As with any endeavor, the public is free to offer a critique of the endeavor ... and in our modern world, if you create something, somebody will critique it. Let's address some of the complaints you'll hear about Digital Profiles.
"If you're going to go to all of this trouble, the least you can do is give us the source code, so that we can replicate the methodology for ourselves and see if it works. Why won't you let us do this for free? What good is this if I can't run the methodology for free?" Here's the funny thing, folks. I've basically given you every single step that you need, so you can truly do this for free. I showed you the variables I collect. I showed you how to standardize the variables. I told you that I use a Factor Analysis / Principal Components Analysis, I told you I use the coefficients from the Rotated Component Matrix to pick the factors that I want to use in the creation of the profiles, I told you that I use the coefficients from the Component Score Coefficient Matrix to score the standardized variables, and I told you that I use 50th percentiles to create 1/0 indicators for four factors, yielding sixteen Digital Profiles. Go ask any vendor on the planet if they give that much proprietary information away, for free. Seriously, go ask them. If you are a cataloger, go tell Abacus that you want to see the computer code behind the algorithm that chooses names for you. If you are an online marketer, go ask Google to share with you how they rank web pages for various search terms. You have the tools to do this for free, so go do it, start programming!
"Your use of Factor Analysis is completely inappropriate and semi-irresponsible." I am not trying to find a cure for cancer via clinical trials, and I am not trying to identify the corn hybrid that is most likely to grow well in Northern Missouri. I'm creating segments of customers who have similar traits, so that we can better understand and act upon customer behavior.
"If you can't use Digital Profiles in a tool like Google Analytics, then your methodology is useless to me. Why can't you create something that I can implement using the software tools that are considered a 'best practice solution'?" At some point, folks, we have to be willing to do a little bit of hard work. If you want simple answers to simple questions, then go ahead and ask Google Analytics to tell you the percentage of Twitter visitors that converted when they visited your website. If you want to know who your Twitter visitors are and you want to know what they are likely to do in the future, you're going to have to do some hard work. We've taught an entire generation of analytics experts to only depend upon simple answers to simple questions via simple software solutions. I know people who wrote computer code on punch cards, using mainframe computers, and they solved more complex questions than are solved by Google Analytics. It's time for all of us to step-up our game a bit!
"All of your examples are from retail or catalog or e-commerce. My business model is different, and my business is different. Why can't you create a methodology that ports well to my industry?" The methodology should work in any industry, though my focus is retail, catalog marketing, and e-commerce.
"My business is unique in that it has a strong seasonal component, so your methodology won't work for my business." Actually, it will work just fine ... just plug in seasonal variables (i.e. 1/0 indicator for customers who buy on Cyber Monday, for example), and the methodology works!
"My business is unique in that it is a gift business. Your experience in fashion retail isn't relevant to my business." Actually, the methodology will work just fine ... you could sell widgets and the methodology will work fine. I hear this argument a lot. Nobody has a business model that is so unique that somebody outside of the business model cannot digest the subtleties.
"Few of the big web analytics vendors ever talk about Digital Profiles. If they don't think much of your methodology, why should I bother to use your methodology?" Ok, give this a thought. In 2007, who talked about the importance of tablet-based computers that allowed you to navigate the web with a simple flick of a finger? And yet, today, the iPod Touch, iPhone, and iPad have fundamentally changed web navigation and use of the web. Apple may lose out to Android in the long run, but so what, they are the ones that changed how we will use the web, forever. Sometimes, folks, you have to do something that is a little bit different, if you fail, so what? If you succeed, well, what is the upside? Just because somebody at Omniture, Coremetrics, or the other big Web Analytics vendors doesn't think much of Digital Profiles doesn't mean your business won't be more profitable by using them!
"Catalogers typically work with Abacus or Merit Direct, depending if they are B2C or B2B. Neither of those vendors are using Digital Profiles to my knowledge, and they are industry leaders. Why aren't they adopting the methodologies you frequently talk about?" I'm here, folks, they can contract with me if they think this is valuable to their clients, and I'll show them how to do it. Or they can read my instructions, word-for-word, and do it themselves.
"How are you going to charge for a Digital Profile project?" My projects are based on the number of twelve-month buyers I have to analyze, the fee is based off of an equation that charges less and less for each incremental twelve month buyer. Small companies pay less, large companies pay more. I require 50% of the project fee up-front. Contact me now for a pricing worksheet.
"Your Digital Profiles do a bad job of incorporating channels. We want to see segments with only retail customers, segments with only online customers, and segments that have customers buying from both retail and online channels. We market via channels, so any Digital Profiles must account for this, right?" There are three easy answers for this. One, simply stick with your channel-based RFM strategy, and reap the benefits of that strategy. Two, you can run Digital Profiles only on merchandise preference, then overlay Digital Profiles across your classic RFM strategy. Three, you can trust a newer methodology that seeks to combine attributes across channels and merchandise divisions and geography and whatever other attributes you want to incorporate, yielding a modern, customer-centric segmentation strategy.
"What do we do when we need to add a new channel or merchandise division?" You probably need to create new Digital Profiles, just like when the online channel arrived and you had to add a channel component to your RFM scheme. If the new merchandise division will only account for less than 5% of annual sales, you probably don't need to make changes in the short-term.
"Why do you have to have sixteen Digital Profiles? That's too many in my opinion." Ok, go ahead and use just three factors, and you have eight profiles. Or use just two factors, and you have four profiles. You use just two factors broken down into low/medium/high criteria, and you have nine segments. You have the flexibility to basically do what you want. I've found, via experience, that sixteen segments based on four factors yields robust and actionable outcomes.
"Why do you have to have sixteen Digital Profiles? That's way too few in my opinion." Executives tell me that sixteen profiles are too many. Web Analytsts tell me that sixteen profiles are too few. You get to pick the number of profiles you want created, how's that?!
"Is the merchandise a customer purchases really that important? Our web analytics solution tells us that marketing channels and referring URLs and search keywords are really important." Your web analytics solution tells you that those attributes are important because the software solution is calibrated to serve online and search marketing applications. Yes, the merchandise a customer purchases is really that important, you've just not been exposed to software solutions that help you see that it is really important. Your web analytics solution only tells you answers to questions that the web analytics solution can answer ... and, by the way, that fact alone is really limiting our ability to learn more about the businesses we work for.
"You use only one year of purchase history, so clearly your methodology is flawed." Well, that's an assumption, isn't it? Here's something the big vendors don't tell you ... they don't tell you that data has a "half-life". In other words, something that happened fifteen months ago doesn't carry the same weight as something that happened two months ago. Now, I've created Digital Profiles that use older data, and here's the problem ... the solution quickly iterates to one that segments customers on the basis of being a customer for a long-time, vs. being a new customer ... the richness of individual channels (Mobile/Social/E-Mail/Search) is washed out, and the richness of merchandise divisions are washed out. So, depending upon what you are trying to accomplish, you may wish to use a lot of customer history ... so be it, do what you want!!! I score each customer on the basis of what that customer did in the year the customer most recently purchased.
"Why are you even charging for a solution that is, to me, nothing more than a glorified RFM scheme? You can get RFM software for a few hundred bucks, but you're going to charge so much more, so tell me why would I ever pay for your solution?" If you believe this to be the case, then go ahead and reap they benefits of your $150 RFM segmentation software solution. If I had an average catalog client that generated $100,000,000 a year in sales, I can find that client $1,000,000 in annual profit without a ton of work by using Digital Profiles in the targeting strategy ... classic RFM variables used in a statistical model account for half of the profit, Digital Profiles account for half of the profit. Basically, the Digital Profiles are pretty important.
"Can't I just pick out customers who bought via e-mail and purchased from merchandise division #8 and send them a trigger-based e-mail campaign when they act in a way that benefits me? I mean, why make things more complicated with Digital Profiles?" Of course you can do that! The majority of clients who are using Digital Profiles are doing so because they want additional business intelligence that cannot be obtained via other tools. They want to know if they have more "A Long Drive" customers than a few years ago, they want to know how "Gold Mine!" customers behave in Search marketing. They have a different set of needs that are well-served by a more-complex segmentation strategy.
"Your methodology misses out on all of the richness of today's Social Media environment. I want to know how networks of engaged fans/followers behave." Collect the right data, and you can easily use this methodology to analyze the richness of today's Social Media environment!
"We're an online startup, and we only have a year of history. Your methodology doesn't work well with our business model." I think it can work pretty well. Use the limited amount of data you have, and then adjust the methodology as you get more data.
"Can we import your Digital Profiles into our customer data warehouse, or into our web analytics solution?" Absolutely! In fact, Digital Profiles provide great results ... say you work with Webtrends or Unica or Coremetrics or Omniture ... yes, they can help you score visitors and help you analyze how well various Digital Profiles convert, or you can import the profiles into your customer data warehouse and yield very interesting results.
Do you have any questions? Please ask them in the comments section of this post.
Ok, let's look at a practical example of Primary / Secondary / Tertiary categorization of merchandise categories. In this case, we ...
Look at the first four rows of our life table (values of 0/1/2/3). These are the first 12-15 weeks after a customer buys for the firs...
We spent the past two weeks talking about the events and influences that shaped what I call "The Great Eight". My Influence...
If you don't like geeky math, please skip this post, because I am about to show you how the sausage is made! I have eight variables in...