Modeling — Taking It Straight From the Experts (Part 2)

Welcome to Part 2 of yesterday's post. (Click here to jump to Part 1.)

Modeling: The more I thought about it, the more I wanted to know what the experts were thinking — so I decided to just ask! For the most part, what is in this article are direct comments/quotes from the folks that I believe are the brightest and the best in our industry around this topic. These are the folks who participated

• Richard Becker, president, Target Analytics
• Jeff Birkner, director of marketing intelligence and strategy, Eleventy Marketing Group
• Caity Craver, CEO, DonorTrends
• Jim Emlet, principal and co-owner, Integral
• John Ernst, chief strategy officer, nonprofit, Paradysz
• Brent Eskew, executive vice president, Wiland Direct
• Yannis Kotziagkiaouridis, vice president, market analytics, Merkle
• Gretchen Littlefield, president, Infogroup Nonprofit
• Chuck Longfield, chief scientist, Blackbaud
• Curt Swindoll, executive vice president, strategy, Pursuant
• Andy Wilder, vice president, analytics consulting group, Epsilon

(You can read the first part of this post here.)

3. How is modeling able to inform the movement to true integrated, multichannel marketing?

Brent Eskew: Modeling is channel agnostic. The right prospect (chosen from a sea of wrong ones) is the right prospect. The solution is to find that right prospect in the sea of eyeballs viewing other channels and integrated across those channels (integrated mail, online display, e-mail and telephone, for example). The records contained in a cooperative database are all multibuyers and have dozens of data points on each one — including the channels through which each individual is most responsive. Proper targeting derived through modeling and based on actual giving behavior across channels helps to realize the efficiencies.

Curt Swindoll: Modeling is used to look for leading indicators and patterns of success, and we've successfully used it to bridge channels. We score and prioritize donors based on multiple factors of engagement and capacity, which then informs our approach to helping our clients leverage the strengths of different channels to engage donors on an even deeper level. Data helps us know which donors are interested, what they are interested in, when they were interested and even how interested they are in what an organization is doing. All those data points are valuable to crafting a multichannel strategy that works.

Richard Becker/Chuck Longfield: Yes, it informs multichannel marketing — by using the more complex data set of integrated, multichannel data! And there's also likely a reorganization/change management piece where silos get broken down as models decide that the next best solicitation or communication comes from another department. The key for nonprofits is to understand the impact of various marketing channels on one another, rather than looking at them uniquely. For example, too often "digital" is evaluated separately from "mail," with decision-making and programs budgets residing in nonprofit organization silos. The ability for data and analytic providers to inform nonprofits with attribution, or the impact of multiple donor touches across channels prior to the final point of engagement/donation, is an area of opportunity that has yet to be fully exploited. For many nonprofits, the value of display advertising, mobile advertising, e-mail, social and SEO have yet to be fully understood or rationalized.

Andy Wilder: This is an area where commercial clients are much more evolved than fundraisers. Most commercial clients have a campaign management solution that leverages predictive analytics and business rules to support an automated, centralized and multichannel campaign management program in real time. The process begins with data integration to ensure that you have a single view of the constituent across the organization (e.g., direct marketing, special events, e-channel, volunteer events). This enables the organization to optimize the donor experience within and across each of the various channels/touchpoints by delivering personalized and relevant content. Models absolutely drive better multichannel campaigns by creating an automated approach to developing, selecting, prioritizing and distributing marketing content based on historical data and interactions.

Caity Craver: Models can be used to inform channel integration in two ways: predicting donor channel preference (what channel or combination of channels a donors prefers: mail, phone, online, face-to-face, etc.) and predicting value (determine how much can be spent on certain donors). Donors that are predicted to generate a small amount of revenue would receive a contact strategy that has a few mail pieces and more low-cost e-communications. Donors that are predicted to generate higher value would receive a contact strategy that has higher-cost channels (mail and phone) with a mix of e-communications to reinforce messaging.

Yannis Kotziagkiaouridis: The ability to connect data across "silos," and especially the offline and digital connections at the constituent level, are the key enablers of true integrated, multichannel marketing. Traditional modeling techniques have focused on optimizing targeting in a silo; for example, direct mail or e-mail. The question has transformed from one that seeks to understand "who is the best person to talk to with direct mail?" to "what is the optimized mix and sequence of personalized experiences we need to deliver to achieve the marketing and business outcomes desired?"

John Ernst: Models, ideally, are able to put the channels on one platform and level the playing field, so to speak. There's no channel politics or budget silos to a math equation. I believe models will inform the movement to multichannel marketing by playing the role of translator, enabling the channels, and their respective owners at each organization, to speak the same language. Models only see data. If the data says donors should be communicated to via e-mail vs. direct mail, or some shared combination of both, then there should be little room for debate.

Gretchen Littlefield: From an acquisition perspective, modeling offers the best ability to match the prospect to the offer and channel of his/her choosing, and deliver the offer when he/she is most likely to respond. The fact that marketers now have the ability to deliver targeted ads online in concert with mail, all driven by models powered with transactional data, is thrilling. The amount of online data is mind-blowing; it seems everyone is talking about "Big Data." With all of that data, how does a marketer find actionable insights? Modeling. It's quite simply the ability to be relevant.

Jim Emlet: If you ask the right question and have the right data, modeling can inform nearly all aspects of direct-marketing decision-making.

Jeff Birkner: Successful modeling must look at all transactions across all channels. However, understanding why people give is a first critical step in the development of a modeling methodology and required to account for the underlying psychographic nature of charitable behavior.

4. What is the future of modeling for fundraising — what's on the horizon?

Eskew: Not to be indelicate, but nonprofit fundraisers are likely years behind other commercial interests in the use of sophisticated modeling. The economic pressures that commercial direct marketers faced 10 to 20 years ago are currently evident in the nonprofit space. It is highly likely that their need to raise funds more effectively and efficiently will lead them to expanded modeling generally and cooperative database models specifically. Not only will models improve results through better selection, they can also be used to identify the best prospects from within the current house file for reactivation, monthly giving, major gifts and planned gifts.

Swindoll: Too many nonprofits aren't even sure what the modeling questions are. You have to figure that out before new capabilities will be of much use. For those who do understand the questions, two areas deserve attention: advanced analytics and better visualizations. Leading nonprofits recognize the gold mine that is available to them in their donor systems if they take the time to capture every piece of information they can — from social-media involvement, to behavioral engagement, as well as traditional RFM and response metrics. They then need tools to analyze that data and look for correlations. We also need new ways to visualize information — beyond bar graphs and pie charts.

Becker/Longfield: The industry will continue to evolve its thinking about models. First with models being run externally and the results uploaded into systems, and soon, with models being directly incorporated into these systems. Vendors and agencies will develop proprietary models that can be plugged in and run (plug-and-play models). The future of modeling for fundraising will consist of improvements to both old and new. Mail-based strategies will continue to be enhanced with attribution-related insights, donor LTV calculations and campaign analytics. Newer channels, and their relationship with one another, will benefit from improved ROI measurement as a result of new transactional and behavioral data being harnessed in a breed of analytics. Ultimately, the opportunity for data and analytic providers serving nonprofits is to evolve from "descriptive" and "predictive" analytics to "prescriptive" analytics, whereby specific action recommendations are delivered that influence program strategy. This will necessarily involve the intersection of software and data, combining rules-based decision engines with data and analytics.

Wilder: To get a sense of where the nonprofit industry will be tomorrow, I think you need to look at where the commercial world is today, and that's unlocking the secrets of "Big Data." Through advanced analytics, organizations are applying social network/sentiment analysis and text mining to reveal hidden patterns and correlations in unstructured Big Data residing outside the consumer database. Combining these insights with traditional data assets is helping marketers identify key influencers for anything from new products to new drugs. The result has been an amplification of targeting as each marketing touch has the potential to reach well beyond an individual consumer. As nonprofits search for innovative ways to attract younger constituents with significant upside potential, Big Data may hold the answer.

Craver: Much like financial institutions started qualifying customers based off a credit score created by Fair Issac Corp. called the FICO score in the late '70s, by 2018 NGOs big and small will use scores to identify, prioritize and cultivate their donors. Scores will be real-time and accessible directly from CRM systems. Fundraisers will come to rely on scoring systems to drive their contact strategies.

Kotziagkiaouridis: Constituents are, above all, consumers. And there is a multibillion-dollar commercial industry that is aggressively spending to better understand their needs and anticipate their all-important "next move." This will have a profound impact on the nonprofit world. Nonprofit organizations that manage to deliver the Zappos or Amazon experience to their donors have a better chance of succeeding in this new world. This is because they better understand who their constituents are; predict what drives emotional connections; and can communicate with relevance and consistency across media, channels and devices to deliver their mission statements.

Ernst: I think the real future for modeling is clearly the ability to access and leverage broader consumer marketing data to model response for an individual organization. We're moving away from the days where an organization's own data is the single best asset to predict future behavior. A potential donor is influenced by marketing efforts from all sectors and in all channels. In order to predict a donor's potential engagement or even their ultimate response, we're going to have to know more about their marketing activity across the spectrum and leverage that information to speak to them in the manner (and channels) they prefer. There's a real "data collection" war happening. The future of modeling is figuring out what data is noise and what is actually relevant predictive information ... and for nonprofits, doing it in a fiscally responsible way.

Littlefield: At a time when so many nonprofit boards of directors are asking (actually demanding) greater efficiency within fundraising, it's likely that modeling will take center stage. It would seem that a more effective utilization of "speaking" to prospects and donors when/where/how they want (revealed through analytics) is likely — and necessary. It's also difficult not to foresee significant growth in the use of prospect databases — the economics of traditional acquisition will force that. "Smart multichannel" marketing decisions will surely be directed by modeling — the dynamic of the 21st century, post-boomer donor is misunderstood by traditional marketing. Also, if anything can solve the mystery of "how to get younger," it is analytics. Conversely, modeling/analytics might prove that this goal is actually a fool's crusade and not worth marketing investment.

Emlet: As it so often does, the future of modeling for fundraising comes back to the data: what data is captured, about whom and how is it stored. So the more data available, the more modeling can be employed.

Birkner: In the future, more value and emphasis will be placed on looking at all of a donor's transactions and interactions within an organization across all channels. In addition to that as we look into the future, the "next generation" of modeling will evolve around causal modeling and where a donor is at in their own personal life cycle and how that affects why (or why not) they become a donor. Through modeling, the industry should be able to predict when someone is settling down and their interests are shifting due to where there are at in terms of discretionary income, owning home, having a family, etc. This type of modeling will put an emphasis on why the person is giving or becoming a donor to the organization.

Summing it all up
I know there are multiple opinions above — even differences in where the industry is overall with the advancement of modeling — but one thing is for sure: This is an area where nonprofits must look. For a nonprofit still using the traditional RFM methods, rest assured, you are not doing something bad but you can be doing so much more. For the nonprofit with lots of data — ask yourself if you are using it for maximum benefit. To have this much input from people across the industry who absolutely know what they are talking about is a clear indication that we have only seen the tip of this iceberg ...