How can TURF analysis optimize my product line?

Let’s say you’re trying to sell some ice cream.

You’re pretty good at inventing ice cream flavors, so you’ve come up with several possible options. Now, let’s assume due to practical considerations you’re limited to selling only four flavors at a time. How should you decide which four ice cream flavors to sell?

In the real world, you can’t produce every product concept at the same time, and selecting the wrong variants can be a costly mistake. Produce too many products, and you’re throwing money at the wind. Produce too few, and you’re letting valuable customers walk away.

Total Unduplicated Reach and Frequency analysis, or TURF analysis, is a survey method developed to answer this question. Two measures are used to select product concepts to be included in the final product line:

  1. Reach. The number of consumers who would be interested in ANY of the products in your final product line.
  2. Frequency. The number of consumers who would be interested in EACH product.

How TURF Works

Let’s go back to the ice cream example. Imagine you’ve surveyed 100 local ice cream consumers and gotten the following results:

Flavor Number of Consumers Who Would Purchase
Chocolate Brownie 80
Chocolate Cookie Explosion 70
Extreme Vanilla 50
Coffee Crunch 40
Mango Madness 15


These numbers are the frequencies for each flavor. To keep things simple, we’re going to assume that all 100 people you surveyed would order least one of your ice cream flavors.

Remember, you’re only able to sell four flavors at once – one of these ice cream flavors won’t make the cut. So which four do you pick?

Your first instinct might be to pick the four ice cream flavors liked by the most people: chocolate brownie, chocolate cookie explosion, extreme vanilla, and coffee crunch. However, this oversimplifies the problem and ignores something crucial.

The same ice cream consumer may enjoy multiple flavors of ice cream. Because chocolate brownie and chocolate cookie explosion are so similar, many of the consumers who liked one flavor would be equally as happy eating the other. On the other hand, some consumers who like mango madness may not enjoy any of the non-fruit flavors. In short, consumers who prefer fruit-flavored ice creams are not being reached!

If you include two chocolate flavors and no fruit flavor, the two chocolate flavors will be competing with each other while the mango ice cream consumers will leave your shop unsatisfied.

There are 5 combinations of four flavors we can make from these ice cream flavors. Let’s look at the reach of two flavor combos:

Combination Number of Consumers Who Would Purchase
Chocolate Brownie, Chocolate Cookie, Explosion, Extreme Vanilla, Coffee Crunch 80
Chocolate Brownie, Extreme Vanilla, Coffee Crunch, Mango Madness 90


While more people liked chocolate cookie explosion than mango madness, replacing it with mango in our final flavor lineup increases the number of consumers who would be satisfied with one of your shop’s flavor offerings.

The number of mango ice cream likers may be relatively small, but leaving mango off the menu in favor of a redundant second chocolate offering will reduce your potential customer base by providing no acceptable options for these consumers. More people may like chocolate cookie explosion, but it is not the ideal choice for one of your four flavors.

TURF analysis can also help you when you don’t know how many options to include in your final set. If we were unsure whether our ice cream shop should sell 4, 3, or 2 flavors at once, we could compare the reach of each bundle of each size and determine the optimal number of flavors.


Conducting a TURF Analysis

So how can you go about collecting data for a TURF analysis?

TURF analysis is an online survey methodology, meaning your respondents will typically answer questions about product concepts. If you would prefer your participants to interact with the product beforehand, it is possible to combine TURF analysis with other consumer research techniques such as central location or home usage testing (although this is more expensive than a survey alone).

You can easily apply TURF analysis to data collected from standardized question types.

Some question types compatible with TURF analysis include:

  1. Liking or purchase intent scales
  2. Maximum Difference Scaling (MaxDiff) exercises
  3. Check all that apply (CATA) questions

When Miaoilis and associates first proposed TURF analysis in 1990, the only data analysis method available was exhaustive enumeration. If using this method, you calculate the total reach of all the possible product combinations and then select the best one.

That’s fine if you’re working with just a few product concepts. But what if you needed to select the best 10 out of 50 flavors? Now you have 10,272,278,170 possible combinations. That analysis is going to take forever.

Today, computer programs find solutions in seconds using validated shortcuts. These methods do not guarantee a single, optimal solution. Instead, they used estimation techniques to find a solution that is “good enough” among millions of options.

Efficient data collection and analysis now makes TURF a cost-effective solution to getting actionable results quickly.


What questions can I answer with a TURF analysis?

Here are a few examples of some business questions a TURF analysis can answer. At the SSC, we are here to help you select the right analysis for your needs. Don’t hesitate to contact us if you’re wondering how TURF analysis might fit in with your next project.

  • Which flavors should I sell in my line extension?
  • How many menu items should I sell to offer something for everyone?

For more information about TURF analysis, we recommend the following technical reports from Daniel and John Ennis at the Institute for Perception.

  • “Large TURF Problems: Finding Custom Solutions” (2017).
  • “eTURF 2.0: From Astronomical Numbers of Portfolios to a Single Optimum” (2016).

Or check out the original: “TURF: a new planning approach for product line extensions” by Miaoulis, Free and Parsons in the Journal of Marketing Research (1990). It’s a classic.

This post was written by SSC graduate student Angelina Schiano.