In the last 10 years, the rise of excellent search engines has changed the way that sites approach findability of their content. On many modern sites, less than 20% of sessions start on the home page, because users are pointed to some internal point in the site. This has in some ways obscured the critical role of websites to assist users in traveling the “last mile” from the site to their desired target…Read More
This project was a collaboration between Daniel O'Neil from The Understanding Group and Eric Kushner from Project Two Paths, Ltd.
It has never been easier to get or analyze data for your organization. Massive data sets that used to be available almost exclusively for the Fortune 500 exist all around us, usually collected almost by default, and at minimal cost, by the systems that we use to run our companies.
From Google Analytics to our merchant and fulfillment systems, insightful customer-centered marketing strategy is just a spreadsheet away! What is key about this explosion of data is that it requires an even greater commitment to story-driven strategic marketing. If data is not judiciously applied to the human narrative, marketing departments can be pushed into reactive, tactical activities against tiny bits of information. We think you should go big instead, and use the data for high-level marketing strategy and planning.
The approach uses three major principles:
Marketing teams know their audience and should tell stories about them.
Those stories should be testable with data.
The collaboration between storytellers and data analysts should end with compelling user stories that drive focused, specific marketing campaigns.
Recently we had an opportunity to apply this framework to GamerSaloon, a brokering service for players who wanted to find opponents for cash video game competition. GamerSaloon’s revenue model is largely driven by frequency so they asked us to determine how to maximize player engagement based on the large dataset that contained ten years about players, the games they played, and their win-loss records.Players who sign up for GamerSaloon have a wide range of usage patterns, from playing multiple times a week for years, to trying only a game or two before moving along. A significant portion of accounts on GamerSaloon were inactive—that is, the player had signed up but had never actually played a game. We were asked to find out what these players needed to continue playing or why they never played in the first place, then develop a marketing strategy to engage them.
Getting Started: The Creation Of Initial Theories
We started by doing stakeholder interviews with the marketing and customer support team, as well as creating accounts and playing a few games. Our findings allowed us to create several customer segments, each one of which represented a story about a kind of player behavior. Our titles were designed to be human and memorable. Here are three examples:
Hey big spender: Players who deposit larger amounts initially are more likely to play again
The power of first impressions: The outcome of the first three games has an impact on long-term player value.
Interval training: The shorter the pause between the first deposit and the first game, the greater the long-term player value.The core concept used for the marketing analysis for GamerSaloon was an engagement funnel, where long-term, repeat game play was the desired end state. The initial theories were chosen with an eye towards inputs that could, based on our experience, be turned into specific messaging campaigns to improve the performance at targeted points in the engagement funnel from initial site visit through early game play.
The Acid Test: Checking The Theories Against User Data
Once the hypotheses were created, we examined user behavior data through three lenses:
How many of the theories were actually testable given the state of the data?
Were the theories supported or refuted by the data?
What additional patterns seemed to emerge from the data analysis that can be applied in actionable marketing campaigns?
The quality of the data on the player platform was extremely high, so we were able to test almost all the theories. As is typical when using this approach, about half the theories were supported by the data. Among the most notable findings was that the size of the initial deposit had a negligible impact on a player’s lifetime value and that a sizable number of depositors had never played a game. We also found a significant correlation between a player’s long-term value and their win-loss record in the first few games.
The most notable pattern that emerged from the data analysis was the identification of an inflection point for lifetime value in new players. We were able to define the numbers of games we needed to foster for a new user to substantially increase the likelihood that he’d be a "golden goose" versus a "one night fling."
Tying It Together: Creating An Actionable Plan
Once we gathered the data, we used them to support initial hypotheses, each carefully crafted with if/then clauses that could be effected through marketing strategy. The answer to each question illuminated a revenue optimization or customer acquisition strategy, and if it didn’t, we set it aside. The findings were valuable because we took the time to identify the addressable variable and approached it with ideas on how findings would impact our marketing recommendations. With a few clear options pre-conceived, we knew that regardless of the findings, we could act on them to positively impact the business. And again, in the instances where that was not the case we tabled the the hypothesis.
Where did these notions come from? We started by examining the mechanics of customer acquisition with our client’s business as well as their revenue model. At the end of the day, there are a finite number of ways to drive more revenue and they are as simple as they are universal:
More frequent transactions
Higher value transactions
More transactions over the life of each customer
There are nuances, of course, such as time to peak spend and cost of customer retention and customer service and the like, but those are the nickels and pennies. Work through those after you’ve uncovered the quarters and dimes. Final tip: Start at the top of the funnel and work your way to the revenue end of it. Enhancements at the top compound on one another and earn the marketer the time to work at the narrow end of the funnel where the stakes are higher and the audience, narrower.
Outcomes And Final Thoughts
Identification of user experience obstacles for first time site visitors resulted in the development of a new user registration wizard that, combined with targeted email interventions, drove conversions up by nearly 7%. Similarly, the data revealed that many users failed to leverage the full breadth of the community. An issue that we mitigated with an email primer series that boosted game play from new users in their first two weeks by more than 100%!
What’s most notable about this project is how, even after years of cheap, widely available data, most companies are not making the effort to address their marketing channels with this approach. There are huge opportunities by creating user stories and big data for companies and organizations of all sizes to better serve their customers, grow their company, and increase their bottom line.
As we described in a previous blog post, websites need guides, the friendly voice helping people discover spaces where they can learn, work, and play. They act as a findability strategy for your users to locate things on your site.This post gets into more detail about the oldest of all guides: the map. Maps are most important when your users need to see the broad context of possibilities related to an idea, especially if the user can’t recall what the original idea was. To put it another way, maps are good for narrowing and remembering things.
The Map: Learn And Remember
Maps teach and inspire. They tell you where you are and where you might be able to go. This depiction of the world also makes assertions about what is important to focus on. They are more than spatial depictions of the world; they are usable tools that make assertions about what is important in context.As a map, your site helps users see the broad context of possibilities related to an idea.
What Makes Maps “Mappy”?
From an information standpoint, a map is a representation of relationships expressed and explained in context.By this reasoning, we are surrounded by excellent maps that aren’t necessarily mappy. In fact, the first really good travel maps didn’t look mappy at all.T
he Roman military created maps that were long, narrow scrolls of 20 feet. They showed a path down Roman roads from city to city. The map depicted distances of thousands of miles, all spooling along these roads. It told a story about a world where Rome was the anchor of any journey across the empire, and the roads were the connective tissue that held the empire together.
The approach clearly resonates in the map finding nature of our world, because in the 20th century it was replicated, in almost exact design, in the AAA TripTik, a multi-page road map that showed your journey from point to point in a series of 80-mile long tiles.
Another pathfinding map archetype is almost any major light rail commuter map. Note that this map of Chicago's light rail system is rotated 90 degrees from the actual orientation in order to more cleanly show the line stops and make room for the map’s annotation.
But maps aren’t always implicitly spatial. One of my favorite maps is of Pullers, a kind of heavy duty wrench used in manufacturing and auto repair shops. This map shows the relationships in context, annotates their use, and prioritizes their selection. It’s not spatial at all, but as a map of use, it is profoundly helpful.
Are Maps Right For You?
So where do maps help us in the online world? They are less common than other guide strategies, but are important in high-complexity, low-knowledge contexts.
There are 3 places where maps work wonderfully: calendars, filtered search menus, like those on a search result page in Amazon, and planning websites.
StubHub.com, for example, is an excellent planning website, with a remarkable map for purchasing tickets. It is the quintessential map: functional, representational, and ably displays information in context.
Beyond those 3 archetypes, there aren’t many explicit maps in the modern internet. This reflects a broader information pattern, which is that most online strategies assume that the assembly of understanding and context happens AFTER things are found.
To put it another way, the internet assumes you are going to start at some level of understanding and then go deeper into the thing you are trying to learn about. Maps then will be the focus at the end of a journey, not as a way to reach it.
Whether or not this should actually be how we manage or process information is, to me, one of the biggest questions about the future of knowledge in the modern world.In the meantime, take a look at your own digital footprint. When would you need maps? Are they necessary to help people make a journey? If so, what would yours look like?
Sites need guides, the friendly voice helping people discover spaces where they can learn, work, and play. Most sites try to do this by adding three major navigation strategies: Direction Signage, Maps, or Indices.
Previously we described each of these strategies at a high level and talked about how you can start to think about the right one for your site. This post gets into more detail about the original search tool of the literate world: the index. Indices are most important when a domain structure is known to the people using it in order to quickly look things up.
The Index: To Look Up In A Known Structure
As an index, your site helps users get more specific understanding of things they already know something about, in a system they are comfortable navigating. This is often in the form of completing a thought in the form of a word, phrase, or discrete metric.The key to the index is that the structure is KNOWN.
We all know, for example, the order of the ABC’s - the classic lookup model of the literate world - or that 5 comes after 4. Some structures are more domain specific, like the sections of the Glass-Steagall Act or the sections of the Dewey Decimal System. By knowing this structure, a user can look things up quickly and accurately.All indices rely on some technology - a printed page, a codex, a card catalog. To start, you need a word native to the system, so that you are able to find that term in whatever internal structure exists. The most powerful modern instantiation of the index - a well-managed search engine - needs ONLY that word, but the user still comes with some notion of what they want.
Indices In Computer Systems
It turns out that, outside of search, indices are not as common as signage, but they are still important parts of our computer systems. The most common index a computer user sees is their music application; if they are using a PC, it might be their file share.
These feel pretty “indexical”: a list of files that can be sorted by a range of attributes and a search box.Indices abound in “expert” systems on the web as well. The most classic expert indices are api and programming references, such as the javadoc file. Its categories are sorted hierarchically, in alphabetical order, and within any given class by repeatable elements that can be quickly used to find information. This is a classic index.
Another index is the menu and ribbon structure of word processing and spreadsheet programs. The nuances of this kind of index are so great and complex that it’s worth another blog post. But the short story is that, once upon a time, the menus represented signage. Now they have so much in them that they aren’t really signs at all, but collections of concepts that people have memorized as structural categories.
Is An Index The Right Guide For You?
Sites rarely just need a single guide, but they should develop and extend them based on the specific needs of their audience and the content at hand. TUG usually starts with a combination of a site’s complexity and the user’s domain knowledge:As we look at this matrix, we see why indices are less common.
High domain knowledge is not something that most systems on the internet assume. In fact, the internet is based on building mastery from LOW domain knowledge. Search is the big exception to this, and the failure of search in many contexts is due to the fact that the searcher doesn’t know the right terms to find what they want, and the search engine cannot help them figure them out.
In short, index guides have been relegated to fairly specialized or high-knowledge domains. But when an index is useful, it is VERY useful, often being the most commonly referenced part of any system or knowledge base. If you are in a high domain knowledge space and want people to find things quickly, an index guide may be for you.