How much information is enough? How much is too much? And, most importantly, how much information is optimal?
Information foraging gives us a way to formally model user trade-offs in deciding how much to read on your website. More precisely, diet selection is a modeling tool that tells us what food animals will eat and what articles users will read. In both scenarios, animals and people decide what to consume in a way that optimizes their benefits relative to the costs.
For example, say a forest is inhabited by large and small rabbits. Which will the wolf eat? The obvious answer might seem to be "large rabbits" because they provide the biggest benefit in terms of filling the stomach. But if the big bunnies run faster and are harder to catch, the benefit decreases. Much better to eat lots of small tasty bites if tiny bunnies are easier to nab.
So, basically, the wolf must eat more calories than it expends pursuing prey. The real question is not which prey provides the most food, but how you get the most food relative to the cost of chasing it down.
The cost/benefit ratio is what matters, not the benefit alone.
Exactly the same is true for informavores. A long article might contain more information, but if it takes too long to read, users will abandon the website and read shorter, easier pieces elsewhere.
Cost/Benefit Metrics for Reading
To formalize the model, we must quantify the costs and benefits of reading different articles.
Cost is easy to model: we calculate it as the amount of time it takes to read an article. For an intranet, this would be a direct cost in dollars, because we're paying employees for every minute they spend reading stuff during working hours. For a website, time is a more indirect cost, because users don't get paid to surf the Web. But still, life is short, and you only have so many hours in the day. Even if users don't get paid, they're still conscious of their time and don't like wasting it.
Benefit can be modeled by hypothetical benefit units that represent whatever value users get from online information. For a B2B user researching a company purchase, the benefit units translate directly into dollars, because they represent the extent to which the company gets a better deal or decides to buy a better product as a result of that user's time on the Web.
For home users, benefits might also have a dollar value. For example, if you're looking into buying airline tickets, the benefit of checking one more site or one more alternate departure time would be the average savings on the airfare that resulted from using a richer data set to decide which ticket to buy.
If people are browsing the news or reading an entertainment site, the benefit units would represent the amount of enjoyment they got from each page.
Example: Long vs. Short Articles
Let's work through an example, using the following values for our cost-benefit metrics:
600 words, meaning a cost of 3 minutes to read (assuming a reading speed of 200 wpm)
7 benefit units gained from reading each article
1,000 words, meaning a cost of 5 minutes to read
10 benefit units gained from reading each article
Finding a new article to read: 1 minute
The following chart shows how the accumulated benefit units increase as a user keeps reading short articles ( blue curve) or long articles ( red curve):
The dots on the curves represent points of change — that is, when the user stops reading one article, starts searching for something else to read, and starts reading the next article. No benefit is gained during the user's search time.
(Here, I use the term "search" to indicate any user activity aimed at finding the next interesting article, whether it's using a search engine, a site's navigation system, or any other method of finding the next thing to read.)
The chart clearly shows that users gain more benefit from sticking to a diet of short articles. The cumulative benefit is as follows:
Short articles: 105 benefit units per hour
Long articles: 100 benefit units per hour
The conclusion is clear: people prefer to read short articles. This is also what we've found in empirical studies of users' behavior while reading websites. People tend to be ruthless in abandoning long-winded sites; they mainly want to skim highlights.
Benefit of Cutting Word Count
If you read my assumptions carefully, you'll notice why the math favors short articles: I assumed that short articles were 60% of the length of the long articles but still provided 70% of the benefit.
Is this realistic? In most cases, I'd say yes. A good editor should be able to cut 40% of the word count while removing only 30% of an article's value. After all, the cuts should target the least valuable information.
When Long Has Value
Now, let's change the assumptions and assume that every third long article is 3 times more valuable than the previous two. In other words, 2/3 of the long articles continue to provide the user with 10 benefit units' worth of information, but 1/3 of the long articles now have a benefit value of 30 .
This scenario corresponds to the occasional situation in which you really, really need to know everything about a problem.
For example, consider a rare disease in which sufferers risk death if they eat 6 particular foods: 4 common foods, and 2 foods that almost nobody eats anyway. If you're reading about the disease out of idle curiosity, you'll probably be satisfied with a short article covering the four common foods. If you just got diagnosed with this disease, however, you won't be content reading an article that says: "there are 6 things that'll kill you, but we won't talk about 2 of them because they're rare." You'll obviously want the long article that will warn you about all the things you need to avoid.
The following chart shows the cost-benefit curves under this new assumption:
The blue line shows the progression of gains from reading only short articles (the same curve as in the previous chart). The red line shows the gains from reading long articles under the new assumptions: for every third article, the benefit jumps up and thus considerably outpaces the blue line.
The obvious conclusion is that long articles are better now that they're sometimes more valuable.
But there's a third behavior users can choose: a mixed diet, where they sometimes read short articles and sometimes read long ones. The green line shows this reading behavior.
For the mixed diet, we have to change the assumptions about the time needed to identify the next article to read. I'll assume that this now takes 1.2 minutes, versus 1 minute for the simpler scenario in which people always read a single type of article. This increase accounts for the extra overhead of having to consider both types of articles and decide when to read what.
Decisions take time, which is why it's often best to offer a simple user interface rather than one with many options. Every extra thing users can do requires consideration, which takes time away from actually using the features.
In this case, the green line is even better than the red line, because users don't waste time on the 2/3 of the long articles that aren't sufficiently valuable.
In the new scenario, users' cumulative gains from the different reading strategies are:
Short articles: 105 benefit units per hour
Long articles: 167 benefit units per hour
2/3 short articles + 1/3 long articles: 181 benefit units per hour
Mathematical Models vs. Real Life
Of course, in real life, you don't need in-depth information exactly every third time you read an article.
But the general idea in my model is extremely realistic:
Reading benefits vary, depending on user circumstances.
Most of the time, short articles contain more value per word.
People sometimes gain higher value from complete or very detailed information about a problem.
The exact numbers in my calculations are merely assumptions for the sake of the exercise. You can run similar calculations for your type of material and your type of users.
What Should You Do?
So: should your website have concise or in-depth content?
If you want many readers, focus on short and scannable content. This is a good strategy for advertising-driven sites or sites that sell impulse buys.
If you want people who really need a solution, focus on comprehensive coverage. This is a good strategy if you sell highly targeted solutions to complicated problems.
Typically, people who really need something are the highest-value users because they're more likely to turn into paying customers. That's why I recommended writing articles instead of blog postings.
But the very best content strategy is one that mirrors the users' mixed diet. There's no reason to limit yourself to only one content type. It's possible to have short overviews for the majority of users and to supplement them with in-depth coverage and white papers for those few users who need to know more.
Of course, the two user types are often the same person — the one who's usually in a hurry, but is sometimes in thorough-research mode. In fact, our studies of B2B users show that business users often aren't very familiar with the complex products or services they're buying and need simple overviews to orient themselves before they begin more in-depth research.
Hypertext to the Rescue
On the Web, you can offer both short and long treatments within a single hyperspace. Start with overviews and short, simplified pages. Then link to long, in-depth coverage on other pages.
With this approach, you can serve both types of users (or the same user in different stages of the buying process).
The more value you offer users each minute they're on your site, the more likely they are to use your site and the longer they're likely to stay. This is why it's so important to optimize your content strategy for your users' needs.