Content Experience Archives - Page 6 of 9

Calendars can rule our lives. Digital ones often seem to determine our choices, instead of reflecting our choices. Yet when structured appropriately, dates can be used in creative ways to enhance value for audiences. To do this requires us to think more conceptually about slices of time.

Dates are an important data format that can be linked to other kinds of data to deliver valuable information to audiences. Dates structure personal experiences, but should not define them. While some dates are certain and non-negotiable, such as a birthday or an anniversary, many dates are contingent on various factors such as cost or convenience. People often don’t want the date to be the main criteria driving what they seek. They would rather prioritize other considerations, and see what dates are best fits.

The dreaded date picker: bane of many users

Content designers often expect people to think in terms of precise calendar dates. A long-established paradigm involves making people specify date ranges. If users don’t like the answer, they are forced to enter a new search with an alternate set of dates. Even visual approaches, involving graphic displays of results and sliders, still require from the user unnecessary interaction. Too often, the user is left to query a database, and fiddle with specifying beginning and ending dates, to discover what’s best or most notable. Computers should be doing more of that discovery work.

People want to know about blips, cycles and trends that can be hard to notice. They don’t want to do lots of work to uncover these insights. Computers generate time-related content by asking users questions about specific dates. Yet people really want answers, not the ability to specify queries. An over-focus on question-asking capabilities can become burdensome. A better experience would provide answers to common questions involving date ranges, without demanding that people enter specific dates when it is not necessary. How can one offer useful answers without audiences asking? How can we stop making people hunt for answers? One key is to frame answers in terms of chunks of time.

Three Kinds of Time-Slicing

Date ranges can be thought of as chunks or slices of time. Three kinds of time slices provide data:

Disjointed
Overlapping
Cumulative

Each orientation surfaces different information of value to audiences. Different forms of date ranges answer different questions. Slices define the characteristics of an event for a person: how long the event is, when it starts and stops, and the advantages or disadvantages of a given period. We can offer different kinds of answers using different time-slicing patterns.

source: Borner & Polley, Visual Insights, MIT Press (redrawn) — source: Börner & Polley, *Visual Insights*, MIT Press (redrawn)

Disjoint (Sequential) Time Patterns

When audiences think about time as calendar months, or weeks that start on Sunday and end on Saturday, they chunk the time into discrete units. Disjoint time patterns apply anytime people can’t do multiple things simultaneously. For example, they can’t be in two separate cities at the same time, or they can’t invest the same fixed sum of money in two different funds at the same time. Such time periods can’t overlap.

Some periods will be predefined, perhaps according to a common convention, such as holiday weekends. In other cases people want to define the length of a chunk, and then be able to arrange the sequencing according to a set of criteria.

Disjoint chunks of time can answer:

What holiday weekend had the highest grossing film this year?
What companies were most frequently mentioned in the business press 48 hours immediately following the last Federal Reserve interest rate decision?
From a cost perspective, is it cheaper to first visit City A for two days, then City B for three days, or vise versa, given hotel availability?

When comparing two different slices of time, the slices may be either spaced apart, or contiguous. Slices that are spaced apart will generally be conceptually similar or equivalent to create a meaningful comparison, such as when comparing holiday weekends.

Contiguous slices are more flexible. They can be used to explore possible causal relationships, such as revealing performance or activities during some duration that follows some other event.

The other use of contiguous time slices is to determine the optimal sequencing of the slices. Trip planning is a common task that involves sequencing chunks of time. It can be frustrating to re-enter arrival and departure dates to calculate options and costs. People want to know that hotels are in short supply in City A during a certain weekend but will be widely available after that weekend. They would rather be in City B on the weekend, and go to City A after. There’s an opportunity to simplify the task for users by highlighting answers based on headline-costs associated with chunks of time, instead of focusing on transactional options and details.

Overlapping (Fluid) Time Patterns

Overlapping time periods look at variations in blocks of time of a set duration. It typically answers question such as when is the best time to do something, or when was the most active period for an issue.

Overlapping chunks of time can answer:

What five-day period will have the lowest average hotel room rate?
What two-week period is statistically most likely to have the most sunshine?
What 30-day period had the highest return for a specific stock?
During what 10-day period last year was a specific product most frequently mentioned on Twitter?

The duration may reflect a length of time that’s significant to the user. Owners of stocks, for example, may gain a tax benefit by holding a stock for a minimum of thirty days, so they are interested in the return for that period. Durations for other topics reflect a sensible “window” to look at a variable, because sustained performance over this period will be considered significant.

Many times an answer to a well-framed question will spark a follow-on action. For someone looking to visit Seattle when it is not foggy or raining, knowing the sunniest days there will be enough to book travel during that period. For answers that reveal periods of high or low performance, the user may be interested in looking into what was happening during that period that would explain the performance.

The user can also explore how changing the date range to an earlier or later time slice changes the results. Ideally, they can simply indicate the concept of “earlier” or “later” to modify the answer, rather than having to enter specific dates.

Cumulative (Maximizing) Time Patterns

Cumulative time patterns answer the question of what period produces optimal value. Sometimes value is determined in terms of a fixed resource that gets depleted: the goal is to maximize the duration before the resource is gone. Other times the value is open ended, and the goal is to locate the best period when the total value of a resource can be maximized.

Cumulative chunks of time can answer:

What’s the longest number of days one can stay at a ski resort for $3000, and when is that?
How long and when was the longest winning streak for a sports team?
In what time periods does a city have a sustained above average number of visitors?

The example of booking a stay at a ski resort flips how making a reservation is typically framed. Some people start with a budget and want to know how they can get the most value from that sum of money, so they can to discover dates, rather than input dates.

The user’s goal can also be to minimize a value. Suppose the goal is to make sure bad timing doesn’t spoil your vacation. You want to make sure that when you visit a medium size city, that it doesn’t correspond to the timing of a big doctor’s convention or trade show for industrial engineers. In the past you’ve found it difficult to make restaurant reservations in cities hosting such events. So you might want to screen for times that are less busy. How that screening is performed in the background is irrelevant to the user, though the content designer could draw on various indicators from airport data to hotel bookings data to provide a signal.

Windows of Opportunity

Think about what information audiences most want to know in relation to a slice of time. What are the windows of opportunity as the audience sees them, and how do they define what’s important? What information, on what issues, can support decision making by an audience, or spark interest in a topic and encourage deeper engagement?

While data plays an important role, the database should be in the background. The goal is not to give audiences the ability to ask any question, or to supply answers to any scenario. Rather, the goal is to identify key issues of interest to audiences, and find ways to answer questions about these issues with a minimum of user effort.

Once issues are identified, content designers need to determine if they have the information available to provide the answers. A powerful combination results when the content designer can integrate internal time-based data, with external time-based data tapped through a third party’s API.

Although the questions answered may be factual and data-oriented, the answers can be enhanced with interpretation. Date-centered questions provide writers with opportunities to provide context to answers. These may be in the form of articles about “best times to” do an activity, or background explaining notable episodes relating to a thing or person.

Content design should look beyond stated requirements to think about opportunities that provide additional value to audiences, in ways they haven’t yet articulated.

— Michael Andrews

Scene setting: the Curatorial Perspective

Scene setting lets people understand common themes in content that aren’t obvious. An author or curator draws on their unique knowledge to construct a theme that unifies different content items. Such themes set expectations about the relationship of content to other content. It is didactic in orientation.

A common label used to announce a theme is the series — for instance, a TV series, or a narrative trilogy. Sometimes the series is just a way to divide up something into smaller parts, but keep them connected: an article becomes a two-part article. A content series can express how different items are related according to the intentions of the author or the interpretation of a curator. They can be a sequence of items presented on a common theme. The series may present the evolution of the item over time, such as versions. A building architect might show a series of images starting with a sketch, then a foam model, and finally a photo of the finished building.

A series presents a collection of items and shows how they belong together. The author/curator draws on their intimate knowledge of the content to point out connections between different content items, which may not be self-evident. We find this in the museum world: an item presented is said to originally belong with other items, that have since been dispersed. A curator might indicate how several items embody a common theme, such as when similar paintings express a recurrent motif.

Art curators identify series of related Van Gogh paintings (via Wikipedia). These three are more similar than others he painted on the same subject.

Any time items are defined by the values and judgments of the author (or curator), the audience must be willing to accept that valuation as relevant. So if a curator identifies items as “new and notable,” then the intended audience needs to buy that labeling.

Mirroring: the Audience Perspective

When mirroring, content reflects themes as seen by the audience. It represents concepts the way audiences think about them to support attraction to the content. Mirroring is different from the authorial perspective, which expresses the content’s intention. The audience perspective expresses how content is imagined.

Brand names are perhaps the purest example of imagined content. Brands have no intrinsic identity: they depend entirely on the perceptions of customers to define what they mean. Even a conglomerate that sells many brand products can’t dictate how consumers view these brands. The French brand house LVHM, which sells numerous luxury brand products, can’t control whether consumers consider Dior is more similar to Givenchy or to Louis Vuitton, even though it owns all three brands. In reality, Chinese consumers may have different opinions about these relationships than Italian consumers would.

Part of a dendrogram showing perceived similarities between different luxury brands, from a study at Woosuk University in Korea

High-level concepts that are meaningful to audiences should reflect how audiences perceive them. For example, people associate different kinds of experiences with different vacation activities. Is bungee jumping active-fun, adventurous, or extreme? It is best to work with the audiences’ framework of values, rather than trying to impose one on them. Card sorting is useful for eliciting subjective perceptions about the identity of things. Yet card sorting is less reliable when defining the identity of concrete things, since it shifts the attention away from the object’s specific properties. Better, more empirical approaches are available to classify concrete items.

Discovery: Perspectives based on Item Properties

Features of items can suggest themes. Object-defined themes let the things featured in the content to speak for themselves. This involves more showing, and less telling. Properties can define identities, and reveal commonalities between different items. It promotes discovery of content relationships.

Faceted search interfaces, such as found on e-commerce sites, are the most familiar implementation of property-driven identification. People choose values for various facets (properties) of items, and get a list of items matching these values. Using properties to identify items is especially valuable for non-text content. Some Digital Asset Management systems allow people to find images that match a certain shade of a color, regardless of what the subject of the image is. Properties can identify similarities and relationships that might not be expected from a higher level label. It can support more criteria-based consideration of identity. For example, when we think of travel items — things to pack — we generally have standard things in mind: toiletries, articles of clothing, etc. But if we start with properties, the universe of travel items expands. We might define travel items as things that are both small and lightweight. We discover small and lightweight versions of things we might not ordinarily pack for travel, but might enjoy having once we become aware of the option.

Generative classification of objects according to properties by P Harni, via Aalto University

Generative classification of objects according to properties by P Harni, screenshot via Aalto University

Leveraging Diverse Viewpoints

There’s more than one way to define the relationship between items of content. I sometimes see people try to make a single hierarchical taxonomy serve as both an authoritative or objective classification of content, and a user-centric classification that reflects the subjective perceptions of users, without realizing they are forcing together different kinds of content identities — one relatively stable, the other contextual and subject to change.

Content can be considered objectively as it is; authoritatively as it is intended; and subjectively as it seems to various audiences. These differences offer thematic lenses for looking at content. They can be used to help audiences connect different items of content together in different ways: setting the scene for audiences so they understand relationships better, reflecting their existing attitudes to promote attraction to items of interest, and helping them discover things they didn’t know.

— Michael Andrews