Niches. By using niches, you can delve into numerous areas of interest deeply enough and use the niche boundaries to stay focused. As you niche hop, the data expand to form a very long and robust tail to that mainstream head.
Then we only need to define these niches to fill out the long tail. But niches aren't just genre-based. A niche can be any category that interests a particular person. Niches definitely crossover, forming a matrix with enough redundancy to maintain accuracy, increase relationships and improve completeness.
You can imagine the following niche categories for TV shows:
And there are a lot of other categories too. Check out TV Acres to see a site that categorizes TV shows into more areas than you can imagine.
As people entered additional data into tvtome.com, they filled out some of the long tail because they were interested in at least one these niches. And those people considered this newly-added data to be quite mainstream (and it was from their perspective) even though it was in the long tail for the masses.
What's next? There are at least four clear challenges with the approach using niches.
1. How do you convince niche experts to provide data for the long tail of their niche? They sometimes provide data for the head of their niche but not always the details for the long tail of their niche. Oftentimes they'd rather be independent than part of a mainstream community. Does that mean that a large organization should setup independently-run, community-based niche aggregation sites with different front-ends but similar data formats so that the data can easily be rolled into a single database?
2. If all of these niche categories do exist, does each category need its own field? (It would make filters work quite nicely.) If one person is interested in a particular niche, it's likely that someone else may be interested as well. But where do you draw the line?
3. How deep into a niche do you catalog? If you go too deep, you are likely to reach a point of diminishing returns. You'll spend time finding and cataloging media that might never be requested or might not satisfy customers often enough.
4. Every company must clearly define the universe of the desired media coverage. Is your goal to document all TV series that aired in the US? Any TV show, including specials and movies? All English speaking countries? Any TV set in the world? Any video monitor for individual consumption?
I can speak about these challenges first-hand because of a fairly simple project. I am compiling a database that tracks national US TV broadcasts (all networks, all delivery outlets, all delivery times, all history). So far this database has over 600,000 records and includes nearly 9000 series, 6000 movies (made-for-TV and TV presentations of theatrical films), and 5000 specials. My initial focus is US primetime network TV over the last 20 years, but I add other data in my defined universe as I find it. This initial focus makes a nice companion to imdb.com and tv.com since it includes date/network/day/time for all airings (and even repeat episode title in many cases).
The "On This Date" feature in the near right margin is generated from this database. To get premiere dates for these series, I've jumped head first into different genres. Earlier this year I spent a chunk of time with gameshows. And then another block with animated shows. For a while I investigated Saturday morning shows. I recently worried about syndicated talk shows of the 2000s. And just last night I was trying to find premiere dates for a few of the older original series on WE: Women's Entertainment. You may need to investigate multiple niches in order to find the info that you're looking for. And every time you focus on a niche, you uncover things in the long tail that you didn't initially know about.
And here's a specific instance of how overlapping niches improve completeness in the long tail. On this blog I published a list of busted pilots. Richard Yokley, who published a book about TV firefighters, found a show on this list that he didn't previously know about. And when I read his book, I expect to find something for the database that I previously didn't know.
No comments:
Post a Comment