1. The WHERE: The advent of the affordable desktop and laptop computer

2. The HOW: The creation of desktop analytics applications

3. The WHAT: A meteoric rise in the availability of data of all kinds

A February, 2012 report by the Aberdeen Group finds the following:

* Average marketing response rates for predictive analytics “leaders” are 67 percent higher than companies in earlier stages of implementation
* Customer retention rates also grew four times faster for these leaders

You can read more in this report by the Aberdeen Group: Divide and Conquer: Using Predictive Analytics to Segment, Target and Optimize Marketing

At MindEcology, we see essentially three ways to approach customer segmentation:

1. Simple segmentation: this is the easiest method, but also the most limited in terms of its potential applications. Simple segmentation involves grouping your current customers by total annual revenue (or some other metric) in order to differentiate between them in your ongoing marketing and CRM efforts. While this method is effective when marketing to your current customers, it does not help you reach out to new prospects since there is no link that can be derived between your segmentation model and those people in your trade area who have not bought from you yet.

2. Traditional demographic data-based segmentation: this method allows you to apply your model to people in your market other than your past/existing customers. It involves appending demographic data such as age, income, sex, etc. to your customer data. Traditional demographic segmentation represents a step up since it gives you the ability to directly target prospects in your market who meet your “best customer profile.”

3. Advanced segmentation: by layering in psychographic data with demographic and geographic data, you can potentially gain much greater lift or leverage from your segmentation work than you can with mere traditional demographic data-based segmentation alone. With advanced segmentation, you can find groupings of customers who are more likely to become your customer – and more likely to spend money with you once they do. Once you have it, this type of information is like money in your pocket. It allows you to only target your marketing and advertising toward those prospects who are 3-5 times more likely to buy from you than would be the average prospect.

The following (roughly 11-minute) video compares the merits of these three types of segmentation using a sample data set. If you like data mining, customer segmentation and marketing,you will enjoy this video. It presents a data-oriented argument for engaging in advanced segmentation techniques:

Video: Advanced Customer Segmentation

“Almost all the growth that’s available to you exists when you aren’t like most people and when you work had to appeal to folks who aren’t most people.”

(“Tribes,” Seth Godin, pp. 69)

There’s a whole lot of wisdom packed into Seth’s statement. At MindEcology, we say something similar: “You succeed when you narrow the focus.” We also say, “Trust the data.” The data points the way.

If your business is trying to be all things to all people, your company will have trouble hitting its true stride. There are two facets to the kind of hyper-targeting that we promote:

1. Skipping over households (B2C) and companies (B2B) in your marketing activities that do not match your company’s ideal customer profile, while directly blasting with intense focus those that do.

2. Tightening up your message around the needs, desires and buy-buttons of your best customers (and therefore, of your best prospects who are most likely to buy from you).

Both of these types of actions will drive your conversion rates sky-high because you are effectively reaching out and touching “not-most-people” in the most effective, targeted and convincing ways possible.

Do this correctly and they will be pulling their wallets out faster than you can fulfill the incoming orders. (And, that’s a very good problem to have).

Similarly, anybody working in business faces a difficult dilemma on a daily basis: we know that we MUST commit money and resources to promoting our business if we are to stay in business. On the other hand, we cannot be 100% certain that we are getting the highest-possible return on our investment on our marketing dollars. In other words: some of our marketing budget is likely being wasted each month.

Faced with this dilemma, most of us simply give into inertia, month after month. Meaning: we just keep doing what we have always done, even though we are far from certain that it is the best thing for our business.

But, giving into intertia is dangerous. Spending money on marketing without making sure we are getting the highest-possible return on our investment dollars is akin to gambling. So is doing no marketing at all. And yet, companies everywhere continue to treat their marketing decisions like a game of roulette, year after year.

The way out of this dilemma is to follow a data-driven decision-making approach to spending your marketing and advertising dollars. In fact, this is the only way to significantly reduce your decision-making uncertainty while moving toward ever-improving marketing ROI. Here are some steps to getting there:

1. Gather together all known marketing options (i.e., channels, media, methods, etc.) that are within your means to afford at this time.
2. Calculate the expected return on investment for each one. Remember, expected ROI is calculated as follows: (Expected Gain from Investment – Cost of Investment) / Cost of Investment
3. Choose one or more marketing options* as a test.
4. Execute your campaign on a modest scale – but make sure you invest enough resources in your campaign in order to yield enough data to gauge results later.
5. Push “pause” on your campaign and evaluate the results of each element you tested in terms of return on investment. This time, however, substitute “Actual Gain” for “Expected Gain” in the ROI equation.
6. Repeat steps #3 through #5.

* Note that you may need to test just one marketing option at a time in order to accurately track how each one affects your revenues.

By following these steps, you can move yourself from being a marketing gambler to being a marketer who is guided by an unwavering commitment to pursuing the best-possible marketing investment ROI.

Contrary to how many companies approach site selection, it needs to be treated as a “process” rather than just a gut-feeling-based, ad hoc decision.

Specifically, most site selection efforts miss the mark in one or more of the following five ways:

1. Places too much emphasis on luck or circumstance in discovering candidate locations for consideration. Example: A colleague says he knows a guy whose lease is expiring soon and suggests you consider taking it over. The chance of that location being your best choice is close to nil.

2. Takes into account site quality factors such as building appearance but ignores the demographic and psychographic makeup of the people who live in the immediate vicinity (the latter is especially important for companies selling directly to consumers).

3. Fails to take into account potential customer cannabilization of other nearby locations operated by the same company.

4. Is unable to appropriately sub-divide the target trade area into meaningful geographical units. Example: Comparing the merits of entire sections of the city (e.g., upper, east side) may be placing too much emphasis on areas that will not influence future business potential. On the other hand, comparing the areas within a 1/4 mile radius of each candidate location likely underestimates the importance of who lives in the area.

5. Does not adequately combine the various predictive success metrics numerically in a way that allows candidate locations to be scored and ranked in order of desirability. Only by reducing the potential success factors for a new location down to a single score can you easily make decisions about which location holds the most revenue potential for you.

Missing the mark in any of these five ways can mean the difference between choosing a star location with big revenue potential and a going with one that never lives up to your revenue expectations.

Do Your Research Before You Open A New Location

Indeed, if you are already asking yourself these types of questions, you are already ahead of the game: most business owners spend far too little time thinking about who their best customers are or how to reach them.

To answer this type of question, of course, you will need to collect some data about your customers. The types of customer data that yield the most insights are:

a. demographic: age, income, marital status, owns vs. rents, etc.
b. psychographic: attitudes and opinions about entertainment, the economy, politics, advertising, etc.
c. media consumption: website usage, TV watching habits, radio listening behaviors, etc.
d. behavioral: how often they buy from you, what they buy, how much money they spend with you, etc.
e. geographic: where they live and work

Once you have collected any of these types of data about your customers, you will next need to find a way of making sense of that data. There are essentially three approaches to putting your customer data into perspective: Descriptive Statistics, Grouping & Prioritizing, and Baseline Comparison. As you will see below, each method can be useful in terms of yielding insights into what kind of people buy from you. But, each method is exponentially more precise than the one before it in terms of yielding insights about whom you should be targeting in your marketing outreach and advertising efforts. Here is a brief overview of each approach:

1. Descriptive Statistics: This approach involves counting or averaging the data you have collected about your customers. For example, if you have collected income data on your customers, you might determine that their average income is $43,343. This is useful data, but it has a crucial limitation: it fails to recognize the likelihood that – taken individually – your customers represent a spectrum of income levels. In other words: the median tells the story of the group as a whole, but it does not yield meaningful information about specific customers.

2. Grouping & Prioritizing: This method entails grouping your customers by sub-category (for each specific category of data, such as income) and then looking for patterns in terms of the relative representation in each sub-category. Often, this approach also involves grouping each sub-category of data according to the relative percentage of the total that each one represents. For example, suppose that further analysis yielded the following income-related sub-categories of data about your customers:

35% have incomes of $20,000 to $30,000 per year
55% have incomes of $30,000 to $50,000 per year
10% have incomes of $50,000 or more per year

Knowing this, you are now in the position to adjust and focus your markeing efforts in order to reach out to those in the $30,000 to $50,000 per year sub-category, since they represent well over half of your customers.

3. Baseline Comparison: However, to find out in a much more precise fashion which customer groupings are the most signficant, you should introduce the concept of a “baseline” to your calculations. To do so, start by turning the data in each sub-category into percentages (as was done in the example in #2 above). Then, do the same for the baseline group (this could be a trade area, for example). Finally, perform a simple division calculation, whereby the numerator is the data from your customer list and the denominator is the data from the baseline population of your trade area as a whole. This simple calculation allows you to perform a very powerful feat: to create an index for each sub-category that determines which are the most well-represented (and therefore, significant), relative to your trade area.

To illustrate this point, and building upon the example in #2 above, if just 3% of your trade area (i.e., the baseline) has incomes in the $50,000 or more per year sub-category, then that income sub-category indexes at (10% / 3% x 100 =) 333. This means that prospects who are members of this income sub-category are 3.33 times more likely to become your customer, relative to the average prospect. Meanwhile, the $30,000 to $50,000 per year sub-category might only index at 75, meaning they are just 0.75 times as likely to become your customer.

Thus, by adding a baseline to your calculations, you have gained additional insights that may change your entire marketing focus – insights that you could not have garnered from merely performing the “Grouping & Prioritizing” work described above. After all, if you had stopped at #2 (Grouping & Prioritizing), you would have made a very different decision about which prospects to go after. Now (and only now) do you have actionable information that you can immediately put to use by focusing your marketing outreach and advertising efforts on this group.

This method becomes even more powerful when you apply it to not only demographic data (as in the above example) but to psychographic, media consumption, and other customer data as well.

One of the outcomes of market segmentation and targeting is that of profiling your best customers. Profiling can take on both quantitative and qualitative forms. An example of quantitative profiling is to divide your customers into groups by demographic or psychographic type, then ranking those types by index score. Meanwhile, qualitative profiling involves creating a detailed description of the habits, motivations, and buying behaviors of your best customers.

Engaging in segmentation, targeting and profiling of your best customers allows you to use that information to locate and reach out directly to those prospects who are most likely to respond to your offers. The result is markedly improved return on investment (ROI) for your marketing and advertising campaigns.

Why don’t more businesses take the time to profile their best customers? Here are 10 common objections that companies raise when faced with profiling their best customers, along with answers those objections:

1. “My customers are unique to any other company or industry niche, so they cannot be profiled”:

A: It is true that every company’s customer list is unique, like a fingerprint or a DNA profile. That is why profiling is so powerful; it converts your unique and specific customer list into a generalizable model that can be used to find more customers just like them.

2. “Too expensive; given the economy, I need to invest in the basics right now”:

A: Customer profiling often pays for itself with just a single marketing project by allowing companies to spend less money on advertising, garner a better response rate, or both. And, the research results are valid for up to a few years in most cases. Meaning: the research can pay for itself again and again.

3. “By only focusing my marketing on my best prospects, I will be missing opportunities to expand my market”:

A: Leveraging your profiling analysis involves placing most of your marketing resources into going after those prospects most likely to respond to your offers. However, you can always set aside a percentage of your resources to continue to market-test new segments as you see fit.

4. “Profiling my best customers is pointless since I already know what my best customers are like”:

A: There is little doubt that you already have a wealth of knowledge on your existing customers. However, unless you can turn this knowledge into a generalizable model that you can apply to prospects who have not bought from you yet, then most or all of the information you have on your existing customers will do you little good when it comes to finding new ones.

5. “My product or service appeals to everybody, so there is no need to profile them”:

A: Unless you are selling fresh water to castaways on a desert isle, it is just not true that everybody in your trade area needs your product equally. Prioritization of your prospects matters, and customer profiling is the best way to prioritize which prospects in your trade area to pursue.

6. “Profiling my best customers makes no sense: what I need is to find new customers, not market to my existing customers”:

A: Building a profile, or model, of your existing customers can certainly help you market more effectively to them. But, where customer profiles really come in handy is in helping you find loads of new prospects who are ready to buy from you – right now.

7. “My customer turnover (or churn) rate is fairly high, so yesterday’s best customers may not be tomorrow’s best customers”:

A: It is true for any business that customers come and go. However, the TYPES of customers who buy from you – your best customer profile – will change only very slowly. We recommend refreshing your profile every 2-5 years, depending upon your business type.

8. “Profiling my best existing customers is akin to looking backward in time; what I need is a forward-thinking solution to my marketing challenges”:

A: As Shakespeare said, “What’s past is prologue.” You stand to benefit from realizing that your historical customer data is gold – or it can be gold if you know how to use it properly. Gaining a detailed understanding of those who have bought from you over the past 2-3 years can yield a wealth of clues about how to find more customers like them in the near future.

9. “I don’t have much data on my existing customers”:

A: If your customers are consumers (householders) and you have records on the full home address (and, preferably, total sales revenue figures) for at least 500 to 1,000 of them, you can leverage hyper-targeting and customer profiling techniques to improve your marketing ROI.

Meanwhile, if you operate in the B2B space, all you need is a list of company names – and in some cases, the job titles of those whom you tend to work with – in order to engage in effective customer profiling. Note that the required database size for B2B profiling is much smaller than it is with B2C profiling, since different techniques are applied.

10. “I am not sure how to get data out of my database(s)”:

A: This is a common concern, since most business owners do not have a high comfort level concerning their data or how to access it. Fortunately, getting data out of your CRM database, point of sales (POS) system, or accounting software is not as difficult as you may think. Experts can guide you through the process, and exporting your data may take as little as a few minutes.

What does this mean? It means that, despite living in a world that is increasingly virtual for so many of us, most of us still care (and should still care) about where our businesses are located in geographical relation to our best customers and prospects. Yes, for many businesses today, there is no mistaking that place matters; geography matters.

At MindEcology, one of our jobs is to show our clients where their best prospects live and work. While we do our “best customer” analyses using data and numbers, we find that we are able to get our analytical findings across to customers most effectively when we use visual representation or “data visualization” for that data. One form of data visualization that we find most effective is the use of color-coded geographical heat maps.

As the name implies, a geographical heat map employs a color scale in order to differentiate between various attributes of a geographical area. Various colors (say: red, orange, yellow, green & blue) are assigned to different geographical units (sub-sections) within a trade area as a way of indicating the attractiveness of each sub-section of the trade area in question. For example, we may designate the color red as “hot” and blue as “not,” with the colors in between falling somewhere along that “hot-to-not” scale.

The two most commonly-used geographical units that we employ for our heat maps are zip codes and block groups. Most people are familar with zip codes – but less so with block groups.

Zip codes were created by the post office years ago as a way to mark off large areas of land for the purposes of sorting and delivering mail more efficiently. Meanwhile, block groups were created by the United States Census Bureau; they represent the smallest unit for which the bureau publishes statistical data on households.

These are two completely independent systems of geographical demarcation, so the block group and zip code boundaries for a given area rarely line up perfectly. In general, block groups are smaller than zip codes (although this is not always the case). On average, the average block group covers an area that is 3-5 times smaller than the average zip code.

So, which of these two systems of geographical units should we use for your heat map project? That depends, of course. In general, zip codes are best if you plan on implementing a direct mail, newspaper, radio, or billboard campaign and you want to know where your best prospects are. However, if you plan to apply the results of our analytics to helping you select the exact spot for your next store location, using block group-level coding may be able to give you better granularity in order to help you choose specific shopping centers or street corners on which to set up shop.

For example, the following two maps show the identical geography at the same zoom level – a location near Aiken, SC. Notice how the zip code view (upper) indicates only 3 zip codes, while the block group view (lower) shows at least 7 distinct block groups. You can see much more of a color variation (and therefore targeting granularity) with the block group view, but using block groups may be too granular if you are interested in considering this area solely as part of a market level analysis.

Two Views of the Same Geography Near Aiken, South Carolina: Zip Code vs. Block Group

Created by Matthew Lemke Productions

1. Companies always need more customers (customer quantity) and/or more revenue per transaction (customer quality)

2. At the same time, companies’ marketing (sales, advertising) resources are finite

One way to survive amidst these conflicting realities is through the practices of segmentation and targeting.

Segmentation and Targeting

In marketing speak, the two related practices of segmentation and targeting go hand-in-hand. Segmentation is the practice of differentiating your prospects before you approach them with marketing outreach in order to separate those most likely to buy from you (your “best prospects”) from those less likely. Meanwhile, targeting is the practice of allocating resources and doing planning in order to tailor your marketing to speak directly to the motivations and needs of your best prospects.

Given that you are most likely competing for the same prospects with one or several other companies, to be successful you need to continually find smarter, more innovative ways to improve your marketing return on investment. Otherwise, your competitors will beat you to it.

You only have a limited number of ways to improve marketing ROI, one of which is to make your marketing more efficient through better customer targeting.

The truth is, you are likely already doing targeting of some sort. This is common sense targeting. For example, if you have a swimming pool installation business that advertises via direct mail, you are likely not sending mail pieces to prospects who live in apartments.

The 4 Levels of Customer Targeting

The way we at MindEcology see it, there are actually 4 levels of targeting, as follows:

Level 1: Common sense targeting

Level 2: Demographics-based targeting: choosing prospects based upon age & income (B2C) or company revenue, geography, other factors (B2B)

Level 3: Proprietary clustering systems (e.g., Prizm, Mosaic) which leverage geo-demographics and psychographics

Level 4: Predictive analytics and data mining

Most companies who engage in targeting today engage in demographics-based (Level 2) targeting. While this is better than merely engaging in common sense (Level 1) targeting, it is limited at best. In other words, it leaves money on the table. However, Level 3 targeting is superior because it leverages your own historical sales data – in conjunction with sophisticated segmentation systems – to build a model of your best prospects. And, Level 4 takes targeting to another level of complexity by combining multiple variables into a more sophisticated model. It can build upon Level 3 data, as well as many other inputs, to create a sophisticated, hyper-targeted model. The result: a hyper-targeted model that allows you to execute your marketing campaign in a much more focused way that yields a better return on investment.

What do you need to get started on Level 3 and Level 4? Here are the required skill sets:

* Data cleansing and organization
* Informed understanding of potentially relevant variables to be selected for further testing
* Understanding of which segmentation systems to leverage (for B2C)
* Sourcing business lists (for B2B)
* Statistical correlation and regression (bivariate and multivariate)
* Artificial neural network creation and data mining
* Knowing how to apply results to actual marketing needs

Setting your sights on Level 3 and Level 4 targeting is a smart, efficient and entirely workable way to improve your marketing ROI.