[This is a draft of a column I wrote for the ACM’s interactions magazine. It appeared in Volume 17 Issue 3, May + June 2010].
A: Do you love me?
A: Will you marry me?
A: Do you love us?
B: click click click
A: Do you love us now?
B: click click click
A: Do you love us?
B: click click click
A: Do you love us now?
A: What about now?
A: Hello? Where did you go?
For me, long gone are the days when the word “engagement” conjured up diamonds, parties, and champagne. These days engagement is all about divining how much love your users or the audience have for product(s) and/or application(s). In the Internet world, the word “engagement” is intimately associated with measurement, metrics, and monetization; it is all about clicks and conversion, visitors, page views and duration. This is the world of media marketing, and of Internet success and failure.
What is Engagement?
Before I dig into Internet engagement measurement, I want to step back and think about what engagement means to me. When first asked about engagement by a colleague at work, I spouted all I knew about the experience of engagement as I understood it from a psychologist’s worldview. Flow. Immersion. Fascination. The swift passing of time as attention is held—tempus fugit. Engagement is the absolute opposite of restless boredom; boredom is the experience of time expanding painfully, a feeling of being itchy and twitchy and ready for any external stimulus beyond what you are doing now.
From a psychologist’s point of view, engagement sits within a framework of arousal states—boredom at one part of the spectrum, engagement at the other. Altertness, forced focus and stress all sit somewhere in between. Perhaps the most famous approach to thinking about engagement is the concept of flow, proposed by Mihály Csíkszentmihályi. Flow is the state of full immersion in what one is doing: a feeling of energized focus, total involvement, and success in the process of the activity. Flow is a positive feeling, reflecting an utterly focused concentration and alignment with what one is doing. In some conceptions, flow is associated with feelings of joy and rapture, an exclusion of all but the central activity. Engagement may not be this rapturous positive feeling; people can be negatively engaged if something is standing between them and their desired state or goal, and for many people, military engagement denotes a negative form of entanglement.
But central to the idea of engagement is the triggering and capturing of attention, of being beguiled and perhaps focused to the exclusion of everything else. Further, lasting engagement can manifest in episodes of focus; perhaps someone will come back for more, again and again and again. This is where engagement in my early training intersects with the world of marketing and media engagement, the focus to which I now turn.
“Engagement” is a favorite word for media marketers and for the designers of products, services, applications. Take a look: The Web is brimming with smart people opining on the topic. In this world, marketers, data miners, statisticians, designers, and a host of other professionals come together. “Engagement is the most essential measure of the success of any medium. Engagement speaks to passion, preference, and habit,” said Charles Buchwalter, vice president of analytics at Nielsen NetRatings, in a 2005 white paper, “Integrated Interactive Marketing: Quantifying the Evolution of Online Engagement.”
In this world, the point is to stimulate engagement, to create it and make it last for whatever period is appropriate for your offering. In most cases, the goal of creating or “driving” engagement is to provide a platform for advertising products or services.
Models of how to engage consumers make assumptions about the level of active involvement of which the consumer is likely to be capable or willing to offer. At one end consumers are seen as being passively engaged with the content; think of the stereotypical couch potato glued to the screen while shows and advertising flicker across the TV screen. At the other end of the spectrum consumers are seen as actively seeking out content, developing their own understandings of that content, and then perhaps sharing with others. The active consumer concept is at the center of viral or word-of-mouth advertising and product engagement. Another example on the active consumer end of the spectrum is engagement marketing (which is also variously called experiential marketing, event marketing, live marketing, or participation marketing).
I like the idea that persuasion architects, who are responsible for creating campaigns to get people actively involved in advertising and engagement, are not completely in control of how the recipient perceives the message or whether or not they are likely to be engaged. I like to bestow upon the reader some agency in their own engagement. Borrowing from Tony Schwarz, a ’60s political advertising guru, who coined the idea of successful campaigns as those that invite people to “participate in their own manipulation,” I call this new world of Internet and cross-channel media play a form of “partipulation.” Perhaps we need “vongagement” to stand in for voluntary engagement? We need a word implies, entails and insists upon reciprocation, a word that acknowledges that the engagement is essentially about a relationship between a person and an activity or an entity.
Measurement and evaluation are key in good design research; the same is true of good audience studies—it is a good thing to be able to evaluate how successful you are at engaging your audience.
In the media world, there have been many scales and measures of engagement, most of them developed for TV viewing. Examples include Q Scores from Marketing Evaluations that measured “likeability,” launched in 1963, and Jack Myers’s Emotional Connections model of viewer engagement, launched in 1999. Since then scales have measured relevance, affinity, comfort and resonance, impression, opinions, recall, awareness of brands and products, memory for program details, emotional involvement, and, of course, moment-by-moment measurements of physiological and neurological effects. Most of these studies are with representative panels of people sampled from the population of viewers as a whole.
In the world of the Internet, measures of engagement focus on features, applications, sites, services, and brands. Here, the Holy Grail is generating engagement measures based on the activity of the whole audience, not a sample; in theory, as all actions take place online, the entire system can be instrumented for activity and therefore measured. The metrics and measures even here, however, are constantly in negotiation and transition. Leaders in the discussions surrounding measurement for engagement are data services like comScore and Neilsen, who work with massive data sets as well as with panels of participants to generate measures and metrics.
A central measure of engagement used to be page views—how many times a webpage was visited. In 2007 the “visits” metric—defined as the number of times a unique person accesses content within a Web entity with breaks between access of at least 30 minutes—was introduced as a way of measuring the frequency with which a person views content, thereby illustrating a key component of user engagement. To determine engagement one looks at total visits, average minutes per visit, average visits per visitor, and average visits per usage day. One of the reasons for changing measures is technological. “With advances like AJAX changing the Internet landscape, certain measures of engagement, such as page views, are diminishing in significance for many Web properties,” said Jack Flanagan, executive Vice President of comScore Media Metrix. “The introduction of these new metrics based on ‘visits’ provides an alternative for measuring user engagement that tells us how frequently visitors are actually returning to the site to view more content.”
Things are also changing as sites become more focused on the social Internet. A more extensive list, based on my review of about six sites dedicated to Web analytics between 2007 and 2009, is this one: There are “root” metrics that appear in all discussions of engagement and Internet use. Some of these are relevant for all Internet sites, and some are relevant only for some sites, e.g., sales apply only to sites that are actively selling something or sites that contain advertising that is intended to lead to a purchase. These “root” metrics are:
• number of unique users
• number of sessions a unique user had with the site/page/application
• duration of visit
• frequency of visit (returning to the site directly—through a URL or bookmark—or indirectly)
• percentage of visits that can be determined to be repeat visits for an identifiable single goal
• recentness of visit
• depth of visit (percent of site visited)
• click-through rate
• “bounce measure” (determined by number of visits that were less than a set number of seconds, which implies the person arrived at the site by mistake or arrived, found nothing they wanted, and left)
• subscriptions, and conversion from free to subscriber service
• sales of product(s) through site and/or referrals that resulted in partner sales
• lifetime value – the last being probably the hardest to evaluate
There are also action metrics, and social-engagement metrics based on elements that are key to social interaction design:
• content contribution (e.g., text, uploads – images, video)
• RSS feed subscriptions
• bookmarks, tags, ratings
• viewing of high-value or medium-value content (as valued from the organization’s point of view—note that depth of visit can be combined with this variable)
• providing personal information, such as filling out profiles
• downloads, content resyndication, and customer reviews
• comments (with some modulation depending on their quality, and some analysis of the ratio between posts and comments plus trackbacks)
Recent applications like Tynt’s Tracer tracks people’s copy-and-paste actions to see what you liked enough to copy and paste it, and many services track site recommendations.
Problems with Internet Engagement Measurement
The artful combination of measures with a clear view of what one is trying to achieve is key. Avinash Kaushik, an analytics practitioner and author of Web Analytics 2.0, calls for good design practice around Web analytics on the issue of engagement. His steps are:
1. Define why your website/application exists, asking, “If there were one thing your website would do, what would that one thing be?”
2. Determine the critical metrics (three or fewer) that will identify exactly how you can measure if your website is successful at delivering against its purpose.
3. Decide what engagement means to you in your context.
4. Don’t call that metric “engagement.”
He urges us to call a spade a spade, saying, “Think very carefully about what you are measuring if you do measure engagement. If engagement to you is repeat visitors by visitors then call it Visit Frequency, don’t call it engagement.” It is crucial to determine ahead of time what your goals are and what constitutes success. This is akin to working on a product and being clear as to what the primary goal of the product is. So many great ideas and great products are hampered by a version of “creeping featurism” that could be called “excitable expansionism.” Whenever I hear someone say “Yes, and we could do this as well!”, I cringe and try to drive the conversation back to the core issue we believe we are addressing—the core goal, need, or hedonistic channel we believe the product should exist to serve.
Engagement cannot be a one-size-fits-all calculation. It varies according to industry, organization, business goals, and so on. Along with this tailoring, one needs to be clear that the relative weighting you give any action depends on your goal. Your metrics and measurement research program depend on whether you are assessing a feature, an application, a site, or a business/brand; whether you are assessing the value of the product in itself for a single user or as a broker between social entities. There are (single) user-engagement measures, but these may not be the same as those that are appropriate for social media. A useful single-user metric might be clicks and content (i.e., how many features are clicked on, how much information and content is uploaded), but a useful measure for social applications may be reciprocity between the people using the application (i.e., how many links do you send me in response to what I send you—how conversational is the application, and how much sharing does it encourage?)
A Broader Yet Still Measurable Notion of Engagement
Before closing, I would like to broaden the idea of engagement still further. Taking my cue from media studies, I am inspired by the “consumer expressions” model and the “connectedness scale” model, both models applied to media like TV and film, but nevertheless fully within the broader and braver new world of partipulation. Or was it vongagement/reciprocative assessment?
The consumer expressions model, devised by researchers from MIT in 2002, suggested the development of a metric that combines qualitative information (the nature of the content, the media consumption environment) with quantitative information (time spent, satisfaction). In 1999 the connectedness model proposed by Cristel Russell and Christopher Puto focused on use of media including self-definition, creative contribution, social interaction around the show, ritualization (organizing and maintaining specific social functions of viewing and using), and the purchase of show-related paraphernalia. This was broadened in 2004, when researchers Russell, Norman, and Heckler conducted a validation of the connectedness scale that examined the extended contribution around a series of constructs—attitude (degree of favor or disfavor), involvement (mental state during viewing), and overall viewing (time spent overall in the medium). Their model, albeit particularly focused on TV and media events, offers six overarching dimensions, some of which perhaps could be extended to the world of digital and physical products and especially to media products:
• escape (immersion into the event)
• fashion (the extent to which the viewer is influenced by the characters’ appearance)
• imitation (the inclination to imitate the characters’ behavior or speech patterns)
• modeling (the degree to which individuals relate their lives to the lives of characters
• aspiration (aspiring to actually be on the show or meet the characters)
• paraphernalia (the degrees to which people collect items to bring the show into their real world)
As connectedness increases, so does frequency of show-related social interaction with others, the development of relationships within the community of co-viewers, and the size of the viewers’ social network of co-consumers. People also have a greater memory for the characters and storylines, and surely this kind of memory is key to long-term engagement as well as in-the-moment engagement. So a deeper sense of overall engagement, and a measure that moves beyond immediate activities into activities beyond the screen, mouse, click and brand awareness into behavior and identity. Indeed, consumption, participation, and identification are key. When I think of a social Internet site/service like Flickr, I immediately see how this level of analysis is useful as a complement beyond assessing clicks and content. The impact of a site or service like Flickr beyond the screen and beyond the session is what makes Flickr a place to see and be seen, not just a place to visit now and then. Avid Flickr users identify with Flickr; they don’t just see it as a place to view images.
My main point is that more integration between different measurement forms is needed to get a clear picture of overall engagement. This is systems thinking; it necessitates measurement perspectives from clicks to brand to identity; it requires a mixed-methods approach and the insight to see how a survey can relate to a field study can relate to a programmatic series of principled data collections … and so on.
In sum, there is so much more to engagement than immediate actions captured by a simplistic action-analytics model or reflected feelings on a survey. Engagement is more than immediate instrumental value and, as a topic, engagement is in need of a broader perspective. Apparently, engagement is itself quite engaging.