TURF Analysis

In brief: Suppose a store sells different flavors of ice cream. To cut costs, it wants to sell only 5 flavors, but its supplier offers 30. Which 5 flavors should it choose for greatest sales?

One strategy is to choose a product line that maximizes the likelihood that customers will find at least one very appealing flavor among the 5 on offer. With at least one such flavor available, a customer will probably buy some ice cream if in the mood.

Suppose that 70% of customers find at least one flavor very appealing in the set of 5 flavors that make up product line A, but only 50% do in product line B. If so, for every 100 customers in the mood to buy ice cream, 20 more of them would be likely to buy some if product line A were in place. We say that the first product line has greater reach, because it extends to a larger segment of consumers in terms of sales. Reach is a useful measure of product line utility.

TURF (Total Unduplicated Reach and Frequency) analysis is the marketing research technique of choice to measure reach. TURF analysis is good for (a) finding the best product lines for sales and (b) deciding how to best extend existing product lines. By “product lines” is meant not just different products, but different flavors of the same product, or different colors, sizes, and so on.

TURF yields easy-to-understand statistics that make it clear which product lines do best and by how much.

Continuing with our ice cream example, suppose the store knows from an earlier survey which flavors its ice cream-purchasing customers would buy if available. The table below is a partial list, showing the top 4 flavors:

Products are made up of combinations of features (e.g., vanilla, $5.00) from different attributes (e.g., flavor, price). Consider laptop computers. A simplified set of attributes is screen size, weight, and price. Each attribute, in turn, might have these features:

Flavor Would buy
  vanilla   60%
  chocolate   40%
  raspberry   24%
  mint   20%

First, note that the percents of just the top 4 add up to way more than 100%. That is, customers buying one flavor may well buy other flavors, too.

Suppose we can only have 2 flavors in stock. Would vanilla (60%) and chocolate (40%) be the best two to have? Do they together reach a 100% of customers? Not necessarily at all, and most probably not. It could be, for example, that all chocolate buyers are also vanilla buyers. The store could make sales to them by just stocking vanilla. So, if the store has only vanilla, or if it has both vanilla and chocolate, it can expect to reach (sell to) at most only 60% of ice cream-buying customers.

Now suppose that mint buyers (20%) don’t buy vanilla, and vanilla buyers don’t buy mint. Then, by adding mint to the product line, the potential reach of customers does increase – from 60% with just vanilla to 80% with both vanilla and mint.

So, the principle here is to have products in the product line that do not “duplicate” each other. Chocolate duplicates what vanilla already does, but mint adds unduplicated sales.

TURF analysis examines all different combinations of n products, taken r at a time, to form particular product lines, and finds the best combinations (i.e., product lines) in terms of reach. In the ice cream example above, we had n = 30 flavors taken r = 5 at a time. This yields 142,506 combinations or product lines. We could obviously not evaluate all these by hand, but TURF does the analysis quickly and easily. Its output typically gives, for example, the top 10 combinations (product lines).

Here’s how a TURF analysis might be done in our ice cream example. A survey is done on 500 consumers. They are asked to rate all 30 flavors in terms of whether they would buy each flavor or not. For each combination of 5 flavors, TURF counts the number of consumers who would buy at least one of the flavors in that combination. The number who would is the total reach score for that combination. After getting the reach scores for all combinations, TURF then typically prints out the top 10 combinations in rank order. The business can then decide, based on these results along with other considerations, which product line to implement.

Now, it could be that one consumer would not buy any of the 5 products in a particular combination, a second consumer would buy only 1 of the products, and a third would buy 4 of the products in that combination. The reach score for the first consumer would be 0, but it would be 1 for both the 2nd and 3rd consumers. This is because “reach” is scored as 0 (would buy none in the product line) or 1 (would buy at least 1 in the product line). But we may also be interested in scoring how many products in a product line consumers would buy. The number is called the frequency. In the example above, the frequency for the first consumer is 0, for the 2nd is 1, and for the 3rd is 4. Just considering these 3 consumers, their total reach score for this particular combination of 5 products is 2 (0 + 1 + 1), and their total frequency score is 5 (0 + 1 + 4).

In TURF, the reach and frequency scores for each combination are provided. The partial table below, based on surveying 500 consumers, illustrates. It shows the best 4 of the many thousands of combinations:

TABLE 2

Product line Reach Frequency
  1   220 542
  2   220 479
  3   215 450
  4   199 387

The reach scores are used to rank-order the product lines. Here, a reach score of 220 was the best, held by product lines 1 and 2. But their frequencies were different: 542 versus 479. So, the frequencies here can be used as a tie-break, making product line 1 the winner.

Example 1 (finding the best product line): A bookstore is opening up and wants to include 10 magazines. Based on recommendations, it considers 50 different magazines that may appeal to its market base. It does a survey to see which 10 magazines it should include. The goal is to maximize reach–to get as many customers as possible to buy at least one of its magazine offerings.

In the survey, 100 consumers who visit the bookstore are approached. They are asked to rate each of 50 magazines in terms of how likely they would be to purchase it. They are given the following scale to indicate this likelihood: 1 = very unlikely; 2 = unlikely; 3 = not sure; 4 = likely; 5 = very likely.

After collecting all the data, a TURF analysis is performed. For a given magazine, consumers checking off 4 or 5 on the purchase-likelihood scale are considered likely buyers and are given a reach score of 1; those checking off 1, 2, or 3 are given a reach score of 0. TURF is asked to report on the best 5 combinations. The results are shown in the table below:

TABLE 3

table-3

  Products (identified by their numbers: 1 to 50) in the product line
 Product line  Reach  Frequency 1 2 3 4 5 6 7 8 9 10
1 72 121 4 33 8 49 1 22 42 5 37 14
2 66 109 31 5 14 37 22 16 33 45 7 12
3 68 99 33 22 31 14 18 38 3 37 5 39
4 59 78 45 14 5 16 37 31 24 36 22 38
5 55 69 37 7 22 31 38 14 33 5 12 2

The best combination (product line) has a total reach score of 72, which means that 72 of the 100 consumers (72%) would probably buy at least one of the magazines in this combination. These magazines include those identified by #4, #33, and so on. The bookstore, however, decides that magazine #33 is too difficult to reliably get in time for sales, so it looks at the 2nd ranked product line. Here, there are two of them, each with a reach of 68. Since product line 2 has a higher frequency than product line 3 (109 versus 99), the bookstore uses frequency as a tie-breaker. It decides to offer product line 2 to its customers and goes ahead and orders magazines #31, #5, and so on.

Example 2 (finding the best product line extension): The bookstore wants to carry 4 particular magazines regardless of reach. But it also wants to add 6 more magazines so as to maximize reach above and beyond the first 4. We can think of the first 4 as its current product line and the additional 6 as its product line extension.

So, it does a survey of 100 consumers who visit the store. Again, it asks them to rate all 50 magazines, as in the first example, in terms of the likelihood that they would purchase each of them.

TURF analysis is then performed on the data. The bookstore wanted to include, regardless of reach, magazines #5, 14, 28, and 48. So, in the table below, the best 5 combinations in terms of reach are listed, with the constraint that each combination must have each of these 4 magazines.

TABLE 4

table-4

  Products (identified by their numbers: 1 to 50) in the product line
 Product line  Reach  Frequency 1 2 3 4 5 6 7 8 9 10
1 63 133 5 14 37 22 16 48 42 45 7 12
2 52 90 48 22 28 14 18 38 3 37 5 39
3 51 60 5 48 42 28 27 21 9 14 15 50
4 40 55 46 14 28 11 34 27 30 5 48 33
5 35 51 34 50 27 5 48 15 14 28 46 11

The 5 best product lines are shown, which all include the required 4 magazines. In the first product line, in addition to the 4 required magazines, magazines #37, #22, and so on are added. This combination gives a total reach of 63, meaning that, for every 100 customers wanting to buy a magazine, 63 on average should find one to buy. Since this combination has the highest reach, the bookstore decides to adopt it as its extended product line.

Noteworthy is that TURF can give incremental changes in product extensions. For example, TURF could be done on all product lines with 4 magazines. When done, suppose that magazines #5, 14, 28, and 48 (i.e., the ones the bookstore wants to have no matter what) have a total reach of 25. Then we can see that the first product line in the table above improves the reach from 25 to 63, a 152% increment.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: