DonorCast NewsWatch

Analytics vs Instinct

2008-08-28T16:10:00.007-05:00

Many thoughts I have introduced in the DonorCast NewsWatch cover the topic of “quality” in data mining and predictive modeling. I came across this article and realized that while I have made suggestions and raised questions about how to, for example, build a model predicting major donor likelihood, I have done little to discuss implementation of this work. I want to use this post to address one of the implementation challenges I encounter most: analytics (i.e. modeling scores) vs. instinct (i.e. VP's institutional experience).

While analytics and predictive modeling is not a completely fresh concept in the philanthropy world, it is young enough to be both misunderstood and mistrusted by some. After all, higher education, health care, and the arts were successfully completing ambitious campaigns long before RFM scores became a standard tool. Many in the philanthropic community still rely heavily on “gut feeling” or instinct for determining a donor's intention or affinity, prospect assignment, or more broadly, campaign readiness and viability.

The post I found discusses a summary of Ian Ayres' conclusion from his best-selling book, Super Crunchers, that “intuition and experiential expertise is losing out time and time again to number crunching.” I agree with the author who asserts that while data mining can offer concrete, and in some cases unforeseen insight, there is still an important role in business (or in our world, philanthropy) for experience, personal understanding, and basic qualitative characteristics.

Josh and I both often recommend that analytics be blended with organizational experience and environment. Achieving an effective balance may prove tricky. Convincing members of the “gut” society to buy into analytics integration may prove trickiest.

To show the value of analytics integration, try a simple control group. If you create an annual giving model, take 100 names at random and make your appeals. Then take the 100 highest scoring in the model not in the control group and offer the same appeal. Compare renewal rates and gift amounts. You may surprise people with the results.

Analytics versus Good, Old-Fashioned Creative Gut Feeling

I really enjoyed a recent post I found on the Precision Marketing online magazine. Jenny Hoffbrand discusses Ian Ayres' new book called Super Crunchers and a quote from the book that really summarizes the value of using analytics in the business as opposed to relying on your “intuition” or gut-feeling: “Intuition and experiential expertise is losing out time and time again to number crunching. Businesses and governments are relying more and more on databases to guide their decisions.”

Read More

Profiling Your Donors: What Data Should You Append?

2008-08-05T12:27:00.010-05:00

Here is thoughtful article that discusses some of the most common external data acquisitions that Josh and I encounter in our work. While Austin does a fair job laying out three basic sources of external data, I wish to add some specific examples where they might be used, as well as some thoughts to consider.

External data acquisition can be a powerful tool for any organization—but like most tools at our disposal—it should be applied strategically. Instead of starting with data, start with some program goals:

Identify new major gift prospects
Increase the participation rate in the annual fund
Discover planned giving opportunities

Once a goal has been identified, review your database to determine which data points are present and which are missing in respect to your goals.

Using the example program goals from above, here are some data acquisition points to consider.

Identify new major gift prospects (Wealth/Capacity Screening)
Increase the participation rate in the annual fund (National Change of Address Screening)
Discover planned giving opportunities (Deceased or Age Overlay)

What is a lesson that can be learned from this? Be very thoughtful when acquiring external data, as it may have more limited applicability than you might think.

Lastly, a development shop should never let external data be the band-aid to record keeping and data entry problems. No one should have better information or a deeper understanding of your donors than you do.

Demographics—Who Are They?
What you should know about profiling your donors
by Don Austin

At some point, most nonprofits ask the question, "Who are my donors?" It seems intuitive that if you know the characteristics of your donors you can market to them more successfully.

Answering this question usually means, "profiling" your donors. While this might sound easy, the process is not always straightforward. Profiling involves, first, overlaying demographic and lifestyle data on your donor file. Second, in the profiling step, you will have to choose between two methods to develop a picture, or pictures, of your donors.

Before you decide to begin this process you should ask yourself how you will specifically use the information and how you will justify the cost. You might find that a simple overlay of donor age will suit your needs.

Book Review—Fundraising Analytics: Using Data to Guide Strategy

2008-08-05T12:21:00.007-05:00

Here is a very in-depth and thoughtful review of Josh's book. The feedback for this work has been tremendous—it has even become standard reading for MBA programs.

If you have not yet had a chance to pick up a copy, read this review, and see if it might be useful in your work/professional development (I am betting it will be).

Fundraising Analytics: Using Data to Guide Strategy
Review by: Gayle L. Gifford, ACFRE, CharityChannel

Fundraising Analytics is a gift to the masses ... a lens into the world of the sophisticated fundraising operations that pump the big bucks into major US institutions. Written by Joshua M. Birkholz, the director of the analytics division of Bentz Whaley Flessner, a major fundraising consulting firm, the book’s subtitle is “Using Data to Guide Strategy” and that’s what the book delivers.

I read this book from the perspective of the majority of US charities (82%) – the ones with the budgets below $1 million. At first glance, this book might seem an irrelevant fantasy fit only for the top strata of charities. None of these nonprofits have the legions of prospect researchers, major gifts officers, data analysts, and annual fund managers discussed in this book. Heck, it’s a lucky find to encounter a small organization that has even one fundraising professional and/or a functioning donor database from which one might extract the kind of information that Birkholz discusses.

But don’t ignore this book. Ease your way. Try jumping ahead to Chapter 5, Data-Driven Prospect Management, and you’ll find a wealth of easily comprehensible wisdom on running a fundraising program that is worth the price of the book.

Read More

Partnerships and Brand Loyalty

2008-07-15T10:35:00.003-05:00

Perhaps as a provider of services in the nonprofit community, it is impossible to write about all of the recent partnerships and brand loyalty campaigns without portraying a sense of bias. Nonetheless, I will make an attempt and encourage you to reach for that proverbial grain of salt. I am often asked to comment about these changes. The following is my brief attempt to do so.

As a resident of the Minneapolis / St. Paul area, I frequently fly Northwest airlines. Since I often need to work at airports, my membership with the WorldClub lounge more than pays for itself in saved internet costs and accessible work space. This membership also enables me to access Delta and Continental clubs. However, when I am in an airport that only has a Delta club, I am enormously frustrated. I have nothing against Delta. However, their club has a partnership with T-Mobile for internet access. I am required to pay additional for my internet access at the club through this arrangement.

My cell phone company has its own power cords made for the phone. The labeling says to use their brand of power cords. Generally, I find less expensive chargers made by other manufactures. These alternatives provide me with flexibility to plug and play other devices as well. There is no need to buy from the cell phone company when a better option exists.

How often do people use Mozilla instead of Internet Explorer because of features or even just principle? How many people have an Apple iPod even though they have a Windows computer? Do you only go to the dealer for the service on your car? Are all of your golf clubs the same brand?

I believe most people are intelligent when it comes to purchasing the right things for their situation. Whether it is for cost, services, convenience, or the overall best fit, people will set aside blind brand loyalty.

When it comes to your organization, do you exercise the same discernment? Do you choose services that are the best fit for you? Or, do you chose services that are the best fit for your software vendor? Do you build your predictive models to maximize the potential of your own existing data? Or, do you purchase models that seem conveniently interchangeably with the ones your peers purchased.

Among the most valuable contributions of analytics is allowing your data to guide your strategies. In this time of partnerships and brand loyalty campaigns, I only encourage you to exercise discernment. Do what is right for you. Do what is right for your organization. Your data is your most valuable asset. Leverage this asset as your advantage. This data, after all, is a reflection of your donors. When your donors are plugged into your decisions, you will make the right choices.

Data vs. Science

2008-07-02T11:00:00.006-05:00

It has been awhile since I have posted on the DonorCast NewsWatch. Alex is so in-tune with the data mining world, I have had little to add. However, as a long-time Wired subscriber, I could not go without mentioning the latest issue, "The End of Science."

Chris Anderson sets a premise followed by several other contributors regarding the modern use of data. One of the most provocative points of the feature is that the scientific method can actually get in the way of data exploration. I believe Chris is correct.

However, as with most debates (endogeneity vs. exogeneity, in-house vs. outsourcing, prospect-based tracking vs. project-based tracking) it is not as simple as one or the other. James Cheng, the brilliant data miner at MIT, presented a compelling case for the scientific method at the APRA data mining symposium this past April. Before changing strategies based on analysis, I use control group tests whenever possible. Kate Chamberlin and Michelle Paladino at Memorial Sloan-Kettering very effectively use controlled study principles in testing the validity and effectiveness of both models and development strategies.

There are times when the method can get in the way, too. Chris Anderson points out that Google does not try and understand "why" before implementing the results of the analysis. It simply moves ahead with it. In the writers own words:

Google's founding philosophy is that we don't know why this page is better than that one: If the statistics of incoming links say it is, that's good enough. No semantic or causal analysis is required. That's why Google can translate languages without actually "knowing" them (given equal corpus data, Google can translate Klingon into Farsi as easily as it can translate French into German). And why it can match ads to content without any knowledge or assumptions about the ads or the content.

When I first began to build predictive models, I always started with a hypothesis. This would influence my data selection as well as my model selection. The more I build, the more I move to allowing the data to guide the process. In fact, the most challenging part of CRISP-DM is the "data understanding" step. If I let my data decisions be guided entirely by what I understand the business question to be, I might miss a hidden pattern.

What is the risk of letting the data guide the process? Well, let me use a major giving model as an example. As I have discussed before, if your goal is to predict giving likelihood to risk-manage a gift pyramid, you may wish to have a large degree of endogeneity. Like a credit score, you really would want to know the probability of the behavior. If your goal is to find new people that might be good major giving prospects, you might choose to minimize endogeneity. The result would be less predictive, but would serve to minimize the identification of names already known to you.

In this scenario, if you were to allow too much endogeneity in the identification model, the risk is small. You would likely exclude researched and assigned names before starting your qualification process anyway. What remains are not known names. But, you might have missed some names that would fit the profile if you had more data about them.

Sometimes, I have Marianne Pelletier's voice in my head, "Well, did you find more prospects?...That's good--isn't it!?" It is very similar to Google's "Did we make more revenue on that ad?...That's good--isn't it!?" Sometimes "why" can get in the way.

Maybe, this is my long way of saying, "Buy this magazine and read the feature." It can be confusing at times since it references "models" in the context of "ways of doing things" as opposed to statistical models. But, I think you will see Chris Anderson's point. It is worth the read.

Read The End of Theory
Link goes to the first essay from the feature by Chris Anderson. See the links on the left side of the page for the other brief essays.

The World of Phone Service is Changing...

2008-06-09T14:44:00.006-05:00

A new study says 3 in 10 get all or most calls on cell phones, and I am certain that number will only rise in the near future.

Nearly 1/3 of those under the age of 30 have cell phones only.

In general, people are more private with their cell phone use. They are often more reserved with giving out this number, and enjoy the decrease of direct marketing calls compared to landlines. There is no "directory" for cell numbers-which is both good and bad (depends on who you are and what you want).

Keeping aware of this technology shift is important for those who do modeling and use "preferred channel" type categories as independent variables. It may also be important to the annual fund folks, where phone solicitation is still a tried and true method of raising money. Perhaps this shift might imply an increase in email or online solicitations to targeted groups as opposed to trying to reach them on the phone? Or a comprehensive program to acquire cell numbers of recent grads?

All the wonderful messaging and strategy in the world is useless if we have no way of contacting our donors. Being aware of trends like these is vital.

For nearly three in 10 households, don't even bother trying to call them on a landline phone. They either only have a cell phone or seldom if ever take calls on their traditional phone.

The federal figures, released Wednesday, showed that reliance on cells is continuing to rise at the expense of wired telephones. In the second half of last year, 16 percent of households only had cell phones, while 13 percent also had landlines but got all or nearly all their calls on their cells.

Read More

Man vs Machine, or "Numbers" vs "Guts"

2008-06-09T14:38:00.010-05:00

Through my analysis and recommendations for a variety of clients, I have seen first hand the tension and complex relationship of what I like to call the “pre-analytics world” and the “post-analytics world.” This convergence of two almost fundamentally different perspectives on organizational and campaign planning is still very fresh in the world of fundraising. Analytics represents progress to many in our industry—insights and capabilities based upon a new process of information gathering and analysis. Unfortunately, this evolution (or some might say revolution) has been strained at times.

Many appreciate the technical ability and metrical sophistication gained from analytics and modeling. For some, it is difficult to grasp the concepts used and understand opportunities for application. For others, it is difficult to embrace and trust the insights gained.

Provided with a reasonably well-stocked database, I could offer not only predictions on an institution's future, but also “blind” insights and analysis on what has been happening to-date. Without knowing the information, I could tease out the shift in annual fund messaging strategy, suggest which gift officers were performing well and why, and even reveal strategy for prospecting and solicitation. Impressive? Perhaps. But what happened to good old fashion “gut feelings.”

In the example I present, experience, the strongest factor used in “gut” decision making, is completely absent. I have never spent an hour inside the institution whose profile I could construct. I may offer new insights and perspectives—but don’t really know XYZ University like the VP does. The VP knows the shop and the donors, and feels the campaign is a “go” despite the reservations I might provide.

I can understand why a VP might feel hesitant to plan campaign strategy around analytics work he/she barely understands from someone who doesn’t know the institution as well as he/she does. It’s the institution's campaign, but ultimately his/her job on the line. Beyond campaign success, part of that job is also embracing new ideas and technologies. While he/she may never want to have a fully analytics-driven campaign—rejecting these tools may brand you as a fundraiser from the “20th century,” a wholly undesirable title.

What is the future for “gut decisions” in our world? I truly hope they never go away—and I doubt they ever will. All the modeling in the world could never replace a highly skilled gift officer, or savvy VP. Yet these two groups: pre-analytics (gut and intuition decision-making) and post-analytics (metrics and analytically rooted strategy) are more and more seen as clashing, especially when considering the increased respect and weight given to analytics in fundraising.

What can we do to bridge this divide, and to integrate the best qualities both these approaches have to offer?

This article posits a similar question. While the author does not attempt a thesis-like response, she does offer one sobering and often overlooked factor: “You can't predict emotion with a machine.”

Last week's episode of The Apprentice, filmed at Ogilvy, proved that marketing does not come naturally to everyone. Which is why decades of admen have been held in great esteem for possessing an instinctive ability to produce great campaigns. But, increasingly, the traditional reliance on intuition as the basis for a successful campaign is being surpassed by evidence-based decision making and 'creative experts' should be on their guard.

Read More

How would you prefer to be sliced and diced?

2008-05-14T11:22:00.012-05:00

Analytics has been pushed to the foreground of American minds by the 2008 election cycle. TV and news media provide seemingly endless hours of pundits and commentators discussing data and predictions. This analysis is based off of complex modeling as well as basic segmentation; political analytics brought us the terms "Soccer Moms" and "NASCAR dads" after all. While not the professional specialty area of most that read this blog, analytics is getting a lot of attention, and in many cases being applied in increasingly prominent ways.

I recently finished the book Microtrends by Political Analyst Svengali Mark Penn. The book offers a provocative analysis of “undiscovered,” yet potentially important populations in America, and promoted strategies on how to engage them and effect change. This idea of almost hyper segmentation has forced me to consider the ways in which I segment data and the resulting application.

I fundamentally believe that studying a heterogeneous group on a more micro level has great benefits, but I believe there can be costs as well. I hope others in our field give thoughtful consideration to the ways we “slice and dice” our data, as well as how “fine” we choose too cut.

You can segment individuals in a variety of ways, but many of these ways may not be useful for the questions you seek to answer. I may be identified as a “mid-twenties jazz music buff,” an “urban chess student and wine lover,” or as someone who “drives American” because I own a Pontiac. These are all accurate segments that connect me with others and offer some snapshots into my interests and purchasing preferences—but is it helpful to you? I feel there is a normal distribution related to the amount of segmentation conducted—a natural sweet spot, after which further division can create more problems than answers, or more incorrect conclusions than accurate ones.

Following the questions of “how do we cut” as well as “how deep” lies the next step: how should we use this information? Does segmentation serve as the sign post for a new fundraising strategy? Or does it simply signal more research? There are successful applications of both I believe, but it depends on the segmentation process and the questions you are trying to answer.

Read this article, consider analytic's emerging seat at the table in our world, and then ask yourself this question:

“How would I want to be identified (segmented) by organizations or causes I care about?”

What’s for Dinner? The pollsters want to know

If there’s butter and white wine in your refrigerator and Fig Newtons in the cookie jar, you’re likely to vote for Hillary Clinton. Prefer olive oil, Bear Naked granola and a latte to go? You probably like Barack Obama, too. And if you’re leaning toward John McCain, it’s all about kicking back with a bourbon and a stuffed crust pizza while you watch the Democrats fight it out next week in Pennsylvania.

Read More

Predictive versus Descriptive Modeling: some points to consider

2008-04-25T15:39:00.004-05:00

This is a fantastic article which I think very clearly describes the difference between descriptive and predictive analytics; I often find these terms blurred and blended very casually when discussing our work.

As the article suggests, understanding the difference along with the appropriate applications is fundamental to any good analytics shop. I personally believe the author is a little too critical on historically based projections and forecasts (basic descriptive analytics), but does raise some important limitations, including resource scarcity (the infamous pipeline), economic influences, and even potential competitors.

Woods also suggests productive applications of descriptive performance metrics such as “identifying broken systems” (perhaps a gift officer portfolio analysis). Many of us invest a great amount of effort in building complex and nuanced predictive models. I find it useful (and sometimes efficient) to conduct some descriptive models (average growth rate formulas, logarithmic projections) at the same time to get a wide analytics perspective. You may surprise yourself with what you might find, or discover something is missing…

Many organizations use historical analytics data as a basis for forecasting future growth, and establishing performance goals and budgets. This applicaton for analytics data can blur the distinction between predictive and descriptive data. Understanding this difference is critical to an effective analytics program. It generally falls to the analytics professional to ensure that the difference is clearly understood within the organization.

I'm going to start out with a couple of definitions. What do I mean when I say predictive versus descriptive modeling?

Read More

Exciting week - Josh's book is out and the Data Mining Summit

2008-03-31T11:17:00.005-05:00

Josh's book, Fundraising Analytics: Using Data To Guide Strategy, has been formally released. He has already received wonderful feedback from people who have purchased it and read it in one afternoon. It is currently sold-out on Amazon last I checked, but you can order a copy that will be delivered once it is in stock again or you can try ordering it directly from the publisher. Pick it up!

Also, I will be in Nashville at the end of the week, attending the inaugural APRA Summit on Data Mining and Modeling. I am very excited to meet people who also have a interest, or even passion, for the work we do. Feel free to say hi, and comments/questions/critiques of this blog are also welcome. Hope to meet you there!

-Alex

P.S. Josh will have a few copies of his book for sale at the APRA Summit.

Can we Build a Better Zip Model?

2008-03-31T11:14:00.003-05:00

Lately I have had a keen interest in demographic data and how it best fits with the tools we have and goals we seek in fundraising analytics. Certainly a plethora of affinity metrics and giving behavior makes our statistical mouths “water,” but demographic data still presents relevance and unique relationships (some good and some bad) when attempting to predict giving behavior.

I have recently posted articles suggesting another long look at demographic data (Why Demographic Data Just Won’t Die) and its benefits (Predictive Modeling the 2008 Elections…) in capturing difficult or complex decisions or choices. This article suggests some of the limitations of a zip model. While many of you may not use them regularly, I think zip-driven models may have utility for annual giving segmentation and mailings, and for institutions that rely heavily on a broad base of public and community support (urban public universities for example).

This article discusses some of the largest issues with zip-focused modeling, including aggregation, and the “self-fulfilling prophecy” phenomenon. It also offers some general but effective advice for anyone considering a zip model as an additional analytical tool.

How to Build a Better Zip Model

The May 2007 postal rate increase sent every direct retailer scrambling. It’s hard to argue the hike’s effectiveness as a catalyst for renewed analytical vigor.

Our clients have been analyzing everything from the impact of page count reductions and co-mailing programs to the most appropriate tools to optimize circulation. And for one, preliminary research indicated that a new zip model might be the right solution at the right time.

Zip modeling is not new. It remains a data-based tool that requires in-the-mail validation, but the postal rate increase was as good a time as any for many retailers to test it.

Predictive Modeling the 2008 Elections...

2008-03-19T16:32:00.000-05:00

In my content research for this blog, I look for specific articles relating to fundraising analytics, broader articles on analytics, or theory that provide either lessons or questions transferrable to our work, as well as other examples of creative minds using past behavior to predict future behavior. Without politicizing this blog, I want to share this article on Ken Strasma, a political analytics guru for a current presidential hopeful.

I was generally unaware of the depth and nuance of this pursuit of analytics. Particularly attractive I believe is the ability to model what are fundamentally just opinions (not financial transactions, such as charitable giving or consumer spending as opinions by proxy). I considered the lack of explicit numeric metrics to be a difficult obstacle to overcome, but Strasma and his colleagues have developed techniques to model not only complex preferences, but also predict what is essentially non-regular behavior (ie voting).

Strasma says:
“..there are a number of basic questions predictive analytics tries to answer for any campaign. These include how likely it is a voter is undecided, what issues undecided voters care about, how likely it is that a voter supports a certain candidate and how likely it is that an individual will contribute if asked.”

For our work, I considered this analysis to be similar to who has interest in giving, what causes do they support, how likely are they to support our organization, how much would they contribute to our organization, or more simply, who is a suspect, a prospect, what is the target, and what is the actual ask amount?

I hope this article enlightens your assumptions of predictive modeling, as it did for me.

Candidates Use Predictive Analytics To Seek Votes

As the primary race grinds on, the candidates are turning to predictive analytics tools to help find voters ready to support them.

A company called VisualCalc provides a free Web site that helps citizens analyze the presidential race through a series of dashboards that chart the status and trends of the primary election.

On the flip side, candidates in this year's historical race for the White House—for the first time a woman and a black man are vying for the Democratic Party nomination alongside a single presumptive Republican nominee—have similar tools to provide information that may help them attract those key undecided voters.

Read More

APRA Summit on Data Mining and Modeling

2008-02-25T16:09:00.006-06:00

I would be negligent in my duties as promoting data mining and predictive modeling in the area of fundraising if I didn't promote this upcoming conference. This is a fantastic new forum that will feature many of the brightest and most creative minds in our field, including my boss Josh Birkholz. The conference also coincides with the release of his new book.

I will be there as well, and hope to connect with those who read this blog for in-person discussions about where data mining and modeling is today in fundraising, and where future directions may take us.

Hope to see you there!

Summit on Prospect Data Mining and Modeling April 3 – 4, 2008

Don’t miss the first-ever APRA Summit on Prospect Data Mining and Modeling - the year's best opportunity to interact with prospect researchers and analysts engaged at the cutting edge of the advancement research field. This two-day symposium will be divided into two groups of sessions: a beginners/management track, and an intermediate/advanced track. The beginners/management track will provide a solid grounding in the goals of, methods for and approaches to data mining. The intermediate/advanced track will showcase new technologies and present case studies of effective applications of statistical methods to prospecting and prospect management.

Whether you’re a proficient data miner, or a researcher or manager contemplating a foray into data mining, this summit will provide you with fresh insights, understanding and tools to help you better understand your constituent base. If you are engaged in building your prospect pool, looking for ways to prioritize and bring focus to an unwieldy database, or seeking to discover diamonds hidden in the rough of a broad annual base of support, this event is for you.

Read More

Why Demographic Data Just Won't Die

2008-02-25T15:50:00.007-06:00

This is a really interesting perspective on what many, myself included, may now consider one of the relic's of predictive modeling: basic demographic data. This data is basic, sometimes clumsy--the data we used in college to learn the techniques of statistics, regression analysis, and econometrics. As analytics junkies today, we all strive to build models and tools to help us fit the contours of the populations we study and to levels much more precise than a zip code or an age group. In modeling, there is “power in numbers,” but there is also an aggregation danger at play when using broad metrics which capture individual behavior and preferences.

I have been posting for some time now on this blog about the frontiers of text-analytics and the raw potential inherent in such custom data mining approaches, that I fear I may have become too nano in my purview.

Behavioral modeling is definitely one of the sharper tools in our toolbox, but read this article and you may find yourself having a similar reaction that I did: reconsidering the benefits and devising new applications for using demographic data.

Demographics: The Targeting Construct That Wouldn't Die

Recently, our customers have communicated a message to us loud and clear. It is a message that may seem counterintuitive here in the 21st century, in the all-digital, micro-targeting, behavioral targeting, contextual targeting age.

Demographics, they tell us, are of paramount importance.

No, seriously. Demographics. Like age, gender, household income. I know; it’s as if I told you I was converting all my MP3s to 8-track, right?

Sentiment Analytics Opportunities

2008-02-14T11:13:00.006-06:00

A colleague provided a link to this article and I loved the title: Sentiment Analysis. This article is another perspective on a theme I have been posting on this forum for some time—moving fundraising analytics beyond simply “who” and “how much” (which are important questions) into more analysis of giving motivations, or "why.”

Presented here is a more in-depth consideration of some of the inherent challenges in using text analytics. The most basic challenge discussed is that opinions (say for example affinity) are harder to describe than facts (I gave $100). This article touches on some basic concepts that may “boost” fuzzy opinions and statements into data with high utility and function. Some of these strategies include:

*Classifying the source for more tailored analysis (gift officer notes, institutional survey, donor pledge card).
*If you have the appropriate software-lexical choice analysis.
*Bayesian methods to identify matching patterns.
*Hybrids of sentiment and account fielded (primarily numeric) analysis to improve sentiment “accuracy.”
*Making “two passes” at text—using automated tools/software, then a set of human eyes to verify results.

This article poses more questions than answers, but I believe with sentiment analytics relatively absence in the fundraising world, questions are the best place to start.

Sentiment Analysis: Opportunities and Challenges

Sentiment analysis is one of the most exciting applications of text analytics today. It may also be the most challenging. The steps involved in sentiment analysis are easy enough to grasp: use automated tools to discern, extract, and process attitudinal information found in text; apply to sources as varied as articles, blog postings, e-mail, call-center notes, and survey responses that capture facts and opinions. What do customers, reviewers, the business community – thought leaders and the public – think about your company and your company's products and services – and about your competitors? What can you learn that will help you improve design and quality, positioning, and messaging and also respond quickly to complaints?

Segmentation and Shakespeare

2008-01-18T15:04:00.000-06:00

Interesting news release out of Stratford England—The Royal Shakespeare Company has developed a successful partnership with an American analytics firm to successfully segment their database to identify and engage different ticketing behavior.

DonorCast has been moving into the ticketing side of predicting modeling and this technique looks promising given adequate data (isn’t that always the case though…)

The Two-Step Cluster feature in SPSS is very powerful—our practice has only just touched the surface of application possibilities. This technique can be used as a finishing “sorting” of records, or can do a sort based on key variables pre-modeling (it can handle both categorical and continuous variables).

I will find some more relevant articles to share about clustering and segmentation techniques in the next edition. In the meantime, play around with this SPSS feature and consider how it might be applied in your work.

Advanced Analytics Move Centre Stage at the Royal Shakespeare Company

SAN FRANCISCO & LONDON--(BUSINESS WIRE)--Analytics software from KXEN is helping boost audiences at Royal Shakespeare Company (RSC) productions in a pioneering arts marketing move. The initiative, an Accenture-led program to segment audiences, has seen a 50% rise in ticket buyers at RSC's Stratford-upon-Avon theatre, a more than 70% increase in regular attendees and significantly earlier sell-outs for London bookings.

Read More

Netflix Contest has Produced Prizes for the Analytics Community

2008-01-07T08:39:00.000-06:00

In June 2007, we posted about the "Netflix Prize" - a contest promoted by analytics savvy movie-rental-house Netflix.

The goal: improve the accuracy of the existing Cinewatch movie recommendation system.

The prize: $1 million

Fifteen months along, and no model has come forward meeting the victory threshold of 10% improvement on matching accuracy. Fortunately, for everyone that doesn't work at Netflix, this contest has produced something of value.

The discussions and attempts conceived from this contest have provided those interested in analytics new perspectives and questions to ponder as we seek to analytically quantify and predict preference and behavior.

This article discusses some of the most interesting insights thus far:

"Open Questions" (text mining) has emerged as a theme to "fine-tune" the specificity of predictive models. Allowing individuals an opportunity to express, instead of forcing them to conform entirely to a pre-defined format, is really emerging as a more nuanced and "high-touch" approach. As I have posted previously, there is software emerging that is making great strides towards allowing text mining to be a pragmatic tool. Discriminate choice models of "ultimate" giving destination preference (athletics, fine arts, brick and mortar) for example, could be greatly enhanced by appropriately applied text mining.

Another model suggested that information about tastes as related genre, language, actors, directors etc, was surprisingly powerless compared to the star ranking of the movie itself. Perhaps this suggests that second tier "affiliation" data (I love Tom Hanks, or in the fundraising field, I was a Sociology major) may be more ambiguous than standard industry assumptions. At minimum, this revelation suggests that more consideration should be given to the importance of the top preference metric (for movies its a star rating, for fundraising, it is giving to the institution).

The $1,000,000 Netflix Prize competition has produced interesting results, even if no winner, 15 months in. Some of those results are a bit surprising; others we should have expected but didn't anticipate. So while participants haven't yet bettered the accuracy of Netflix's Cinematch recommendation algorithm by 10%, the threshold to win the $1 million prize, we can still take away lessons about predictive-analytics fundamentals.

Read More

Web Analytics Primer from an "Evangelist"

2007-12-20T15:38:00.000-06:00

This is a great new article by Avinash Kaushik, the “Analytics Evangelist” for Google. I have posted a few other articles that touch on the topic of web analytics because I consider this a relatively untapped, but potentially rich source of information.

This is a very good primer for web analytics. Kaushik describes basic concepts in how website usage, or “visit” data, has utility. These concepts are fundamental, but certainly are still the most widely used in website analytics.

The applications for analysis for the six basic measures mentioned:

■ Visits
■ Page views
■ Pages/visit
■ Bounce rate
■ Average time on site
■ % new visits

These are universal in creating core metrics for a website—you need to have some place to start to know where you are going.

Basic ideas off the top of my head for these simple applications include:
1) Basic web stats for an online donations page—what is the “close” rate of those who visit?
2) Tracking sourcing from online pages—what are the most effective and least effective “links” sending people to your online donations page?
3) Identifying other interest areas through usage stats—are there other surprising sources on your site that have generated strong interest? Special events, news, messages? Possible affinities or, at the very least, interests may lie undetected.

And this is just a start. Obviously, as you layer and link pages, data, etc., the specificity of the analysis can increase sharply. This is a basic start.

Try it out. Show a colleague—see if they are interested…

New to Web Analytics? Confused about Web Analytics? Think it is too hard? Scared of tools and consultants?

This post is for you, its goal: Web Analytics Demystified! Yeah!

Web Analytics is complex. That is what it is. Complex.

Get the nuance? Complex. Mysterious. Inviting. Come in. Sit down. See what’s there. No free rides. You’ll do your part, your efforts will have a rich payback.

Complex holds the promise that you’ll get it. Nay, you can get it. Come in, welcome.
Start with this post.

DELTA Force

2007-11-29T08:11:00.001-06:00

Perhaps a misleading, if not corny title. The "spirit" however is relevant to this article.

Thomas Davenport, a respected leader in the field of predictive analytics, spoke at the SPSS Directions conference last month in Orlando, Florida.

DELTA is an acronym Davenport created to capture the life cycle, as well as the environment necessary, for successful predictive analytics ventures. If you have read his book "Competing on Analytics: The New Science of Winning," the concepts will be familiar. If you have not picked up a copy, I strongly suggest you purchase it.

Either way this review of his keynote is informative.

ORLANDO, FLA. -- Walking on stage here yesterday at SPSS's Directions 2007 North American Conference, author Tom Davenport sported a Boston Red Sox cap and used the 2007 World Series Champions as an example of how predictive analytics can give organizations a competitive advantage.

"The Oakland A's had analytics and no money," Davenport said, referring to A's general manager Billy Beane, who introduced the power of mathematics and statistical analysis to the day-to-day operations of running a major league baseball team. "The Yankees had money and no analytics," he added. "The Red Sox have both money and analytics," which he believed contributed to the team's second championship in four years. Not without taking a few additional jabs at Yankees fans in the audience, Davenport, as part of his presentation, "Competing on Analytics: How Fact-Based Decisions and Business Intelligence Drive Performance," proceeded to emphasize the importance of predictive analytics. His formula, he said, could be broken down using the acronym DELTA:

Read More

Online Fundraising - How is Behavior Different?

2007-11-29T07:41:00.000-06:00

With its relative ease of operation, low overhead costs, and the increasing role of the Internet replacing previously in-person transactions in our daily lives, online fundraising is now a major player in fundraising. While working on a recent project regarding various giving channels I asked myself this question:

How is online giving behavior different from offline?

While this might not satisfy any requirements as "breaking news" (it is nearly a year old), I found this study regarding online fundraising behavior incredibly informative.

Some interesting findings:

The Internet can serve as an effective acquisition source
Online donors tend to be younger and wealthier than offline donors
Online donors have lower renewal rates than offline donors
Multiple channel donors (online and phone or mail or personal solicitation) have higher revenue and retention rates

This article does a fantastic job summarizing the study, and I suggest you read it.

Online Fundraising on the Rise - Target Analysis Group and Donordigital Report finds

While the Internet, broadband networks and email have grown to be the new fundraising tools for non-profits over the past several years, their potential has not been reached - in terms of the amount of money raised and the number of organizations fundraising online.

SPSS Directions Panelist Notes

2007-11-29T07:26:00.000-06:00

The SPSS Directions conference, held in Orlando last month-featured predictive analytics industry leaders, including Bentz Whaley Flessner's Josh Birkholz and the "Grandfather" of predictive analytics, Thomas Davenport.

This article reviews a keynote panelist discussion, revolved not around statistical techniques, but the presence of predictive modeling in the business industry today.

Understanding how predictive modeling is viewed within your organization, and developing ways for further integration of your work were central themes; from "simple" language to helping your organization where your predictive resources can be applied to where there might be limitations that are not obvious to others.

Finally:

"You can never have enough data" - Thomas Davenport

ORLANDO, FLA. -- As part of SPSS's Directions North American Conference here Monday, all of the keynote panelists portrayed themselves as the visionaries of their respective companies. Each speaker strongly described predictive analytics as a means to elevate a company above its competition -- and, ultimately, to better serve its customers -- regardless of any corporate obstacles.

"If you know this is right, you need to just take [other executives'] criticism. Don't let them win the battle!" said Mike Hayes, senior vice president of The Bon-Ton Stores, a Pennsylvania-based operator of over 200 department stores.

Read More

(Re)-emerging strategies for the “narrative” or “unstructured data” problem.

2007-11-12T15:10:00.000-06:00

This article discusses a re-emerging field in predictive analytics called Text Analytics. I say re-emerging, because as the author points out, narrative analysis was a cornerstone of the earliest business intelligence strategies. Today this concept may have utility especially when combined with segmentation or donor-targeting strategies. From prospect management report sheets, phone-a-thon caller logs, to the infamous “other” box on a simple survey question, Text Analytics can provide opportunities for more nuanced insight into the “narrative” data we do have—as well as applications to quantitative models we construct.

One of the fundamental problems of using mathematics to analyze human behavior is the unstructured, or as I like to call it, “narrative” data problem. The amount of purely numerical or quantifiable information available to those in the predictive analytics field is limited—and what this quantifiable information available can tell you is variable as well. I consider non-profit or fundraising analytics to be more opaque than for-profit sectors in respect to this reality. Individuals, on a basic level, need to purchase goods and services. Therefore intent and preference are more transparent. In for-profits, purchasing a product can imply a variety of affinity relationships; this product is a necessity, I prefer this product to other similar products, etc.

Philanthropic giving, monetary or in-kind, is less clear in respect to quantifiable variables producing specific affinity. Attitudes towards institutions or missions may often be more personal than the type of soap you buy, so a donation may imply high affinity. The source of affinity however, can differ greatly: I am an alumnus, my child was a patient, the institution is important to the community, I like the sports teams, etc. Also the absence of immediately available options (there are no supermarkets to choose between charitable organizations) makes comparisons difficult as well. Giving data, capacity rating, alumni classification are all quantifiable values, but some more “narrative” fields like the basic question, “why is giving to us important to you” are more complex.

While the technology for Text Analysis may be more complex and costly than many organizations care to absorb, I believe this represents a very exciting frontier; making predictive modeling more accurate, dynamic, and relevant.

Text analytics is a new IT discipline that has already proved itself in applications ranging from pharmaceutical drug discovery to counter-terrorism to survey analysis, in science, government, and industry. It is poised to break out into the broader analytics market, in workbench form, integrated with business intelligence solutions, embedded in line-of-business applications, and enabling semantic search.

Text analytics is an answer to the “unstructured data” problem, which is best expressed by the truism that eighty percent of enterprise information originates and is locked in “unstructured” form. That problem has been recognized for decades. In fact, the first definition of business intelligence (BI) itself, in an October 1958 IBM Journal article by H.P. Luhn, A Business Intelligence System, describes a system that will:

“…utilize data-processing machines for auto-abstracting and auto-encoding of documents and for creating interest profiles for each of the ‘action points’ in an organization. Both incoming and internally generated documents are automatically abstracted, characterized by a word pattern, and sent automatically to appropriate action points.”

So we see that the earliest BI focus was on text – on extraction, categorization, and classification rather than on numerical data!

Read More

Lets not forget about the Annual Fund; Behavioral Targeting

2007-10-26T14:56:00.000-05:00

This article comes from a for-profit sector perspective and discusses an optimization technique using web data, commonly referred to as "behavioral targeting." Many of the articles and techniques shared here relate to predictive modeling, primarily for major giving—what about other giving populations who may yield smaller dollar amounts, but have more consistent patterns of philanthropy? Obviously the high-reward potential of accurately identifying transformative or major gift prospects is very attractive. There are other opportunities however, in which to apply predictive modeling techniques to support increasing the effectiveness of your giving programs at all gift levels.

The requirements (significant longitudinal data) and benefits (modeling of consistently stated preferences) of behavioral targeting make it an interesting technique when applied to the examination of annual giving behavior.

For example, if your institution had a well-developed online annual giving program, elements and principles of behavioral targeting could be applied. Inserting one simple, but well-designed affinity question into the process of submitting an online annual giving donation could produce some informative trends. From these trends, the annual giving program could be more specialized in targeting and messaging, as they seek to engage new constituents or increasing giving levels of current annual giving donors by identifying effective priorities and factors for giving.

The other benefit about these strategies is that they are relatively simple, when compared to complex major gift models with cluster analysis, etc.; the time invested may just be appropriately proportional to increase in dollars from focusing on populations of "base givers."

IN MY PAST FEW COLUMNS, I have set out to clarify optimization, a term that is often bandied about and regularly misunderstood.

I first covered testing, the most frequently used method of improving consumer response, which includes A/B testing and multivariate testing. With the targeting article, I covered how systems based on rules can be used to create more relevant experiences with better outcomes.

The third type is perhaps the most seductive -- and misunderstood -- form of optimization, behavioral targeting. (The fourth, social optimization, I will explain in the near future.)
What Is Behavioral Targeting?

The holy grail of direct marketing has been a system that detects consumer behavior and changes offers. The first incarnation of this approach was called data mining, and was focused on using data to drive strategic planning. There is an apocryphal story about Wal-Mart: "By scanning each sale into a data warehouse, grocery stores have determined that men in their 20s who purchase beer on Fridays after work are also likely to buy a pack of diapers. Thus, a display of Pampers or another brand might be set up in the beer aisle, or merchants will put one (but not both) of the products on sale on Friday evenings."

Read more

Predictive Analytics frontiers: Web Analytics

2007-09-27T10:30:00.000-05:00

Predictive Modeling and Analytics is a new and exciting tool in the development and advancement services world. Many organizations are now not only integrating these powerful tools into their current work, but also seek new opportunities to improve their organizations capacity to understand relationships and model behavior. By implementing new strategies for data collection and information, organizations recognize the gains which could increase Predictive Modeling’s utility.

The newest frontier in Predictive Analytics is Web Analytics; information gathered from website traffic is considered by many a fertile and relatively un-tapped resource of behavioral data. Applications for Universities, Colleges, Hospitals, and other non-profits vary from gauging the effectiveness of a campaign message, to the success of online giving solicitation versus more traditional methods.

This is a very informative article from Avinash Kaushik, the Analytics Evangelist at Google.com. Kaushik considers the opportunities in Web Analytics, in addition to current struggles in its application, along with ideas on how to improve Web Analytics data in the future.

Data Mining and Predictive Analytics on Web Data Works? Nyet!

Strong Russian word: Nyet (No). By the end of this post I hope you’ll agree. Worst case you’ll have food for thought.

This in-depth post covers a complex topic that might not apply to everyone, but it covers an area where companies have struggled to try to show return on the investments made in skills, technology and time. The post promises clarity and guidance that hopefully will result in you saving tons of aggravation and yes even a nice chunk of change.

Data Mining and Predictive Analytics have promised a the earth, the moon and the sun for sometime now, in all channels we do business in. My personal point of view is that on the web they fall far short of even the most pessimistic promises. For now…

Read more

Data Mining: Three Steps To Mining Unstructured Data

2007-08-09T11:02:00.000-05:00

This article presents some informative perspectives on effective ways to capture and integrate non-static, or "unstructured" data. The term unstructured data is often applied to information regarding an individuals preferences or tastes. This information is viewed as more susceptible to variation and more difficult to predict, yet powerful as a predictor.

One field of non-static data in prospect research may be rating an individuals willingness to give during a campaign. This rating may be tied to a series of factors which often have uncorrelated relationships to each other (affinity for the institution and stock market performance for example) and can change independently. However if they are appropriately aggregated and correctly understood, can be dynamic informers towards predicting the likelihood of giving at all levels.

In our journey of discovery, we have seen one mistake made repeatedly. We have seen static business models and static data models try to be used to model inherently dynamic business processes, particularly at the point of interaction. For example, virtually every customer relationship management system we have come across has a manual classification scheme (or taxonomy) that is meant to be used by the service agent to classify the nature of the customer interaction. This approach has two major flaws.

First, as soon as the classification scheme is published, it is out of date, because interactions with your customers are unpredictable and continually changing. Second, even if the classification scheme was representative of your customer interactions, it is unreasonable to expect any number of service agents to classify their interactions with their customers in a consistent way and with high quality. This very often makes such classification data completely useless, or, more dangerously, misleading. This issue is true throughout the business ecosystem where unstructured information exists.

Read More