**Version Notice:** This article covers features in our **r9/IS Pro** platform. If you're looking for information on this topic related to r8, see TURF Analysis.

The **TURF Analysis** applet allows users to score checkbox questions and MaxDiff exercises to identify the best *combination* of products, flavors, or other varieties, not just the best-performing individual ones.

## What is TURF?

**TURF** stands for "Total Unduplicated Reach and Frequency." Originating from the media planning world, but since having been adopted by the market research industry, the initial goal of TURF was two-fold – to maximize *reach* (the percent of the target audience that sees at least one ad) and to maximize *frequency* (the average number of exposures, or number of times, an ad is seen by a member of the target audience).

In market research, the focus is more on maximizing the "reach" of a product line (i.e., what percent of customers would buy *any *of the items in the line), especially in the case of attempting to expand the options in a particular product line. In this situation, TURF examines how adding one or more new product options can increase the number of potential customers "reached" by the product line, assuming that not all of the newly proposed options will be kept. Frequency, i.e., the average number of portfolio items customers would buy, is often used as a tiebreaker between combinations with the same reach.

As an example, think of ice cream flavors. An ice cream company can only offer so many flavors based on production, shelf space, etc.. If they conduct a survey that asks potential customers what flavors they purchase most often, they could potentially have results like this:

Just looking at the percentages, you might recommend Item 1, Item 6, and Item 5 (with Item 8 as the fourth, if needed).

But what if those top three items are Chocolate, Chocolate Chip Cookie Dough, and Cookies & Cream? If a company chose those three flavors, they wouldn’t be able to sell any ice cream to people who don’t like chocolate, have chocolate allergies, love other flavors, etc. While chocolate lovers will have multiple options, they won’t necessarily purchase *more* ice cream to make up for the consumers that are alienated by the "choco-centric" assortment.

Instead, a TURF analysis identifies what *combination* of flavors maximizes reach, i.e., how to get the most people to buy *any* of your options. Results would likely recommend an assortment of flavors to fit different consumers’ preferences.

Although there is no technical minimum to run a TURF analysis so long as there is more than one complete, it is recommended to have a minimum of 100 respondents for a TURF analysis (the more, the better). Also, it is best to run any analysis after data collection is complete.

## When to perform a TURF analysis

TURF might be the right analysis when you hear these key terms:

**Combination**– When clients are interested not just in the highest performing option(s), but those that work best together in market/on shelf. Note that not all combination questions are a fit for TURF, though; for example, TURF is often not recommended for marketing messages or features within a single product (see below).**Portfolio Strategy**– TURF use cases include situations where product assortment is constrained by*shelf space*,*portfolio rationalization*(i.e., removing underperforming portfolio items or varieties), or*line extensions*(e.g., which varieties to add to existing products in market). TURF can answer both*how many*varieties to go to market with (i.e., at what number reach starts to plateau), as well as*which ones*make the optimal combination. It can also reveal the*incremental reach*of introducing each new variety (i.e., the additional share your brand will attract beyond what it already owns).**Cannibalization**– TURF can be used to determine whether a new product or line extension will attract new customers or just steal consumers from your own current products.

When to potentially *not* use TURF:

**Marketing messages**– While TURF originated in the marketing space, as it is currently used, it may not be a fit for marketing use cases. If messages cannot be targeted to the appropriate consumers, they may potentially see more than one message identified by TURF analysis, including one(s) that do not resonate with them personally, potentially counteracting their preferred message(s).**Features within a product**– When consumers purchase a product, they are purchasing all the features – they cannot pick and choose the way they can with flavors or varieties. Thus, if a product includes lower-performing features, it could alienate consumers and discourage purchase rather than attracting different consumers who prefer different features. This is a better use case for*conjoint analysis*.

## How to perform a TURF analysis

To access the **TURF Analysis** applet, select the **TURF Analysis **option in the **Analytics **applet group.

A TURF analysis can be performed on multi-select questions/variables and MaxDiff utility score variables. Only one question can be analyzed at a time.

For MaxDiff exercises, the TURF analysis is performed on the MaxDiff's utility scores. Before running a TURF analysis, you will first need to score the MaxDiff using the **MCMC Scoring** applet so that the utility scores are present in the survey data.

For multi-select questions, the process is the same, except the "Convert utility scores" step is skipped. After selecting the records and field, you only need to define the portfolio sizes to include in the analysis.

**Note:** When analyzing a MaxDiff, be sure to score the MaxDiff exercise before conducting a TURF analysis.

### Select records and field to analyze

First, select which records to include in the analysis, and which field to analyze. The **Select records** options function similarly to the **Report Builder** used for reporting applets.

Next, select a field to analyze using the **Field** dropdown menu. Only fields eligible for TURF analysis will be listed.

After specifying which records and field to include, click **Next** to continue.

#### Records with missing data

If any records are missing data from the selected field, you may choose to either exclude them from analysis (default) or assign a '0' for the missing values.

Whether or not to exclude these records depends on the wording of the question and survey logic. For some, blanks mean "not seen," but for others, blanks mean "did not select this item based on previous survey answers," so the researcher can make this decision based on their interpretation of the data.

After making a selection, click **Next** to continue.

### Convert utility scores

**Note:** This step is only necessary for analyzing MaxDiff exercises. For checkbox questions, proceed to Define portfolio sizes below.

If you are analyzing a MaxDiff, the next step is to select a **Reach Assessment Technique**. The technique chosen determines how the items are counted as "reached" based on the MaxDiff utility scores. Options include *Threshold*, *First Choice*, or *Top Two* and are detailed below.

#### Threshold

The Threshold reach assessment technique recodes MaxDiff utility scores into binary values, with those above the threshold counting as "1", or "reached," and below the threshold as "0", or "not reached." The calculations will depend on the threshold that you enter.

Enter a threshold in the **Threshold** field, then click **Compute**. A **Recoded** table will appear that indicates how many respondents were "reached" by each item using that threshold value.

If you're unsatisfied with the threshold, enter a new value, and click **Compute** again to recalculate the **Recoded** table. You may recalculate the threshold as many times as needed. Once you're satisfied with the threshold, click **Next** to continue, or click **Back** to choose a different reach assessment technique.

Try to find a threshold where reach numbers are useful. A good starting point for a threshold is 2-3 times the average utility (i.e., if utility were evenly divided among all options). If the reach numbers are too high or low, (e.g., the **Reached N** is >98% or <5% of the **N** for each item), there is little to differentiate results, and the numbers will not reflect actual market behaviors (e.g., more than 5% of respondents probably buy ice cream).

Note that a threshold above the maximum for too many items could result in not enough items being counted as "reached" for analysis. The **Reached N** must be at least 1 for at least one item in the **Recoded** table, or the TURF analysis can't be run.

**Tip!** As a starting point, common practice suggests 2-3x average utility, i.e., (100/n items) * 2 or 3.

#### First Choice

First Choice only counts an item as "reached" for a given respondent if the item has the highest utility score. This approach makes less efficient use of the granular preference information provided by a MaxDiff, so small changes to the data have relatively more impact on the rank-order of the outcomes. However, this method may be the easiest to interpret and describe compared to the MaxDiff utility scores.

Review the **Recoded** table, which indicates how many respondents were "reached" by each item using this criteria. Then, click **Next** to continue with this calculation, or click **Back** to choose a different reach assessment technique.

#### Top Two

Top Two counts an item as "reached" if it has one of the given respondent's two highest utility scores, providing more granularity than First Choice.

Review the **Recoded** table, which indicates how many respondents were "reached" by each item using this criteria. Then, click **Next** to continue with this calculation, or click **Back** to choose a different reach assessment technique.

### Define portfolio sizes

On the next screen, enter a name to use for the Excel output's filename. Then, enter the number of items to include in each portfolio. *Portfolios* are groupings of items which TURF is asked to analyze in order to maximize the reach within the population.

Enter a single number (e.g., "3") to run the analysis for one portfolio size. Enter a range of numbers in the format "2-3" or "2:3" to run the analysis for each portfolio size in that range. For example, if the range "3-5" is specified, reach will be separately calculated for a portfolio of 3 items, 4 items, and 5 items.

Note that the minimum portfolio size is 2, and the maximum is one less than the total number of items. If you want the reach of 1 item, you can simply refer to its percentage in a Frequency report.

When ready, click **Next** to run the analysis.

### Interpret the results

After processing, you will see the results of the TURF analysis. The data and composition of the top portfolio for each of the selected sizes are displayed in a table.

The following columns are included in the table:

Column | Description |

Portfolio |
Number of items in the portfolio. |

Reach |
Percentage of respondents who chose any of the items included in the portfolio. |

Reach Δ (delta) |
The difference in reach between that portfolio and the one listed above it. |

Freq |
The average number of items in that portfolio that were chosen by respondents. |

Items |
A column is shown for each item in the dataset. A '1' indicates that the item is included in that portfolio. |

In the example above, the best reach for a portfolio of 3 items is 0.93, meaning 93% of respondents chose at least one item in that portfolio. That portfolio includes Items 4, 5, and 9. Comparing Portfolio 3 to Portfolios 4 and 5, you can see that the reach and frequency increase as the portfolio size increases.

From here, click the **Back** button to return to the previous step, or click **Download** to download an Excel export with more detailed results.

**Tip!** The Excel download includes a key mapping the option text to the item numbers.

#### Download to Excel

The Excel file includes a summary tab and additional tabs for each portfolio size analyzed. The tabs for each portfolio size list all combinations of items and their respective reaches and frequencies. Note that calculations can take time to process if there is a large number of items or combinations.

The Summary tab of the Excel output provides details on the analysis, including a key mapping the item numbers to option text. It also provides the results table from the platform, showing the "winning" combinations for each portfolio size.

The Excel output also includes tabs for each portfolio size included in the analysis. These tabs list every combination of items in order of highest reach followed by highest frequency. Reach is rounded to the nearest hundredth decimal point (.01) for reporting purposes; however, the full decimal string is considered when ordering the combinations from greatest to least.

Examples of each tab are shown below. The combination in the first row should match the "winning" combination on the Summary tab. Note that the **Combo** column simply assigns each combination a number for reference and does not factor into the results.

**Portfolio of 3 Items in detail (first 7 rows):**

**Portfolio of 4 Items in detail (first 7 rows):**

**Portfolio of 5 Items in detail (first 7 rows):**

## What else TURF analysis involves

Clients may use TURF outputs in different ways. First of all, clients might have parameters for their portfolio; for instance, they must retain two existing varieties. This is also called "forcing" or "pinning" items to a portfolio. This can be easily accomplished with Excel filters.

One important question might be to identify optimal portfolio size, i.e., the level at which reach plateaus. Plotting the data from the Summary tab of the Excel output should show the portfolio size where incremental reach is too low to justify an additional variety. While there is no clear threshold, typically you might see an inflection point where reach delta falls into the low single digits. In the example below, we might recommend a bundle size of 4, since adding a fifth item only adds ~3% reach.

## Comments

0 comments

Please sign in to leave a comment.