Great Moments In Database Marketing #4: Interpolating Iowa

We go back to 1989 in the great State of Iowa, for this Great Moment in Database Marketing.

Iowa was a state burgeoning with opportunity in 1989. Kevin Costner was filming "Field of Dreams". And the Iowa State Fair boasted the World's Largest Pig.

At The Garst Seed Company, a little-known Statistical Analyst named Kevin Hillstrom was working on measuring corn hybrid performance in Iowa.

One might think that corn grows well in Iowa. But there are subtle differences in performance by geography. We executed numerous tests across the state, trying to understand which hybrids performed best across various geographies.

We used a tool called "linear interpolation". In essence, if you didn't conduct a test for a hybrid near Mason City, IA, you "interpolated" the results of the test, based on surrounding test plots North, South, East, and West of Mason City. You "averaged" the performance based on surrounding tests.

Interpolation yielded beautiful three-dimensional maps of Iowa (or any other state), with high terrain representing high yields, and valleys representing poor yields.

Fast forward to 2008. Our "on demand" world of rapid metrics gives us an endless array of fascinating insight into online customer behavior.

What Web Analytics fails to give us is interpolation.

Look at the following pair of tables, measuring conversion rate by prior visits to the website, and by depth of visit into the website (an "x" means there isn't enough data to obtain a valid estimate).

Before Interpolation
			Visit Depth
Visits	One	Two	Three	Four	Five+
One	1.0%	x	5.0%	4.0%	x
Two	x	6.0%	x	x	x
Three	x	x	7.0%	x	4.0%
Four	3.0%	4.0%	6.0%	x	x
Five+	2.0%	x	x	7.0%	6.0%
After Interpolation
			Visit Depth
Visits	One	Two	Three	Four	Five+
One	1.0%	3.0%	5.0%	4.0%	3.0%
Two	2.5%	6.0%	7.5%	6.0%	3.5%
Three	2.8%	5.0%	7.0%	6.0%	4.0%
Four	3.0%	4.0%	6.0%	6.5%	5.0%
Five+	2.0%	3.0%	6.0%	7.0%	6.0%

Interpolation helps one visualize the peaks and valleys inherent in all of our data. In this case, the Web Analytics experts observes optimal conversion among customers with modest number of visits and modest amount of site depth. Folks visiting only once but browsing deep into the site do not convert at a high rate. Folks visiting the site often, but only viewing the home page, do not convert at a high rate.

It's been a theme across this series. Just because we have instant access to hundreds of on-demand metrics in our Web Analytics package doesn't mean we have genuine insight into how our customers behave --- we simply have a lot of metrics! There is an art to transforming incomplete data into a compelling and actionable story. Interpolation is one of the tools that can be used to tell the story.

Great Moments In Database Marketing #5: Free Friday

If you are a database marketer, one who specializes on answering questions using SPSS/SAS/SQL, you've met up face-to-face with the concept of the "time crunch".

You begin your work day with a clear vision of what you want to accomplish. By 8:43am, your day has gone sideways. Elsie in Customer Service wants you to query the customer records of an individual who is upset that she no longer receives e-mail campaigns. The angry VP down the hall wants you to count how many customers responded to his promotion using a specific discount code last Wednesday. The earnest search marketer wants you to calculate how many customers visiting your site with a specific keyword place subsequent orders.

Before you know it, you spend your entire day writing queries for folks who do not have the skills necessary to query a database. If you go down this path, you lose your identify. You are no longer a valuable employee --- instead, you become a hybridized version of software.

Worse, you spend almost no time working on strategic issues. Everything you do is calibrated toward answering random "point in time" questions.

How do you fix this problem?

You don't!

But you can do something to mitigate the problem. I called it "Free Friday".

The idea isn't a new one. In 1996 at Eddie Bauer, I was stuck in a rut similar to the one I mentioned earlier. So I declared Friday to be "Free Friday". I would not answer any business question on Friday ... ANY Friday! Friday was "my time", time to research strategic issues and obtuse problems others didn't find important.

There was a price to pay for this freedom. I had to work forty hours a week Monday - Thursday to meet the needs of my co-workers. But then Friday was all mine.

Almost every key strategy that we employed in catalog marketing in 1998 - 2000 came from the "Free Friday" days of 1996 - 1997. Multichannel Forensics are almost entirely the result of "Free Friday". Our new store scoring algorithm and cannibalization metrics were the result of "Free Friday". Understanding online cannibalization was the result of "Free Friday". Interestingly, none of these topics were requested by co-workers or Sr. Management. They came from having the time to think clearly, to research ideas, to freely explore concepts.

Consider the concept of "Free Friday" in your work environment.

Great Moments In Database Marketing #6: Long-Term Impact of Promotions at Eddie Bauer

We go back to 1998 for this Great Moment in Database Marketing.

At the time, I was Director of Circulation at Eddie Bauer, a brand that was punch-drunk on promotions. Anytime a customer failed to purchase in six months, the "CRM/Circulation" process offered the customer a "20% off $100" promotion ... twenty percent off your next order of one-hundred dollars or more".

We tested these promotions until we were blue in the face. Continually, they showed that the customer spent about twenty percent more if offered this promotion.

So, the promotions became part of "what we did". And then my team decided to execute a long-term test. For the next six months, we would not offer a segment of lapsed customers a single promotion.

What do you think happened?

Take a look at the following table, a table that approximates the actual results of the test.

Eddie Bauer Six Month Promotion Test: 1998
	Receive	No	Incr-
	Promos	Promos	ement
Month 1	$10.80	$9.00	$1.80
Month 2	$9.00	$9.30	($0.30)
Month 3	$10.80	$9.60	$1.20
Month 4	$9.00	$9.90	($0.90)
Month 5	$10.80	$10.20	$0.60
Month 6	$9.00	$10.50	($1.50)
Demand	$59.40	$58.50	$0.90
Net Sales	$41.58	$40.95	$0.63
Gross Margin	$22.87	$22.52	$0.35
Marketing	$9.00	$9.00	$0.00
Promos	$4.07	$0.00	$4.07
Pick/Pack/Ship	$4.99	$4.91	$0.08
Profit	$4.80	$8.61	($3.80)
% of Sales	11.6%	21.0%	-9.5%

Oh oh.

Here's the 411 folks. When customers are continually promoted to, they delay purchases until the promotion is offered to them.

In our test, if customers were not offered promotions, they slowly began to "build momentum". Instead of the every-other-month cadence of promotions to this audience (the actual test had a different rhythm than illustrated above), the customer waited for promotions, did not receive them, then started spending more.

After six months, we noticed that customer spend in the two groups was nearly identical!

Now look at profit. Sure, the group that received promotions appeared profitable --- they appeared profitable via every system we had in the company, via every A/B test we executed.

But when viewed via a long-term A/B test, the results were significantly different. We were losing a boatload of money promoting to customers who would ultimately spend the same amount of money if we didn't execute the promotion.

In 1999, we dramatically pulled back on promotions. Total Net Sales decreased by maybe five or six percent. Total profit hit an all-time record high.

The core fundamentals of direct marketing are often violated in the world of "instant metrics" we've created. Our e-mail marketing friends read open rates and conversion rates from a "Free Shipping" e-mail within an hour of blasting the campaign. The adrenaline rush felt from obtaining instant access to customer behavior fuels strategy.

My challenge to the e-mail marketing and web analytics community, two communities that live and die by a steady diet of exhilarating and instantaneous metrics, is this ... do your metrics allow you to understand if what we observed at Eddie Bauer in 1998 is happening in your business? And if your visit-specific metrics don't allow you to observe a trend like this, what kind of systems/software/human investment is needed to allow for this style of measurement?


Hillstrom's Multichannel Secrets: Fifty-Nine Facts For CEOs!

Great Moments In Database Marketing #7: S&P 500

Back in 2001, I was brought in to help change the trajectory of Nordstrom Direct. The direct-to-consumer arm of Nordstrom struggled through another amazingly unprofitable year in 2000. A host of former Lands' End leaders partnered with in-house talent, tasked with turning things around.

The business began to crater in November 2000, with performance hitting rock bottom in Spring 2001. By September 11, any glimmer of hope was shattered by the events of the day. Soon thereafter, the focus shifted to enabling donation buttons for the Red Cross, and to anthrax in the mail system.

Our President was a plucky data hound named Mike Smith, current leader of Bag, Borrow or Steal. Mr. Smith felt that any problem could be solved by analyzing customer information. Mr. Smith charged me with figuring out "what was wrong with business".

Business is influenced by multiple factors. You self-inflict damage with the dumb things you do to yourself. Your competitors inflict damage with their strategies. And the economy can inflict damage.

You might recall that 2000 - 2001 was known as the "dot com bubble". Surprisingly, these bubbles just keep on happening!

I was able to develop a non-linear set of equations that correlated changes in the S&P 500 with changes in Nordstrom Direct sales performance. In fact, the equations explained sixty percent of the shortfall in business.

Not easily impressed by answers that fail to immediately fix the business, Mr. Smith panned the analysis. It is the responsibility of business leaders to improve business today, so the response was directionally appropriate.

But that doesn't mean you stop quantifying the impact external factors play in the trajectory of your business. Mired in yet another self-inflicted bubble, a business leader might want to know when we're "about to pull out of this mess". Pundits chattering on CNN probably won't give you the information you need.

So build relationships that correlate economic factors with changes in business performance. Use the equations as a forward-looking indicator to tell you when you're ready to pull out of an economic slump. Understand how much of the damage is created by the economy, and understand how much is self-inflicted.

Bob Bly vs. Robert Scoble: Old School Ideas vs. Newbie Technology

Bly's old-school thoughts on Facebook and Social Media and Direct Mail vs. E-Mail. Scoble's response.

Let's transition some of that discussion into the context of running a multichannel brand. Who's up for a test?

Who would be willing to execute this test within the context of their own brand?

• Customer group #1 is marketed to via direct mail and catalogs.
  
• Customer group #2 is marketed to via e-mail.
• Customer group #3 is marketed to via direct mail and catalogs and e-mail.
  
• Customer group #4 receives no direct mail, catalogs, or e-mail. They are simply left to the seductive wiles of social media.

Place your bets, folks ... which group drives the most sales and the most profit?

Social Media folks ... would you be willing to stand behind customer group #4? Or do you need the help of direct mail, catalogs, and e-mail (and paid search and portal advertising and affiliate advertising and shopping comparison sites) to succeed?

Your thoughts?

Great Moments In Database Marketing #8: The Rolling 12 Month File

The rolling twelve month file is one of the most under-utilized metrics in all of Database Marketing.

I was exposed to the metric in the early 1990s at Lands' End, after a few folks from Fingerhut made their way onto the Dodgeville campus.

The metric was put into use at Eddie Bauer in the mid 1990s. At the time, Eddie Bauer was a highly profitable brand, a brand going through an amazingly brisk retail transformation.

The rolling twelve month file is the "Dow Jones Industrial Index" of the Database Marketing world. It is a trailing indicator, telling you how many customers purchased from a product, brand, channel or store in the past twelve months. During times of change, the metric tells you what happened. It cannot tell you why something happened --- it is your job to figure out why!

We'd pick a market, say Omaha. Omaha did not have a retail store until the mid 1990s. We would measure how many customers purchased via the catalog, online, and retail channels in Omaha on a rolling twelve month basis.

When a store opened in a new market, the market was transformed. Catalog buyers decreased, online buyers flattened out, and retail buyer grew at a rapid rate. In other words, the new retail store was cannibalizing direct-to-consumer customers. And this is the way it generally works at brands that have a strong direct channel, then choose to add stores to the mix. Sure, you're now a multichannel brand, but the transformation comes with a cost.

In a new market, the rolling twelve month file for the retail store would stabilize within fifteen to eighteen month after a new store opened. After eighteen months, the market maintained a new balance between the channels, with each channel able to once again grow or thrive at the rate previous to the opening of the new store.

When you open a bunch of new stores, it is a good thing to conduct rolling twelve month file analysis for each market. The charts are also telling in saturated markets ... open a new store in a market that already has five stores, and watch what happens!

Any metric can be tabulated on a rolling twelve month basis. The key is to look for changes in the metric over time, then, dig into the data to understand why the metric is changing.

At Eddie Bauer, market saturation became obvious when viewing channels and stores through the rolling twelve month metric.



Hillstrom's Multichannel Secrets, Now Available!!

Great Moments In Database Marketing #9: The Square Root Rule

This is the second in a ten part series of key database marketing moments that shaped the focus of The MineThatData Blog.

We go back to 1991 for the origin of The Square Root rule.

"Back in the day", catalogers were experimenting with statistical models. At Lands' End, statistical models had been used to decide who received catalog mailings since at least the mid 1980s.

But statistical models can make deciding how much a segment of customers might spend a real headache.

Consider this example.

• A segment has 10,000 customers.
• Only 8,000 customers were selected by the statistical model to receive the mailing.
• The segment of 8,000 customers spent $3.00 per customer, $24,000 total, when mailed the catalog.

How much would the total segment of 10,000 customers have spent, if all customers were mailed the catalog?

It would take nearly seven years to find a simple solution to this problem.

Honestly, one can use "rules of thumb", or statisticians can create unique models for every catalog to solve this problem.

After witnessing the results of maybe 300 catalog mailings over a seven year period of time, a simple solution was created. In 1998 at Eddie Bauer, we developed "The Square Root Rule".

• The 8,000 customers above spent $3.00 per catalog, $24,000 total.
• If 10,000 customers would have been mailed, total spend would increase by the following factor:
• (10,000 / 8,000) ^ 0.5 = 1.25 ^ 0.5 = 1.118.
• In other words, 10,000 customers would have spent $24,000 * 1.118 = $26,832.
• Dollar per catalog for 10,000 customers = $26,832 / 10,000 = $2.68.
• The remaining 2,000 customers would have spent $2,832 / 2,000 = $1.42 each.

By simply knowing the percentage of customers mailed in each segment, this simple rule allowed the circulation analyst to "guesstimate" what might have happened if all customers in the segment were mailed.

The Square Root Rule applies to advertising budgets and page counts and items offered per e-mail campaign ... basically any situation where you have limited information and no good history to estimate what might happen.

Of course, the approximation has numerous limitations. Don't use it to extrapolate too far ... if you only mail 10% of the customers in a segment, the equation might fail. If you want to mail 3x as many customers as were mailed last year, the equation will fail.

But for many instances, this equation solves problems, especially if you have limited data and you don't have a statistician sitting next to you, awaiting your beckon call!