Mining Data

Dear Catalog CEOs:

Now that folks are terrified of the looming postage rate increase, you're probably going to turn to the Multichannel Forensics methodology to save mailing expense.

Wise decision!

As you already know, I score every customer in your file. Every customer is evaluated on the basis of their spending potential over the next twelve months. It looks something like this:

• Probability of Buying, Next 12 Months = 40%.
• Amount Spent, Next 12 Months, if Customer Purchases = $150.
• 12 Month Value = 0.40 * $150 = $60.00.

Of course, this $60.00 figure isn't important. What is important is how much the customer will spend if you stop mailing catalogs altogether.

The methodology predicts the percentage of demand generated without catalog mailings. This means that customers are evaluated only on their ability to spend money because of catalog marketing. This methodology is different than the matchback analytics that you've all used.

Here are two customers who my methodology considers to be identical in value.

• Customer #1 is worth $100 next year, with 10% coming from catalog marketing, 90% organic. Catalog Value = $100 * 0.10 = $10.00.

• Customer #2 is worth $15 next year, with 66.7% coming form catalog marketing, 33.3% organic. Catalog Value = $15.00 * 0.667 = $10.00.

Odds are that your statistical modeler or database provider demands that you mail customer #1, ranking that customer as being really, REALLY good.

My methodology views the two customers as being equal. Brand loyalty drives Customer #1, catalog marketing drives Customer #2. Both are equally valuable, when viewed via catalog marketing.

As mentioned last week, spots for Fall are starting to fill up, as smart Catalog Executives ask to have their Multichannel Forensics projects completed before the postage increase hits. Contact me now to reserve your project date!

I'm re-reading Mosteller's Fifty Challenging Problems in Probability With Solutions while traveling in Japan. This is absolutely one of my favourite books (not least because it is such a tiny volume). Fredrick Mosteller was "one of the most eminent statisticians of the 20th century" (according to Wikipedia), and a statistician who was passionate about education. The book actually has 56 problems, generally simply stated, like the following:

A drawer contains red socks and black socks. When two socks are drawn at random, the probability that they are red is 1/2. (a) How small can the number of socks in the drawer be? (b) How small if the number of black socks is even?

Reading the problems in this book lead me to thinking about what constitutes a 'good' puzzle. Personally, I like those problems that have one or more of the following qualities:

• Counter-intuitive: the birthday problem is like this, 23 sounds like a low number.
• Relies on basic number theory: Mosteller's problems require things like working with geometric series, binomial coefficents, etc.
• Motivates you to think probablistically about things that you don't generally consider in that way: for me, those problems involving randomly throwing sticks on a table are fun (but I do remember Johnny Ball performing this experiment live once on the BBC...)
• Problems that seem recursive but can be solved simply.
• Entertainingly cast as a brief story (e.g., a three person dual).

Intruigingly, Mosteller's solutions, while nicely written with an engaging informality, still require concentration, especially when they skip a few steps on route to the prize. Generally, he demonstrates a great way to attack the problems by first playing with a few examples, and then running through one or more full solutions (often employing induction and negation tactics).

If you are looking for other good sources of puzzles, Car Talk (on NPR, Saturday at 9) offers not just amusing car problems, but also the occasional Mosteller like puzzler...

I've just watched the video introducing Firefox's TabCandy:

An Introduction to Firefox's Tab Candy from Aza Raskin on Vimeo.

Much of what is presented is not novel - but novelty is not the heart of innovation. Execution is, and TabCandy looks like it has a great chance of taking off - the Firefox userbase and the simplicity of the user experience.

Bing's whole approach to differentiation has been to focus on tasks, and a big part of TabCandy is to help the user scale over tasks, especially those with a long term nature. Should the browser embrace the information sources, or should the search engine embrace browser like state?

Here's a link to the story in the Wall St. Journal.

Here's a big hint. When your website is down for several days, you have a GOLDEN opportunity to analyze the impact of your e-commerce business on your total business.

• Did comp store sales increase, or did they decrease?
• Did customers try to order merchandise over the telephone?
  
• In the seven days after the website was down, did website sales increase vs. plan (i.e. recapturing sales) or did sales stay flat vs. plan (i.e. you lost all sales during the outage)?

If your comp store sales didn't change while the website was down, then what does that say about how your website drives sales to your stores?

These are magical moments for an analytics individual, or for a marketer. When your website is down, you have the most fertile analysis environment you're ever going to have!

Once again, I enter Lauren Fetzer's office, as we prepare for her weekly Executive meeting.

Kevin: "What are you listening to?"

Lauren Fetzer (CEO): "'I Got A Man' by Positive K. It doesn't get much better than that, does it? Memories of college.

Kevin: "Alright".

Lauren Fetzer: "Here's what we're dealing with today. Penny Parker has her own Twitter account, and I pretty much let her run with it, I mean, what damage can a Twitter account do with 37,000 followers, of which only 27 really pay attention to anything that Penny says? Well, the rest of my team seems to think that Penny is a few french fries short of a happy meal, they hate what she tells her followers."

Kevin: "Didn't you hire me to help you provide a data-driven roadmap to 2015?"

Lauren Fetzer: "Yup, and we get there by resolving small issues that folks perceive to be big problems. Even Patton had to figure out where latrines should be stationed."

Kevin: "Alright".

We enter the Executive boardroom.

Lauren Fetzer: "Ok team, let's get down to business. Penny, how was business yesterday?"

Penny Parker (VP Marketing): "The website was 8% behind last year, our suite of mobile apps were 200% over plan, so in total, the direct channel was 5% ahead of last year. Retail comps were +4%, with Downtown Seattle leading the way at +11% and Alderwood bringing up the rear at -9%"

Bart Cox (VP Stores): "That's the best Alderwood has done in two months, folks. And Penny, while we're at it, why exactly did you tell everybody on Twitter that Alderwood is running double-digit negative comps. That's proprietary information, you can't share that!"

Ashley Zimmerman (VP Merchandise): "Yeah, and last week you told your followers that our Fall assortment delivery is delayed by a week due to problems at the Port of Tacoma. That's proprietary information, you can't share that!"

Connie Simpson (VP Finance): "Yeah, and what was up with your comment about us taking a bath on Streamline Yukon Hiker gross margins? That's proprietary information, you can't share that!"

Bill Bledsoe (VP Logistics): "Yeah, and you told your followers that you were giving a full refund to @butterbean411 because his shipment took nine days to be delivered. Now everybody who has a late shipment will want a refund, you can't share that!"

Penny Parker: "Kevin, help me out here. You analyze our comp customer segment. Those customer spent $82 last year, year-to-date, what are they spending this year, year-to-date?"

Kevin: "Your comp customer segment is spending $91 this year, year-to-date".

Penny Parker: "Customers are spending more this year than last year, hmmmmmm. Kevin, help me out here. What is the conversion rate for our average customer, and what is the conversion rate for customers who we know have a Twitter account?"

Kevin: "Your conversion rate last week was 8.2%, the conversion rate last week for customers with a Twitter account was 18.4%."

Penny Parker: "Where's the damage to the business, folks?"

Bart Cox: "Those are psuedo-metrics. Comp customer spend might have been $93 this year, year-to-date, had you not been out there blathering about company secrets."

Connie Simpson: "And you can't compare conversion rates between those two audiences, because the Twitter audience self-selects itself --- those are already our better customers. And why is it acceptable to share the fact we're taking a bloodbath on the Streamline Yukon Hiker? That gives our competition an advantage?"

Penny Parker: "What advantage? They browse our website every day, they knew the item was $129 and then was marked down to $99 and was marked down again to $79. Are you telling me that our competition is so stupid that they can't figure out that a 39% price reduction isn't damaging to the gross margin of that item?"

Bart Cox: "Maybe."

Ashley Zimmerman: "You just can't announce that we're having problems with product delivery. Aren't we supposed to be a multi-channel brand? We should do all things across all channels at the same time. We'll tell customers via e-mail and the website and stores on August 15 that the fall assortment is ready. The customer doesn't have to know that we were supposed to have the fall assortment ready on August 8."

Penny Parker: "Remember last February, I told our audience how awesome the spring assortment was, I featured key items in e-mail before they were available, we advertised key items online weeks before they were available. When the assortment was available, we had website comps of +80% for three consecutive days. Nobody complained about that."

Ashley Zimmerman: "Those comps happened because customers loved my merchandise."

Penny Parker: "Those comps happened because I told my customers that they had to love your merchandise."

Ashley Zimmerman: "Those aren't your customers, Penny. Without merchandise, you don't have anything to market."

Penny Parker: "And without my marketing efforts, how would anybody ever know you were even selling merchandise?"

Bill Bledsoe: "Don't we have guidelines for what employees can share with the public?"

Penny Parker: "Of course we do. But what I am doing is different. I'm not tweeting about how great my Crab Benedict was at the Rocky Shore Cafe this morning. And I'm not tweeting that we made $1.7 million of pre-tax profit last quarter. I'm tweeting the Fetzer's Footwear lifestyle. I am giving our customer an insider's view of our company, helping our customer feel a bit closer to us. I don't sell shoes on the same web page that I sell Ad Words on like Target does. I sell Fetzer's Footwear, good, bad, or indifferent. Customers don't trust tweets that are like '... another great meeting, we have the best employees'. Customers follow us because we sell a realistic life experience congruent with their life experience. Heck, we have 37,000 followers on Twitter, Nordstrom only has 31,000 followers, and they sell, what, a billion dollars of shoes each year or more? Come on, something is working here."

Connie Simpson: "Is it working? What is the ROI of your Twitter antics? Why don't you prove to us that you drove a million in sales last quarter because of your comments? Heck, we have a morning dashboard that proves that Bart Cox is failing in his efforts to revive Alderbrook, he gets nailed for that every morning. Who holds you accountable for your comments? Show me the ROI of your Twitter activities."

Bill Bledsoe: "That's a good point, Connie. I get hammered each time our call center and distribution center expenses exceed 10% of sales ... it's a simple, easy-to-understand metric. Kevin, what is the simple, easy-to-understand metric that proves that Social Media is paying the freight?"

Ashley Zimmerman: "That IS a good point, Connie. My merchandise sales are tallied at a divisional level. If a division is failing, I know it first thing every morning. Lauren, you look at me and you ask why sandals were down 4% vs. last year, and I have to explain myself. We are all held to certain levels of accountability. So yes, Kevin, what is the metric that proves that Penny is contributing to the bottom line?"

Kevin: "Don't let anybody tell you that there is a metric that proves that Penny is contributing to the bottom line via Social Media. Social Media is, by definition, not measurable. Oh sure, you can measure followers or click-through rates or coupon redemptions, that's all measurable. Say a customer purchased via a coupon from Twitter. How do you prove that the customer wouldn't have ordered anyway? The majority of Social Media metrics are faux metrics, providing the illusion of accountability. The metrics are not of the family of metrics that the rest of you are measured by, and that's just a reality of life, sorry to say."

Bart Cox: "Then Penny should stop tweeting company secrets, because we can't prove that her comments generate ROI."

Penny Parker: "No, I should tweet more, because my tweets may be responsible for a two or three percent increase in sales."

Lauren Fetzer: "Ok, I've listened to enough of this garbage. Bill, did I ever ask you to prove to me that warehouse robotics would generate a positive return on investment? You know you can't prove that our customer retention rates are better because of robotics. And yet, you lobbied for them and I gave you a ton of capital to get that done, right? How much capital does it cost Penny to tweet her comments to our customers? And Connie, you wanted a new ERP system installed last year. Did I ever once ask you to prove how much more customers would spend because we linked our Human Resources system to our Finance systems? Do you honestly believe that because HR can talk digitally to Finance that a customer in Spokane says 'oh goodness, that's neat, I'll buy an extra pair of shoes because of their neat back-end systems integration process'? Bart, you ordered 275 mannequins last week, did I ask you to prove to me that the ROI on the mannequins would offset the cost? Did you ever think that Penny might view your purchase as being stupid, that your purchase won't possibly sell another pair of shoes? And Ashley, come on girlfriend. If you think our customers cannot live without your merchandise, go work for somebody else, and we'll be able to prove your value in comparison to the merchant we hire to replace you."

Ashley Zimmerman: "Why are you siding with Penny?"

Lauren Fetzer: "I'm not siding with any employee, I am siding with the customer. I always take the side of the customer. If we have 37,000 customers who are riveted with what Penny has to say, she must be doing something right, and I'm willing to take a leap of faith that her low-cost method for engaging with customers is a risk worth taking. Most of you weren't here in 1995 when I told a store-based brand that we had to move online. Bart, you remember those days. You used to mock me, you'd ask if I sold 9 or 10 pair of shoes a day, then you'd laugh and walk down the hall and tell me to enjoy 'fun time'. Now the website is four times as big as our stores are, and you don't mock me anymore, do you? We can all see the future. The future clearly isn't about our website. The future is some fusion of social, mobile, our website, and our stores. And we can't predict what that will look like, can we? So we have no choice but to let Penny experiment. We're not going to legislate her comments by committee. If I think she is out of line, I'll whop her upside the head and get her in line. Otherwise, give Penny room to innovate. We just wasted a perfectly good hour of time, and we are no closer to having a plan for what our business will look like in 2015, are we? Go get bus, go invent the future. This meeting is adjourned."

Ashley Zimmerman (to Connie Simpson): "I think Lauren got up on the wrong side of the bed this morning. Take her out to drop off some crab pots and calm her down!"

For many of us, testing website strategies is a matter of running simple "A/B" tests.

This process seems easy enough. There are automated algorithms that compute how many conversions are required before one strategy can be labeled as the "winner" against another strategy.

Most of the time, this process is done on the basis of "conversions" ... did a customer buy something, did a customer subscribe to something ... you get the picture.

And most of the time, this strategy yields an outcome that is wrong.

Yup, you heard it here first. Your conversion-based A/B tests, while yielding a statistically significant outcome from a conversion rate standpoint, are yielding an "incorrect" outcome when evaluating what matters ... spend per visitor.

Here's why ... for those of us in e-commerce, we are measuring "conversion", when we should be measuring "dollars per visitor". And "dollars per visitor" is the component of two metrics.

• Conversion Rate.
• Dollars per Conversion.

When you multiply variability (conversion rate) by variability (dollars per conversion), you get ... VARIABILITY! Lots of variability!

Take a look at this example. This is a test, eleven customers in each group (for illustrative purposes only).







































































































The test group clearly outperforms the control group (7/11 conversions vs. 2/11 conversions).

Also notice that the average amount spent per conversion is $100 per conversion, the same across each group.

However, the variability associated with spending amounts (they vary between $30 and $170 per order in both the test group and the control group) causes the t-test to yield a result that is not nearly as statistically significant as when measuring conversion rate alone.

Well guess what? This happens ... all of the time! Every time you run one of those software-based automated conversion rate A/B tests that measures responses until one outcome is statistically significant, you guarantee that measurement based on dollars per visitor is not going to be significant ... in other words, every test you are running gives you a statistically insignificant outcome!!

So do me a favor ... please, I beseech thee ... please run your A/B conversion rate tests against the statistical significance of dollars per visitor, not on the basis of conversion rate.

You will learn that you have to have 2x - 4x as many customers in your test, but your results will be valid. As being executed now, so many of the tests being executed out there are yielding garbage for results ... and this partially explains why a decade of conversion rate optimization has yielded websites that, by and large, convert fewer customers than in the past.

Logging on to Facebook from Japan triggered a lengthy authentication process. Having only used the site from US IPs logging in from a Japanese address forced me to go through an authentication process which involved answering a number of face identification questions in a multiple choice format. This was quite un-nerving for the following reasons:

1. Some of the people I knew from face to face interactions, but I hadn't seen in a long time (people change over time).
2. Some of the people I had never met and was barely able to figure out who they were. If it weren't for the multiple choice format of the challenge I would have failed.
3. Some of the photos - this is true - had tagged people who were wearing fancy dress costumes including a full mask. The face was literally not visible, and I wouldn't be getting any points for recognizing a dead president. Fortunately Facebook showed two different images per person.

Facebook gets full points for an innovative application of tagged pictures, but it wouldn't have surprised me at all if I had failed to 'reactivate' my account (following the language used, logging in from a different country must 'deactivate' your account).

The irony for me was that I had actually meant to log on to Foursquare but my fingers weren't listening to my brain. I was actually already in the authentication process before I realised - this doesn't look like Foursquare...

One of the easiest segmentation strategies you can employ relates to new items.

See, it turns out that you have customers in your database that love buying new merchandise ... these customers hate boring companies, they crave something interesting and exciting.

Conversely, you have customers who love the same-old-same-old. When it comes to e-mail marketing, for instance, these customers want something they can trust. Why not serve them up what they want?

Segment your customer base by merchandise preference ... create a variable that divides customers into those who love new items, those who buy a mix of product, and those who buy tried-and-true favorites. Use the segmentation variable to serve up home pages and landing pages that are congruent with customer interests. Use the segmentation variable to deliver e-mail campaigns that have merchandise aligned with customer interests!

I've just done my first Japan checkin on Foursquare - it was non trivial. Firstly, I had to add a location via Foursquare's website. This was a real challenge as the site is not really aligned with the way in which Japanese addresses are expressed. After a lot of hacking (throughout which Foursquare kept telling me it couldn't find the location) I figured out that if I could tickle the Google location search via the Foursquare interaction, I could get the right place to show up. Finally, this succeeded and I added 京急富岡駅.

This would have been a lot easier if I could have directly located the venue on the map (as a lat/lon position) and then simply named it. That mode of entry seems like a pretty general backup plan for any problems, and easy to implement, so I'm assuming it is not available due to some potential for abuse.

Then I wanted to check in. This is can be done via the Foursquare mobile web page (first you need to update your location).

,

Stop over at the Web Analytics Demystified blog, where Eric T. Peterson was kind enough to publish my essay about five opportunities for Web Analytics professionals to expand our "analytical worldview". Thank you, Eric!!

And if you have a need for some Web Analytics consulting, give 'em a holler!

Dear Catalog CEOs:

Another week, another article about a looming postage increase, courtesy of DMNews.

Regardless of your angst, you're going to cut circulation, aren't you? And you need to figure out who should receive fewer catalogs in the future. One population that makes sense are the customers who so love your brand that they will shop online without being prompted with catalog mailings.

I usually execute two Multichannel Forensics projects per month ... in those projects, I grade all customers in your database ... "A", "B", "C", "D", and "F". You cut back, cut way back, on the Cs, Ds, and Fs ... with little impact on demand and a nice boost to your p&l.

The typical project yields between a 0.5% and 1.0% profit to demand ratio. In other words, if you are a $50,000,000 business, I'll find you between $250,000 and $500,000 in profit opportunity. Some projects go way beyond that ... I found $1,500,000 for a $75,000,000 business by simply adjusting the contact frequency to average housefile buyers.

I'm taking project reservations now for September and October ... with the postage increase looming on the horizon, you want to have the analysis done this fall, so you are ready to react in 2011, correct? So contact me now to get your project into the queue.

Last year I wrote about some work done by Sune Lehmann and colleagues at the Barabasi Labs which explored the relationship between location and affect signals in Twitter (here). Sune pointed me to a recent update which extends the work to a cartogram approach to visualizing moods and volume: Mood, Twitter and the new shape of America.

The work explores both visualization methods and the data itself. It also demands answers to some interesting questions, the least of which is the apparent difference between the coasts and the bits in the middle. I'd be interested in seeing an analysis of these distinctions,

A year ago, this blog essentially had no mobile visitors.

Three months ago, 3% of visitors came from mobile devices.

Today, 5% of the visitors come from mobile devices.

• #1 = iPhone
• #2 = iPad
• #3 = Android
• #4 = iPod
• #5 = Blackberry

69% of the visitors are new (vs. the site-wide average of 60%).

All "engagement" metrics trend worse than the site-wide average (a 'duh', given that the site isn't optimized for a mobile experience).

RSS subscribers, of course, must be using mobile devices at a far greater rate than the average blog visitor via the website.

The world is changin', folks.

Today, I am joining Lauren Fetzer, CEO of Fetzer's Footwear, at her Executive Team meeting. As always, I enter Lauren's office, where she is listening to her iPod Touch.

Kevin: "What are you listening to?"

Lauren Fetzer: "The Boy Is Mine ... it was that Brandy / Monica song from the late 1990s, don't ask. I got this song from Napster back in 2000. You might say I stole the song, lawyers would certainly feel that way. And that's our topic at our Executive meeting today."

Kevin: "Illegal file sharing?"

Lauren Fetzer: "Close. How do you deal with competitors that knock off your product."

Kevin: "Alright."

We walk into the Boardroom.

Lauren Fetzer: "Ok Penny, what did sales look like yesterday?"

Penny Parker (VP Marketing): "E-Commerce was up 3% vs. plan and up 4% vs. last year. The website was down 11%, mobile apps continue their meteoric growth. Retail comps were +5%, led by Redmond at +16. Alderwood trailed the field, at -14%."

Lauren Fetzer: "Bart, what in the name of S.S. Kresge is going on with Alderwood?"

Bart Cox (VP Stores): "Why don't you ask Penny what is going on with the website?"

Penny Parker: "We already know the answer to that one, crankasaurus, the mobile app is cannibalizing the living daylights out of the website. You know that nobody is cannibalizing the living daylights out of Alderwood."

Bart Cox: "So how, Penny, are you going to drive increased website sales? And Lauren, why don't you ride Penny like you ride the Alderwood store? And when are we going to renovate Alderwood? That store hasn't been renovated since 1998? Maybe the reason that store performs so bad is that it looks like it is preparing for Y2K."

Lauren Fetzer: "Ok Bart, you made your point."

Connie Simpson (VP Finance): "Bart, you already know that Alderwood isn't on the renovation schedule until 2012."

Bart Cox: "By then, the store will have lost 25% market share. Smart. Retail is like a zit to you folks, e-commerce is all that matters around here."

Ashley Zimmerman (VP Merchandising): "No, merchandise is what matters around here, we all know that. Here's the problem of the day. Did you see that Zappos is selling the Hi-Tec Women's V-Lite Altitude Ultra? One-hundred and twenty-four bucks. And Fred Meyer has a knockoff of that item, a store brand selling for $49.99. We invented that shoe last year, remember? It is our Womens Q-Max Mountain Climber, and we sell it for $199. It's the same shoe. We have a fully gusseted leather tongue, moisture wicking textile lining, rustproof hardware, ours is 16oz, I mean, it's a complete knock-off by Hi-Tec, and I heard that Zappos is selling a ton of the things.

Connie Simpson: "What happened to sales of our Q-Max Mountain Climber since Zappos began to sell the Hi-Tec knockoff?"

Ashley Zimmerman: "Sales are down 39% to last year."

Bart Cox: "Lauren, what are you going to do about the sales decreases in the Q-Max line of shoes?"

Bill Bledsoe (VP Logistics): "We probably need to price match here, right? Maybe we can make up margin dollars by selling more."

Ashley Zimmerman: "That's a hard one. We'll sell 500 units at $199 and our cost of goods is $85, so we make $57,000 of gross margin. Say we lower the price to $119 and we sell 50% more units. That's only $25,500 of gross margin. Say we sell 100% more units, that's only $34,000 of gross margin. In fact, let me run the numbers here ... ... ... ... ... we'd have to sell 3.4 times as many shoes in order to break-even on the gross margin line. That won't happen."

Bill Bledsoe: "But you'd have more market share, and that counts for something, right?"

Ashley Zimmerman: "Sure, it counts for a reduction in profit of $23,000."

Bill Bledsoe: "Can't we sue them? I mean, we have a patent on the Q-Max line of Mountain Climbers, right?"

Ashley Zimmerman: "No, their shoe is just different enough to make it a unique line. It's a 16oz shoe, ours is 15oz. They have the exclusive Vibram hiking outsole, too."

Connie Simpson: "Penny, what is a viable marketing strategy here? I mean, how do we communicate to the customer that this shoe at $199 is fundamentally better than the shoe they can get at Zappos for $124?"

Penny Parker: "We can do an e-mail blast with a customer testimonial. Let's see ... we have an e-mail list of 100,000, and we'd get a 20% open rate, and 35% of those would click-through to the website, and 5% of those customers would buy the shoes, so that's ... what ... maybe 100,000 * 0.20 * 0.35 * 0.05 = 350 customers purchasing the shoes, so that's something."

Bart Cox: "Are you telling me that we have an e-mail list of 100,000 names and only 350 would buy something if we send an e-mail marketing message? 350? I thought e-mail marketing had the best ROI of any marketing channel? 350? I mean, why bother?"

Ashley Zimmerman: "350 units will translate to 250 units after accounting for returns, so we'll increase sales of the item by 50% over the 500 we'll sell without e-mail marketing support."

Bart Cox: "But that's a one-time fix. We send out something like 100 e-mail campaigns a year, so we have to advertise other products as well, right? So this e-mail campaign doesn't solve the core problem, here. The core problem is that Fred Meyer stores can sell a knockoff at $50 and Zappos can sell a knockoff from Hi-Tec at $124, and both win while we lose. So, what is our strategy? How do we deal with this?"

Connie Simpson: "Locally at least, we have a multi-channel advantage, don't we? I mean, nobody shops the Fred Meyer website, and Zappos doesn't have stores, so we should be well-positioned locally. Kevin, that should work to our advantage, right?"

Kevin: "On a local level, it could help you. All of your stores are here in the Pacific Northwest, so I can take a look at customers in Northwest Washington to see if they behave differently than in the rest of the country because of the Fetzer's Footwear multi-channel presence."

Penny Parker: "And Woodside Research said that by 2015 they expect multi-channel brands to see a 4% increase in market share over online brands".

Lauren Fetzer: "Ok, team, that's wonderful, but all you discussed were tactics and analytics. That makes it sound like you are busy or strategic, but not one person mentioned a solution to our problem, not one person even offered a suggestion. Allow me to restate the problem for you. What is our response when a big brand offers a knockoff item at only 25% of the cost? What is our response when an online brand offers a branded knockoff item at 37% off? Be honest, team. What is the reason that you would buy our item at $199 instead of buying the knockoff item for $125 at Zappos? It can't be customer service, because Zappos will get it to you tomorrow while we'll take 5 days to deliver it to the customer, and we'll charge the customer $5 to ship it to her. It can't be quality, because the customer realizes that for a 37% price decrease she'll accept lower quality. And we can't price-match big brands because that will just drive us out of business. So, again, I ask all of you, what is the reason that a customer would buy our shoe for $199 over the same shoe for $125 at Zappos or $50 at Fred Meyer?"

Bill Bledsoe: "We can't ever win that battle, can we?"

Lauren Fetzer: "I guess we can't. Oh well, let's just shut down operations today and go out of business. I'm sure Nordstrom is hiring in their shoe department, so we'll all enjoy selling the Hi-Tec Women's V-Lite Altitude at $125 a pop, earning a 7% commission on every single unit we sell. Sounds good to me."

Connie Simpson: "Can't we talk up our history? I mean, Zappos is a decade old business."

Lauren Fetzer: "Customers don't care about history. If they did, we'd all be shopping in Montgomery Wards stores, and the downtown corridor of any mid-sized city would have a vibrant collection of Gimbels department stores. No, we need to offer the customer a compelling reason to shop us over our competition. What is the compelling reason for a customer to shop us and not shop the competition? I need a compelling marketing answer to this question. I need a merchandising answer to this question. I need a logistics answer to this question. I need a store answer to this question. And I need the financials to work to our advantage. I don't need one-off e-mail campaigns, I don't need product development cycles that are too long to please the modern customer, I don't need it to take 5 days to deliver a pair of shoes to a customer, I don't need multi-channel bricks 'n clicks marketing theory that only benefits consultants and bankers, no offense Kevin, and I don't need to keep seeing the Alderwood store pulling -14% comps. I'm paying all of you $200,000 a year plus a 60% bonus to provide real solutions. So let's get to work on providing real solutions."

If business is not performing to expectations, where do you invest the most attention, and why? Prioritize these areas in order from most important to least important.

• Catalog performance.
• Website performance.
• Merchandise performance.

Discuss.

You've probably already broached this subject, right? I mean, you've used your web analytics package to thoroughly analyze the future value of customers, right?

AMEX buyers are different than Visa buyers ... Visa buyers are different than Master Card buyers ... Master Card buyers are different than Discover Card buyers (pay attention to the Discover Card buyer), and Discover Card buyers are different than PayPal buyers.

And then, if you have proprietary credit, you've got a whole 'nother set of exciting challenges.

It turns out that the method you choose to pay for your merchandise plays a role in determining the future value of a customer. And this is such an easy thing to analyze, isn't it?

So just do it!

Dear Catalog CEOs:

If you are in the B2B world, then you are probably attending the Merit Direct Co-Op this week. This is a wonderful forum for B2B leaders to discuss various topics and issues.

True catalog brands are really not competing with each other anymore. Heck, there's enough trouble out there on the horizon from low-price online brands and retail big-box monsters, all of whom are eroding your market share! There's never been a better time to be collaborative, to work with fellow catalogers and to learn from each other.

It's time for a shift in thought ... work with your fellow catalogers, and learn from each other. There are too many forces out there looking to take business from catalogers via other channels, so use forums like the co-op this week to foster relationships with like-minded individuals.

Recently, I've been working through a number of planning processes with varying degrees of uncertainty. Some thoughts:

• There are a number of different areas of uncertainty with any systems plan. Two key areas of uncertainty are engineering, or technical, and data and inference.
• Uncertainty around engineering involves questions about technical capabilities of platforms and architectures with respect to requirements.
• Uncertainty around data and inference involves issues to do with the characteristics of the data (does it even include the stuff that I want?), the nature of the inference problem (is it mathematically attainable to the degree required?).
• Engineering uncertainty can be reduced by the application of technical knowledge and rapid prototyping.
• Data and inference uncertainty requires exploration of data (assuming it is novel to the team).

In general, one can't take a rationalist approach to planning and, in my opinion, it is better to reduce uncertainty by acting - whether that acting be prototyping, data mining, or whatever. In this regard, the exercise of building an end to end system (where some thought and commitment is given to the gross functional architectural design) is useful.

Barney recommended reading 'Made to Stick' by the brothers Heath. Generally, I'm somewhat wary of bestseller big idea books, but I'm finding this one already engaging.

In the first chapter, there is an account of the military's approach to planning. Colonel Tom Kolditz is quoted:

No plan survives contact with the enemy. [a phrase attributed to Helmuth von Moltke the Elder]

The book then goes on to describe the notion of Commander's Intent - meaning what is important is the intended result. It is imperative that this is crisp and the rest of planning flows from the structure of the organization and the adaptation of actions in the course of the activity up to the culmination of the goal.

Of course, in the fierce world of internet applications and services, one might regard the competition as the enemy. But the analogy works well with respect to planning where the data is the 'enemy', thus:

No plan survives contact with the data.

I was talking with a colleague today who is taking a business-oriented data mining course, and there was a list of neural network books recommended by the instructor. It was fascinating looking at the books in the list because I didn't know several of them. When I examined several of the recommended books on amazon.com, I found they contained what I would call "academic" treatments of neural networks. That means they covered all kinds of varieties of neural networks, including brain-state-in-a-box, Boltzmann machines, Hebbian networks, Adaline, ART1, ART2, and many more. Now I have nothing against learning about these techniques on the graduate school level, or even on the undergraduate level. But for practitioners, I see absolutely no advantage here because they aren't used in practice. Nearly always, when someone says they are building a "neural network" they mean a Multi-layered perceptron (MLP).

When I use neural networks in major software packages, such as IBM-SPSS Modeler, Statistica, Tibco Spotfire Miner, SAS Enterprise Miner, JMP, Affinium Predictive Insight, and I can go on... I am building MLPs, not ART3 models. So why teach professionals how these other algorithms work? I don't know.

Now neural network experts I'm sure will find times and places to build esoteric varieties of neural nets. But because of the way most practitioners actually build neural networks, I recommend sticking with the MLP, and understanding the vast numbers of options one has just with this algorithm. This is one reason I like the Christopher Bishop Neural Networks for Pattern Recognition. Check out the table of contents--I think these topics are more helpful to understand than learning more neural network algorithms.

Another option for spinning up on neural nets is the excellent SAS Neural Network FAQ which is old, but still a very clear introduction to the subject. Finally, for backpropagation, I also like the Richard Lippmann 1987 classic "An Introduction to Computing with Neural Nets (8MB here).

For the past decade, we've been taught how critically important it is to be "multi-channel".

Intuitively, the concept makes perfect sense. Maybe I want to rent a movie. Theoretically, it makes logical sense that I'd drive four minutes to my local Blockbuster store to pick up a copy of, say, "Idiocracy", because I want to see the movie now, not in two days when Netflix delivers it to my home.

Go visit the Multichannel Merchant website, and browse the hundreds of articles that have been written about Blockbuster multichannel strategy over the past decade. You'll be riveted by the strategic choices that Blockbuster made, choices that any pundit would laud:

• Blockbuster makes a bid for Circuit City.
• Blockbuster Canada launches a loyalty program.
• Rent a movie online, return in stores.
• Integration and Segmentation, a Blockbuster concept.
• Blockbuster Incents Netflix Customers With Free Rentals.
• Blockbuster Offers Free Rentals Via Expanded Rental Program.
• Blockbuster Lends Marketing Muscle For The Happy Feet Film.
• Blockbuster Combines Movie Cards With Meal Card Promotion.
• Blockbuster E-Zines Get People In The Store.
• Blockbuster Plays Ball With MLB In-Store.
• Blockbuster Settles With 48 State Attorney Generals Over "No Late Fees" Policy.
• Blockbuster And MasterCard Team Up For Movie Pass Program.
• Blockbuster UK Chooses MicroStrategy to Provide CRM Functions.

That's a lot of multi-channel strategy ... bricks 'n clicks and promos and CRM and co-branding and channels and freebies and segmentation and even loyalty programs. And yet, last week, Blockbuster was delisted, as the brand struggles to make interest payments on more than $900,000,000 in crippling debt.

Be honest with yourself, especially if you promote multichannel strategies:

• Why do you choose Netflix over Blockbuster?
• Why do you choose Amazon over Barnes & Noble or Borders?
• Why do you choose Zappos over Nordstrom?

If you can answer those questions, you've solved a marketing riddle that escaped the multichannel generation of business leaders.

(Hint #1: A good part of answering the riddle involves prudent financial management of a business. It doesn't matter if a customer can order a movie and return it in a store if a business chews up half of all profit dollars or more on interest payments, does it?).

(Hint #2: Focus on merchandise and service and product delivery before you focus on channels and CRM and loyalty programs and promotions and co-branding ... maybe that is the true answer to the marketing riddle of why multichannel marketing doesn't work).