Lets face it, sales professionals like to sell based on what is in front of them. They don't like to look backwards. Yet looking back can provide great ideas for going forward. I'm talking about your data and what it is trying to tell you about your customers, who you should be selling to, where to look for growth and profitability. Financial services firms, and Wall Street in particular, while much maligned by those in the industrial world, know how to use data to drive profits. Sales and marketing professionals also need to learn how to use data to profitably grow sales. Instead of selling based on what you think, sell based on what you know!
This article will case study an industrial company (an outdoor picnic furniture manufacturer) as they identified the factors that predict profitable customers. They used this information in their sales and marketing activities to productively grow sales by pursuing prospects that were most likely to be highly profitable. Because this is a case study article, it is very specific to this one company. Of course what they learned is only applicable to this one company in this niche industry.
The problem this team solved was predicting which prospects are likely to be highly profitable customers. By understanding the attributes of a highly profitable customer, this manufacturing company could effectively target its marketing campaigns and resources. The marketing activities at this company are primarily mailing out printed catalogues to customers and to prospects via rented mailing lists. Targeting their marketing efforts would improve sales per mailing lists rental fees and catalogues mailed. In addition, they wanted to use the predictions from this analysis to re-assign existing sales representatives to higher profit potential prospects. The goal was to build a model that would help the company achieve greater profit and improve sales per marketing expenses.
Factors that May Predict Profitable Customers
The team brainstormed the variables that would help predict gross profit per customer. A key consideration for the selection of a variable was the availability of data, either within the company database or in the public domain of the web. The variables selected are as follows:
1. Key customer: Key customer was a binary variable (1 = key customer, 0 = not). Key customers were assigned by sales management. If a customer had $20,000 in sales in a single year or $5,000 per year over three years they were designated a key customer.
2. Whether the customer ordered product via the company's web site or only over the phone: Web Y/N was a binary variable (1 = order via web site, 0 = order only via phone).
3. Whether or not the customer was located in a coastal state: The company's outdoor furnishings are highly corrosion resistant. Therefore, management wondered if customers in coastal states, where outdoor furnishings are more prone to salt water corrosion, were the more profitable customers.
4. The sunshine percentage of the state the customer was located within: They also wondered the effect of sunshine, i.e. how sunny/rainy the location of the customer was on total profit. We used sunshine percent of the state as a proxy for sunniness of each customer's locale.
5. The number of employees of the customer: We used this as a proxy for customer size, wondering if larger or smaller companies were more profitable.
The team also wanted to know if income and population affected customer profit. The hypothesis would be that companies ordering from areas having higher median incomes and companies ordering from more populated areas purchased more products. We were able to acquire median income and population by zip code from the 2000 census. This data was matched to customer billing zip code.
6. The median income of the zip code that the customer was located within,
7. The population of the zip code that the customer was located within.
These variables were used in single and multiple-variable regression models to predict gross profit per customer.
Data Summary
Our original database had 50,505 customers. Here is some data about the customer base, to put the results of the analysis into perspective:
The average gross profit per customer was $1,333
The average orders per customer were 1.57
There were 284 key customers in our data set of 50,505 customers
13,641 customers ordered via the web
26,331 customers were located in coastal states
The least sunny state had sunshine percentage of 37.6%, the sunniest state had sunshine percentage of 84.5%, and the average was 60%
The fewest employees per customer was zero, the most was 11,000, and the average was 51.9; Note that a privately owned business for which the owner performed all of the work would have an employee count of zero
The lowest median income for a customer billing zip code was $2,499, the highest was $200,001 and the average was $47,714
The lowest population for a customer billing zip code was 5 people, the highest was 143,987 and the average was 24,818
Results
Five of the seven independent variables proved to be statistically significant. That means we are able to predict customer profitability based on these factors. Here are the results from our analysis:
Key Customer: A key customer will have an average gross profit $19,977 higher than non-key customers. This verified that key customers were being correctly chosen.
Web-Yes: Customers that use the web will have an average gross profit $741.94 higher than customers that call in orders
Coastal-Yes: Customers located in coastal states will have an average gross profit $332.83 higher than non-coastal state customers
Customer-Size: For each additional customer employee (again a proxy for the size of the customer) a customer will on average have gross profit $1.29 higher
Median-Income: For each $10,000 increase of median income in a customer's zip code a customer in that zip code will on average have $74 higher gross profit. Because the average profit per customer is $1,332, even $74 is economically meaningful. This is about a 5% increase in gross profit per customer for every $10,000 in median income of their zip code.
Conclusion
This analysis began by asking if the seven variables could predict profitable customers. Can this company use historical data to improve their sales and marketing strategy and profitably grow sales? The intent was to use this information to more effectively rent direct mail lists, choose customers and prospects to receive a mailed catalogue; and to determine which customers and prospects get assigned to a sales representative. Because of the statistical significance of the variables and this analysis overall, these four variables (customers that purchase on the web, customers that are located in coastal states, greater number employees at a customer and greater median income of the customers' billing zip code) will be used to select where to expend marketing and sales resources. Each variable predicts a more profitable customer.
By taking a step back, and analyzing their sales history, they learned to sell based on what they know, versus what they think. By looking backward they can now plan forward with greater confidence that they are not just shooting from the hip.
For more detailed information about this subject goto http://www.supplyvelocity.com/ for our white papers.
0 comments:
Post a Comment