Multichannel Forensics And The Five Customer Types

By now, you know that I advocate segmenting customers into one of five different segments, as illustrated here.

Why?

Because the five segment definitions accurately portray levels of profitability and loyalty. Even more important, you get to see how your business is evolving.

There are several dangerous combinations in this structure.

Advertising and Algorithms interact in a bad way. The cataloger knows this all-too-well, right? The cataloger sends a million catalogs out into the ether, only to drive the customer to Google, where the customer has fifteen choices for nearly identical product at comparable prices.

Honestly, there's nothing Google likes more than traditional advertising, because traditional advertising sends a customer to Google, where the customer clicks on paid search links. One brand spends money, Google intercepts the demand, and benefits financially by re-distributing the demand to paying advertisers.

Advertising and Begging interact in a bad way, unless you are an inventory manager. The marketing executive creates demand by begging via advertising vehicles. Customer loyalty is now shot, as the customer is being paid to buy something --- worse, Advertising and Begging can drive the customer to an Algorithm. Google loves Advertising stuffed with Begging!

Social shopping is something we're only starting to consider. Retailers have known the power of social shopping for centuries --- but online leaders only recently discovered social media. Advertising can interact with Social shopping in a positive way. Begging can interact with Social shopping in a positive way. Algorithms can steer customers to interactions with other customers, benefiting Social shopping. Algorithms can also steer your customers elsewhere. This is why it is so important to know your Net Google Score.

Organic customers buy from you because they are loyal to you. This is the most important customer to cultivate. Zappos and Amazon and Starbucks and Apple are loaded with organic customers, customers who buy from those brands regardless of Begging, Advertising, Algorithms, or Social opportunities.

From a Multichannel Forensics standpoint, we assign customers into one of these five classifications. We then measure the long-term evolution of customers, as the customer flows in and out of each classification, increasing or decreasing profitability.

We especially pay close attention to what happens when we migrate customers into the Begging segment ... do customers transfer into Begging, then stay isolated in Begging, not wanting to shop unless a discount or promotion is available? That's something we must closely monitor in 2009, given the fire sales we are having this fall to move inventory.

Multichannel Customers And Advertising

Please click on the image to enlarge it.

The most popular question asked of me is "When can I stop advertising to or reduce advertising to a segment of customers?"

Often, catalogers and retailers are unwilling to execute tests that will answer this question, opting instead to maximize short-term sales.

If you fall into that camp, try this:

• Identify any customer who purchased from your brand in the past twelve months, and purchased at least ten times (since the first purchase).

• Sum how many of the ten most recent purchases occurred, by channel.

• Graph the frequency of occurrence, using one channel as the x-axis.

In the image at the start of this post, there are three distributions.

• The U-Shaped Distribution occurs when your customer prefers shopping from one channel. Almost all customers buy from, say, the online channel, or the catalog channel, and tend to purchase in equal rates. When this happens, nearly half of your audience is eligible for a reduction in advertising, as nearly half of the audience buys online, and may be in the habit of shopping online without the need for catalogs.

• The Bell-Shaped Distribution occurs when customers swap back and forth, between channels, not exhibiting a preference for any one channel. In this case, you probably need to continue catalog advertising.

• The Skewed-Shape Distribution happens when customers generally shop in just one channel, and show a clear preference for just one channel. This frequently happens in online/retail situations, where customers inevitably shift their preference to the retail channel.

If you want to reduce advertising (catalog advertising in particular), you don't want to see the bell-shaped distribution. When customers switch back and forth between channels, they are likely to be semi-dependent upon catalog advertising. Instead, you want to see a lot of customers who are 100% or nearly 100% loyal to the online channel. This is the audience that could stomach a reduction in advertising, without adversely impacting the top line.

How Online Advertising Channels Fit Together

Please click on the image to enlarge it.

E-commerce provides us with the richest array of information ever available to direct marketers. Let's use the Multichannel Forensics framework to explore the interaction between advertising and customer purchases on your website.

Here's what you do.

Step 1: Create a "spreadsheet", if you will, with each row representing a customer who purchased in 2005. Each column in your spreadsheet is populated with a series of "1s" or "0s". Each column represents a form of advertising. A one is populated if the customer purchased in 2005 using that form of advertising. A zero is populated if the customer did not purchase in 2005 via that form of advertising.

A list of potential "advertising" channels include:

• Brand Buyers: Customers who purchased, and we cannot identify any advertising campaign that caused the customer to purchase. This is the most important "channel" ... we want our customers to purchase from us because they "love" us, not because we advertise to them.
  
• Catalog: Customers who use a catalog source code, or are identified as having purchased via catalog from a matchback analysis.
• E-Mail: Customers who clicked through an e-mail and purchased merchandise.
• Social Media: Customers who purchased and had a referring URL from a blog or social media site like MySpace or Facebook. Some companies comprise a list of the top 5,000 referring URLs, then visit each URL, and identify if the URL is a blog/social-media site. In addition, these companies look for URLs with phrases like "Blogspot" or "Typepad" in the URL, to identify that the referring URL is a blog.
  
• Portals: Customers who purchased and had a referring URL from an advertisement on a portal like MSN or Yahoo!.
• Paid Search: Customers who purchased and had a referring URL from a paid search ad.
• Natural Search: Customers who purchased and had a referring URL from a search engine, a URL that isn't from a paid search campaign.
• Affiliates: Customers who purchased via an established affiliate.
• Shopping Comparison: Customers who purchased via a shopping comparison site like MySimon.

Step 2: Repeat Step 1, but this time, create the file for calendar year 2006.

Step 3: Match the files together for Step 1 and Step 2.

Step 4: Conduct a Multichannel Forensics analysis.

Step 5: Create the map at the start of this post. Any channels that are in "Equilibrium" or "Transfer" are linked together on the map.

The analysis illustrates how your advertising channels work together.

In this example, there are numerous findings.

• New online customers are recruited through paid search, catalog marketing, and "brand buyers", customers who purchase without having used advertising.

• Shopping Comparison and Affiliate customers appear to be largely "cut off" from the ecosystem. In other words, these customers do not eventually become customers who purchase "on their own", or via other advertising channels. One would evaluate the LTV of these customers, to understand if these marketing activities are worthwhile.

• Social Media sites and Portals are leveraged by "brand buyers". In other words, best customers utilize these sites to make purchases.

• Paid Search and Natural Search become "links" that ultimately facilitate the transition from new customer to a loyal customer. The data in this example suggest that these customers can eventually become "brand buyers", customers who don't need advertising to purchase.

• Catalog Marketing is an important way to acquire new customers. The data also suggest that catalog customers use search to complement future purchase activities.

• E-Mail is a "dependent channel" in this example. E-Mail marketing might be effective, but it appears it depends upon other marketing activities (search, catalog) to "acquire" the e-mail address that will be used for e-mail marketing purposes. In this example, the link between "brand buyers" and e-mail is one way. Brand buyers offer an e-mail address, and then respond to e-mail. E-Mail customers are less likely to become "brand buyers", in this example.

Now that the advertising ecosystem has been fully mapped, you sit down with your marketing folks, and consider appropriate linkages between marketing programs.

It is also important to understand which advertising channels drive a customer toward loyalty. Notice that catalog, search and e-mail marketing are all interconnected, whereas affiliate and shopping comparison marketing activities deliver a fundamentally "different" customer to the brand (in this example ... your mileage may vary).

Television And Multichannel Forensics

Are you watching less television?

I sure am. During the past three years, the DVR fundamentally changed how I use my time.

This is fascinating, because the VCR existed for decades, but it didn't change my behavior. I recorded a handful of shows. Maybe I was attending a dinner, and would miss a new episode of Seinfeld. The VCR kept me current with programs I missed.

For me, three things happened that caused me to stop watching television.

• Television became awful. I'm not sure television was ever great, but it sure used to be better. Last night, I scanned two hundred channels before settling on a thirty year old episode of MASH that I viewed a hundred times previously. I personally think September 11 changed television. How many comedies are on television anymore? Life wasn't funny after September 11 ... at least that's my opinion. You have to balance the half-dozen CSI and Law and Order franchises with something that makes you laugh. Instead, television gave us "reality" programming, an inexpensive, unsustainable, short-term profit solution to comedy. Instead of improving the quality of the programming, television improved the quality of the image (HDTV).
  

• If television is awful, then there isn't a reason to sit in front of the television for three straight hours. To me, the DVR didn't represent a fundamental change, because the VCR always existed. The DVR coupled with awful television represented a fundamental change. When there are only a handful of quality choices (quality is defined differently by each individual), one can use the DVR to create a personalized "evening of television". For instance, this morning, I am going through a week of programs like "Charlie Rose", Dan Rather Reports, and CBS Sunday Morning. Television folks call this "time shifting". I call it "convenience". These days, I can watch "Lost" on Thursday evening, when both my wife and I are at home.
  

• Television folks say the DVR allows you to skip commercials, a bad thing for the health of the television industry. We've been able to skip commercials for three decades. Bad television coupled with technology changed consumer behavior. When the price of a DVR became affordable to the average consumer, a mass shift in behavior occurred. Cable companies, the very middlemen who connect viewers to programming, hastened this shift by offering low-cost or no-cost DVRs to consumers in response to competition. I purchased my first DVR three years ago, my first HD-DVR last year. Everything changed, especially after buying the HD-DVR.
  

Programming, advertising, cable/satellite providers, technology, the internet and pricing all interacted in a complex slurry, yielding the odd response to television we see today, an odd response ideally suited for a Multichannel Forensics analysis. Only seventeen percent of Americans own a DVR. What happens when fifty percent of Americans can afford a DVR?

Maybe most interesting is what I do with my extra time. I write books, I write this blog, I volunteer. Arguably, the quality of my life is better today than pre-DVR, thanks to the "quality" of television today. It is a zero-sum game.

Optimal Online Marketing Budget

We've previously discussed the importance of the "square root" rule in analyzing marketing campaigns, when solid test-based data is not available to the analyst.

Assume you spent $20,000 on an online marketing campaign, yielding $60,000 net sales, and a net loss of $2,000 (assuming 30% of sales flow-through to profit). You want to know what might have happened, had you spent more or less than $20,000.

Square Root Rule --- Assume you wanted to only spend $10,000. Sales will change by the following factor: ($10,000 / $20,000) ^ 0.5 = 0.707. Net Sales of $60,000 will change by 0.707, or $60,000 * 0.707 = $42,426. Profit = $42,426 * 0.30 - $10,000 = $2,728.

Again, if you don't have good test-based data to make comparisons with, use this rule as a quick shortcut.

The table below illustrates different spend levels, associated sales, and profit.

Spending Level	Net Sales	Estimated Profit
$10,000	$42,426	$2,728
$12,500	$47,434	$1,730
$15,000	$51,962	$588
$17,500	$56,125	($663)
$20,000	$60,000	($2,000)
$22,500	$63,640	($3,408)
$25,000	$67,082	($4,875)
$27,500	$70,356	($6,393)
$30,000	$73,485	($7,955)

Blue Nile, Chicos, PetSmart and NASCAR

A few tidbits from recent investor conference calls, courtesy of 123Jump.

At Blue Nile, management states that it measures market productivity by summing sales in a market (San Francisco), and dividing sales by the number of people living in the market. In markets like San Francisco, they claim to generate $2.00 sales per person. In rural markets, they claim to drive maybe $0.30 per person.

At Chicos, management discussed a change in advertising strategy, yet another one that benefits compiled databases, and spells doom for television and magazines. A quote, "the company was not able to quantify any real positive effect from television advertising". Management also states that coupons are the number one driver of sales for the company.

At PetSmart, the business generated 5.5% EBIT on sales of just over $3 billion dollars, through nine months. Management stated that they spend 2.4% of sales on marketing. That sounds small until you realize that, though nine months, the total is over $60 million dollars.

Ever wonder if NASCAR is a cash cow? International Motorsports, which owns many of the tracks NASCAR races at, earned $182.4 million in operating income on net sales of $544.9 million, through nine fiscal months. My goodness. How do I get a job there? That's simply stunning.

What Is More Important, Sales Increases Or Increased Profit?

In a recent comment, Ray S. asked if it is acceptable to have an advertising campaign that increases net sales, but does not increase profit.

Let me pose two scenarios to you.

Campaign Result, Scenario #1: Net Sales increase by $100,000 verses last year's campaign. Profit is flat, compared with last year's campaign.

Campaign Result, Scenario #2: Net Sales decrease by $100,000 verses last year's campaign. Profit increases by $25,000, compared with last year's campaign.

Ok, you are the CEO of this company. Which of the these two scenarios is preferable to you?

Coldwater Creek: The Little Engine That Could

Coldwater Creek provides a great example of how management transitioned a catalog company into a retail company with an online channel, and a catalog advertising arm.

After reading comments from Coldwater Creek's third quarter "conference call", it is obvious management continues to transition this business model into one dominated by the retail channel. During the third quarter, retail sales grew by 48%, and now represents 65% of the total business. Online sales grew by 29% from last year, and now represent 67% of the sales within the direct channel. Catalog now represents just over 10% of the total business. Wow.

Over the past six years, management shifted their corporate strategy. As the company began to invest in stores, management began to tear apart the tradtional catalog marketing strategy. During the call, management stated that the catalog now primarily drives traffic to the stores, not to the catalog channel, not to the online channel. More important is the comment that the internet is independent, and has other ways of using marketing to drive sales to the online channel.

Once again, we learn that the multichannel environment is a big ecosystem, one with inter-dependencies, and interactions that pundits do not understand very well.

In the case of Coldwater Creek, a brand was built via the catalog channel. As management shifted the focus of the business from a catalog company to a multichannel company, several things occured. Customers transferred out of the catalog channel to the online and retail channels. Catalog advertising drove business to the retail channel more than the online channel. The online channel has marketing channels that it can leverage to drive its own business, independent of catalog marketing. Magazine advertising, coupled with discounts and promotions, drive business to Coldwater Creek stores. The catalog/telephone-channel has been reduced to about 10% of the total business.

Executives at multichannel retailers would be well-served to study the evolution of the Coldwater Creek business model. I'm not suggesting your business will evolve exactly like the business evolved at Coldwater Creek. But it will probably evolve in a way that is different than the pundits tell us it will evolve.

Return on Investment Formulas In Multichannel Retailing

Let's talk about some of the equations that individuals use to measure advertising return on investment in the multichannel retailing industry.

Ad to Sales Ratio: This is one of the most frequently used equations. Assume you spent $10,000 on an online marketing campaign, and generated $50,000 net sales. The ad to sales ratio is calculated as ($10,000 / $50,000) = 20%. Obviously, the lower this percentage is, the better your advertising performed. Multichannel retailers compare advertising efforts against each other with this metric.

Sales per Ad Dollar: Some industry publications like to use this metric. In the above example, we simply calculate the inverse of the ad to sales ratio. ($50,000 / $10,000) = 5.00. In this case, you get five dollars of sales for every dollar of advertising spent. The higher the metric, the better your advertising performed. E-Mail pundits like to use this measure, since e-mail has virtually no cost, thereby insuring that it has a good "return on investment".

Cost per Order: Online marketers enjoy using this metric, one that is maybe the least effective metric of all. Assume that the $10,000 spent in our previous examples generated 400 orders. Cost per Order (sometimes labeled "CPA" for cost per acquisition) is ($10,000 / 400) = $25.00. Each advertising strategy is compared, with lower metrics preferred. This metric is highly skewed, because the metric doesn't account for how much was spent, per order.

Profit per Order: A more effective, but less-used metric, is profit per order. Let's assume that, in the example above, twenty-five percent of the sales generated are converted to profit. In this case, ($50,000 * 0.25 - $10,000) = $2,500 of profit is generated. Next, divide the $2,500 profit by 400 orders. This yields $6.25 profit per order. This is one of the better ROI measures, because all aspects of the profit equation, sales, margin, and marketing cost, are included. Better yet, this measure can be stacked-up against long-term value metrics. For instance, if a marketer loses $10.00 profit per order, but expects to get $50.00 lifetime value back, the marketer should invest in the marketing activity.

Internal Rate of Return: This metric is not frequently used, but reflects what happens if marketing dollars are continuously invested over the course of a year. In the Profit per Order equation, we netted $2,500 profit on an investment of $10,000. Let's assume that this marketing effort took place over a twenty-six week period of time. The internal rate of return is calculated as ($12,500 / $10,000) ^ (52 / 26) = (1.25 ^ 2) = 1.56. In other words, on an annual basis, this investment has a fifty-six percent interest rate. The interest rate can be compared against all other marketing activities (many of which have a different time window --- e-mail may have just seven days, for example).

Your turn! What return on investment metrics do you like to use to evaluate marketing activities at your company?