Data should be the bedrock of any political campaign. But, it's not. I'm not talking about just polling data. Let's face it: pollsters don't want to tell you their craft is limited, finicky, and generally more of an art than a science.
What I'm talking about is customer relationship management style (CRM) data management. Replace the "client" in CRM with "Voter" or "Donor", and you get where I'm going. Minimally, all user online activity, voterfile data, donor history, should have an integrated back end. Then, you should hire a couple of smart people who understand data tools like Crystal Reports and let them tell you what's going on. You might just be surprised.
Look - there's an oft quoted statistic in the CRM industry, that something like 70% of CRM implementations fail. Political campaigns aren't ready to go for full CRM--process is the cornerstone of CRM, and campaigns aren't amenable to heavy process documentation. There are also enormous data privacy/security issues to be grappled with.
Andrew Raseij Rasiej said something interesting to me yesterday,
"Harish, it won't be until your generation turns 40 and runs for political office will any of this (heavy CRM, data mining.) be used by political candidates."
Andrew's probably right. Regardless, the political candidates and organizations that figure this data thing out today will have a tremendous political advantage. Don't get me wrong--I think there are pretty smart people that have figured out a lot of this. In the mean time, the rest of us are behind the curve.
With all of the computing power, bandwidth, and understanding of data mining techniques across client relationship management platforms, winning candidates will get with the program and leverage this technology, at least minimally.
Legacy Comments
There are several reasons why large, enterprise-wide CRM implementations frequently don't work well. The biggest is that they are large, enterprise-wide software implementations.
Here are three ways to almost guarantee that a CRM system will be successful:
The other benefit of gathering this data in one database is that the people running Crystal Reports have much more high-quality data to use in reporting and forecasting.
Having a successful CRM system doesn't take a rocket scientist, but it does take a desire to help the team listen, observe, and learn about constituents' needs and concerns.
Hi Cliff
Insightful comment. I think your last paragraph, having a successful CRM system (snip) does takea desire to help the team listed, observe, and learn..." is critical. We need to get more end-users involved in the process, and we need to be able to adapt our own thinking as we develop the best tools and solutions possible.
HRR
Hi HRR!
I've been told "X won't happen until my generation turns 40 and Y" for many situations:
(X,Y)
("IT taken seriously", "become CEOs")
("Products that use the internet", "become CEOs")
("Technology education in schools won't be effective", "become principals")
("Fundamental changes in higher-ed", "become Deans")
("Databases won't be designed in proper 3rd normal form", "run projects")
etc.
etc.
The point that so many CRM deployments fail is getting a lot of study right now. Fundamentally CRM deployments fail because people at the top are shielded from finding out what happens in the trenches. The AT&T Wireless failure was due to upper management making dumb technology decisions (we saved money last year by not upgrading, let's not upgrade again this year!) followed by an environment that discourages bad news.
There was a great article recently (sorry, I can't find the link) about the VP of Customer Service at AOL being shocked when she decided to sit with the people that answer the phones. 80% of the calls they get are people asking why AOL sent them this CD that doesn't play any music when they put it in their stereo. Gosh, could their low new-subscriber rate be related to the fact that most CDs get sent to people without computers?
When working on a local campaign, the script that the phone callers were using was extremely ineffective, but the volunteers stuck to it because they had complained but the campaign manager thought it was great and wasn't willing to try the script for herself.
(On the Dean campaign, the "dial for Dean" system had no quality control. Oops.)
Another big problem is that CRM systems just aren't that flexible. You design what you think will work, and its a crapshoot if you are right or wrong.
I don't know a lot about CRMs, but I do know a lot about orchestrating good rollouts. Starting with a small group, growing, growing, adding features, growing, growing, etc. It can take years, and a campaign doesn't have that kind of time. Oh wait, maybe that's why the Republicans started designing theirs 4 years ago.
Tom
Great thread Harish.
To my mind the problem begins with the traditional firewalls of power and information in Democratic campaigns. Field, Political, Communications, Policy, Correspondence, and Fundraising don't talk to one another much, and cloister information in order to accrete and retain power within the campaign. Furthermore, power and budget flow sequentially from fundraising at the beginning of a campaign to policy to political to field etc. as we draw closer to election day. Traditional commercial technology vendors to Democratic campaigns have typically developed, understandably, focused tools and services geared towards meeting the needs of and winning business from one of these departments.
Ushering in a new era of wholistic campaign technology would require, among other things:
-development of a great open-source highly scalable CRM system at the core
-integrated and fully productized and supported toolsets that leverage this CRM system and that meet the traditional needs (e.g.: voter file systems, donor tracking, web CMS, etc.) of each Department at least as well as today's stand-alone tools do
-vendors that are willing and able to provide and support these integrated systems at a price point that small campaigns as well as large can afford
-lowering the barriers (legal and cultural and technical) between political campaigns non-campaign grassroots advocacy organizations so that constituents can effectively participate in and be leveraged by both
-case studies that visibly demonstrate that sharing information within a campaign yields success far more than cloistering it
-a means to fund the development of all this that does not depend on contracts from campaigns (because the internal advocates for and budget for such product will not be there before the product is there)
-visionary and powerful campaign managers and candidates who will hire staff that are willing to take on the risk of doing things in a very new way, and a track record of those campaigns winning in order to prove it to those less visionary and more risk-adverse.
This is a tall order-- but there are reasons for hope:
1) My sense is that NGP is going to be a powerful advocate here because they get the opportunity available in providing wholistic solutions, and they are already in the door in most Democratic campaigns with the Department that gets to call all the shots in the early days of a campaign-- Fundraising.
2) The currently-happening integration of CivicSpace with Advokit with a new open-source CRM back-end will bring a lot of these capabilities together, and low-cost open source committed vendors/consultants are standing by, ready to implement and support and to do the heavy lifting on hosting potentially massive databases of voters and street networks and the like... The biggest questions on this approach are how well it will be productized, how effectively it can be marketed to traditional campaigns, and whether campaigns will budget sufficiently to support viable business models for these vendors.
3) Some traditional campaign technology vendors have committed themselves to handling the heavy lifting of hosting massive data sets while newly opening up their systems to interoperate with 3rd party tools and services through coherent APIs.
4) The one size fits all approach of the commercial ASPs and the high-cost proprietary custom solution approach of others, while they do a good job with their existing markets, is not gaining the kind of traction one might expect in an era of exploding online donation processing, while campaigns and grassroots organizations are signing up for open-source solutions in droves, even before they're fully productized and complete.
Cheers,
Andrew Hoppin
Disclaimer: I work with CivicSpace and with Trellon
TOM!
YOU'RE ALIVE! CALL ME! EMAIL ME! SOMETHING!
You're quite right. People in power like to shield themselves from the Truth. I remember you trying to help the volunteers who would answer email on the Dean campaign (our voter [client] service people, in effect), who used to sit right next to our office: 6 volunteers answering anywhere between 4-8,000 inbound email questions (not spam!) per day. On a day where the Governor would get on TV, 20-30,000 inbound email inquiries. On a good day, I think they would get to respond to about 20% of those emails.
WTF?
But the Dean campaign was probably better at this type of service than almost any other campaign out there. And it's not like other political campaigns are any better. They're not listening to the data that is all around them. They're not even collecting the data. They're just really looking for the right message from their pollsters, rather than paying attention to the people who are screaming at them to do something. How many hundreds of phone calls would we get in a day? What about all of that data? Not even a minimal logging system was in place, from what I recall.
Seth Godin has a very accurate summary:
http://sethgodin.typepad.com/seths_blog/2005/01/and_its_even_a_.html
Harish,
CRM offers a flawed model for understanding what voters are saying. While there certainly are operational problems with CRM systems, the larger problem behind CRM is that it is constructed of data points indicating past purchase behavior. It is in a sense a rear-view mirror. It tells you that people who bought a certain toaster in the past are likely to buy a certain coffee-maker in the future or people who bought Porches in the past and say they are Republican means that Porsche owners are more likely to vote for Republicans in the future. If this were true, you wouldn't be quoting statistics that say 70% of all CRM systems fail. Insights on past purchase behavior are not true indicators of future behavior. CRM and polling data largely result from numbers or information attained through closed-ended and internally written formats.
Responses to polls that say a certain percentage of voters somewhat agree with a candidate's statements tell you nothing (often determined on a 1 to 6 scale) because they don't tell you "why" they agree nor the difference between one person's 3 rating and another's 4 on the agreement scale. Further, the range of answers creates bias because it is internally determined by the candidate or polling company. A candidate can only understand how to motivate voters if he/she knows why they believe what they do. This lends itself largely to using open-ended questions to query voters on past and future behavior, what would cause them to change and why they would do so.
The problem is that coding perceptions, ideas and motivations from thousands of open-ended, conversational responses, aggregating it into usable and understandable data is an onerous task. Candidates resort to focus groups and their many flaws. Candidates or interest groups will not get it right until they begin to utilize systems that use open-ended questions. The idea is not to think about how CRM does it in the commercial world because they are doing it wrong.
Using that Porsche story again that supposedly turned Karl Rove into a genius: If, for example, we were to learn from Porsche owners what messages they would like to hear from a candidate to capture their vote and why those would be important to them, we'd have something to work with. Further, with enough open-ended data, we could segment these people by their unmet needs and understand the core and how messaging can motivate people on the fringes. Then you've got something of value for turning the political process around. The right messages are not determined internally but by listening to what people are saying - and it is manageable to listen and respond.
Jon Stamell
Chairman, Communitas Online
We had access to the tools to start to do some of this while I was working for ACT Ohio. We had a bunch of consumer marketing data from Axciom that we could in theory have used to target mail and calls to very specific groups of people and demographics - based on their past purchasing behavior, income, ethnicity, organization membership, etc.
But we didn't really use the tools, and i'm not sure why -- even with someone from Axciom in house helping us out. We ended up using Axciom to help clean up our voterfile (their data was better than that from the secretary of state), but we did not do the finegrained message targeting that we might have.
Part of it might be mindset, as Harish indicates. The people making decisions about how to spend money are not yet used to thinking about how to do niche message targetting. Telling everyone exactly what they want to hear is a powerful concept, if someone can figure out how to do it effectively.
We did not however, have CRM style tools. We were engaging in an enormous number of voter contacts through paid calls, mailings, and our paid and volunteer canvasses. These were all recorded in the voterfile -- but if someone were to call our offices with a question, or comment - that was not recorded in the voterfile. In general this data was only really used to send out additional calls or mailing to people who had identified themselves as Kerry supporters, or undecided. The idea of managing the relationship between ACT and a voter over a series of interactions really was not implemented.
~Nathan
I just remembered a time when we did use a CRM style interaction.
During the week or two before the election, ACT Ohio ran a 800 line for voter questions -- how to get to the polls, voter fraud, registration problems, etc.
I adapted a trouble ticketing system written to manage a computer network into a system we used to track the calls, and the problems that were reported to us. The phone bank staff entered calls as they were recieved, and then marked them as closed, as problems were resolved. Serious problems were flagged in the database, and were reviewed by our lawyer.
We called voters back to make sure that problems had either been resolved, by us or by the voter before closing the ticket. This really was not CRM -- since it was not integrated with the Voterfile, or allow us to keep a log of our other interactions.
The basic problem is that CRM is hard, it requires a lot of interaction between different systems (the proprietary software used by predictive dialers for example). Campaigns are so transitory that is difficult to justify the time and expense necessary. Open standards for data interoperability and open source software to do the heavy lifting might solve those problems -- but they still require expertise to implement well, something most campaigns at the state and local level just don't have.
~Nathan
http://www.acxiom.com/, actually.
I've been talking a bit about the need for a good, inexpensive web platform for political campaigns. The way I see it, there are two stumbling blocks for widespread adoption of this idea: the first is a
Hey Harish, nice discussion.
I agree with Jon Stamell, but I think his criticisms can be addressed by breaking down the problem. CRM is appropriate for donors, volunteers and activists but probably not the right approach for voters generally (prospect list, GOTV, persuasion). I think of donors and activists as the house file, in direct marketing terms, the people the campaign or organization has a relationship with, and CRM makes sense there. The whole universe of voters, or the voting age population for that matter, is more like a prospect file, and a target for GOTV and persuasion.
The US population is about 295 million, roughly 220 million are of voting age and about 170 million are registered to vote. A national campaign has interactions with only about 5% of the voting population (interactions being an actual two-way communication). Handling prospecting, GOTV and persuasion is a problem of an entirely different scale from interacting with donors and activists.