Louis Gray has a dynamite post today listing his view of what the 5 ways people find relevant information are. He missed some, which I will add to my list, but this is a valuable list. In fairness, I’m not sure Louis sees this list as extending to search, he is focused on filtering what we already have, but I think search goes hand in hand with the process and fits very naturally. You can tell because it’s easy to imagine combining some or all of these techniques to improve search and filtering for some specific problem.
This is important because having an exhaustive taxonomy or framework is the first step in analyzing or gaming what’s going on.
Here are the different ways people find relevant information (the first 5 are from Gray’s post):
1. Editorial Filtering. Some expert is nominated who will tell you what’s important. You follow that expert because you like their choices, or you just want the comfort of knowing those are the choices of an expert. This has to be one of the earliest mechanisms and one we still see very commonly. The New York Times is a great example of Expert Filtering.
2. Global Popularity Filtering. If a lot of people like a thing, it must be a good thing. This one is just as old if not older than Editorial Filtering. Examples are all around us ranging from our system of Government (though they sure can make themselves unpopular at times!) to American Idol. Online services frequently present ranked listings based on popularity.
3. Social Filtering: If my friends liked it, I probably will too. Hey, this is what Facebook is all about.
4. Explicit Personalization: Tell us what you like and don’t like and we’ll use that to help filter. Netflix and Amazon want you to rate what you have already consumed. They use that information to find new things you will like.
5. Implicit Personalization: We will watch you and infer what you like. In Amazon, “people who like this book also bought these books.”
6. Keyword Filtering: If the same words are used with the same frequency as something you like, the words probably describe something else you like. Hey, that’s the life of a Search Engine.
7. Crowdsourced Filtering: Get on a forum and ask peeps what they think. These are not necessarily your friends. Anyone on the forum can answer. Hopefully the place you’ve chosen is frequented by people who are more likely to know your kind of answers than a random person selected off the street. I really like StackOverflow for programming-related questions, for example.
8. Location Filtering: Geolocation is big these days. We can assume you are where you are for a reason. Maybe knowing where you are helps us filter certain kinds of information. Yelp is a beautiful thing for that reason.
9. Demographic Filtering: This is a very old mechanism for marketers to use when targeting, and often it is the first thing they reach for. It’s effectiveness is surpisingly limited (Explicit and Implicit personalization can be as much as 10x more predictive, for example) but it can add value. Demographics include age, sex, ethnic background, and the region you live in. Dell wanting to know whether you want a computer for home use, small business, big business, or whatever is their way of trying to apply a little bit of demographic filtering to help you find what you want.
10. Link Network Filtering: If the Internet has taken the time to create links to a particular answer of some kind, it is probably a better answer. This one is clearly related to some of the others, but I wanted to call it out on its own simply because of the importance of things like Google Page Rank and Retweets. These are subtley different than a pure popularity score, for example. One difference is people have to work harder to make their input known. They may have to create a web site with links to the content, for example. Another is that this type of relevance weighting usually requires a fair amount of analysis to collect. Popularity contests are usually very obvious and up front. Perhaps this could be referred to as implicit popularity too, or reference to authority.
There are probably other methods, but this is a strong set. What can we do with it?
Product designers can ask themselves whether their product benefits from the addition of one or more. That’s cool. One I am more fascinated with is Marketing.
At its heart, marketing is the art of getting on people’s radar screens and passing muster once on the radar as something they have to have. At a level of abstraction, this means gaming their information filtering strategies to a greater or lesser extent. I think it would make a fascinating marketing offsite to take the list of 10 and go figure out how to move the needle in a favorable direction for each and every one of them. Don’t think it’s important? Don’t think it’s marketing?
Consider that SEO is basically the process of gaming Keyword Filtering and Link Network Filtering. If you don’t think SEO is important to marketers, I can’t help you. But one thing I believe wholeheartedly is that efficient marketing is the process of finding ways to market that are different. When everyone is marketing in exactly the same way (i.e. using exactly the same channel to send the message and sending very similar messages), the noise level is just too high. What if, during the course of that offsite, you discover that for your market some of these filtering mechanisms are hardly being considered and its pretty straightforward to walk right in that door?
Just to help with understanding, here is the linkage between the 10 and various kinds of marketing:
– Keyword Filtering + Link Filtering = SEO
– Editorial Filtering = PR
– Global Popularity Filtering: Sometimes I think this is all simple-minded marketing thinks about. Yeah, we want to be popular, we want to be cool!
– Social Filtering: Of course Social Media marketing goes here. But so do referral programs. How can you get friends to tell friends they gotta have your product.
– Explicit Personalization: Can you get prospects to qualify themselves in some way? Is there a referral service for your industry that asks a series of key questions and then suggests products that fit the answers? And what is the impact of saying, “We don’t think you’re a customer.” No Soup for You may just increase the appetite for it, oddly enough.
– Implicit Personalization: Targeting like-minded people is an old marketing technique. You must have sat around a table discussing where your customers like to get their information, their likes and dislikes, and how to find more people just like that.
– Crowdsourced Filtering: This is a fascinating area that I’ve had a lot of experience with. If you frequent social forums, there are people in them all the time asking for advice on what to buy. It is rare that you see companies responding there, yet if it’s done tastefully and without being too pushy, it can be very effective. Do you have a strategy to cover those venues? Do you have a strategy to convince your customers to speak up for you in those venues?
– Demographics: I won’t teach you how to suck eggs. Marketers have been focused on demographics for ages.
– Location: Does your business depend on location? It’s a function of walk-in traffic. If you depend on it, you have to avail yourself of every possible means of getting on the radar based on location.
So there you have it. A nice cut at a taxonomy for filtering information by relevance. Doing that job better than anyone and monetizing it is what made Google successful. Figure out how to make sure you’re successful at it too.
Thanks Louis for the post!