What is TURF Analysis and When to Use It?
##### Published on 18 November 2019.
TURF Analysis Simulator
TURF

Conduct automated TURF analysis on results of any Conjointly experiment (or an outside dataset) using this user-friendly TURF analysis tool.

TURF analysis (Total Unduplicated Reach and Frequency) is a statistical technique that ranks combinations of products by how many people will like these combinations.

TURF
Range optimisation
Features and claims

### TURF Analysis Simulator

Conduct automated TURF analysis on results of any Conjointly experiment (or an outside dataset) using this user-friendly TURF analysis tool.

TURF stands for Total Unduplicated Reach and Frequency. It is a technique that came into prominence during the 1950s in the space of media planning. In this article, we will focus on the “reach” component of this technique.

Reach is the percentage of respondents for whom at least one of the claims in a particular combination is their most preferred claim. That is, it is a measure of how many respondents can be “activated” by a combination of claims (or products with certain claims).

Conjointly’s user-friendly TURF Analysis Simulator is an automated tool that helps automatically find top claim combinations. It can be used for product claim selection, range optimisation, and other applications (both using data from Conjointly experiments and from other sources.

## A simple example of TURF analysis

Imagine you are launching a new brand of πΉ vegetable juices. As you are preparing for the launch, you want to have a range of flavours that will appeal to (“reach”) the largest number of potential customers. But what flavours should you offer if your budget allows only two flavours?

Let’s start by listing all possibilities:

• π₯ potato
• π₯ carrot
• π½ corn
• πΆ pepper
• π₯ cucumber
• π₯¦ broccoli
• π mushroom
• π° chestnut

What do you do? One way to solve this problem is to run a Product Variant Selector study, which will help you rank flavours by consumer preference. The same study will also give you individual-level preferences, such as:

Respondent IDπ₯π₯π₯π½πΆπ₯π₯¦ππ°
14.91.50.6-0.81.21.83.9-8.54.0
24.50.8-1.30.1-0.21.73.8-8.42.0
35.5-1.00.3-0.31.40.73.6-7.32.9
45.7-1.2-1.06.00.00.15.2-8.93.1
54.3-1.5-0.3-1.5-0.80.85.9-7.83.7
63.20.8-1.0-0.20.20.23.7-8.53.5
73.10.21.2-0.2-0.90.53.1-7.51.5
85.10.5-0.1-1.4-1.40.33.1-6.53.8
94.60.9-0.9-1.41.00.13.4-8.01.2
103.20.90.5-1.00.40.33.6-7.81.7
113.4-0.5-1.45.00.21.55.8-7.92.2
124.8-0.90.50.3-1.02.93.0-7.23.7

Each cell shows the partworth utility of a certain flavour for a particular respondent. We can then assume that if a particular flavour is among the top two most liked by a person, then we call it appealing to them β. Therefore, these scores can be used to identify which flavour will be the most or the second most liked by each respondent:

Respondent IDπ₯π₯π₯π½πΆπ₯π₯¦ππ°
1ββ
2ββ
3ββ
4ββ
5ββ
6ββ
7ββ
8ββ
9ββ
10ββ
11ββ
12ββ
β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦

Next, we assemble several possible combinations of flavours and calculate a couple of metrics:

• (Unduplicated) Reach, the percentage of people for whom at least one of the flavours is appealing.
• Frequency, the average number of appealing flavours per respondent.
Respondent IDπ₯ + π½π₯ + π₯¦π₯ + π°π½ + π₯¦π½ + π°π₯¦ + π°
1ββββββ
2ββββββ
3ββββββ
4ββββββ
5ββββββ
6ββββββ
7ββββββ
8ββββββ
9ββββββ
10ββββββ
11ββββββ
12ββββββ
β¦β¦β¦β¦β¦β¦β¦
Reach1112119611
Reach %92%100%92%75%50%92%
Frequency1.11.51.31.11.01.1

As you see, everyone likes at least one of π₯ avocado + π₯¦ broccoli (Reach = 100%). This combination is a winner!

If you have more budget to launch three combinations, you can do the same analysis with three-way combinations:

Respondent IDπ₯ + π½ + π₯¦π₯ + π½ + π°π₯ + π₯¦ + π°π½ + π₯¦ + π°
1ββββββ
2ββββββ
3ββββββ
4ββββββ
5ββββββ
6ββββββ
7ββββββ
8ββββββ
9ββββββ
10ββββββ
11ββββββ
12ββββββ
β¦β¦β¦β¦
Reach12121212
Reach %100%100%100%100%
Frequency1.71.31.81.2

This time, all combinations have equally good reach. Now you need to look at frequency. If you offer the combination of π₯ avocado + π₯¦ broccoli + π° chestnut, for an average consumer, there will 1.8 liked flavours from your brand. This is the way to go.

## Deciding number of SKUs to launch through TURF

TURF analysis can be used to determine an efficient number of SKUs to launch for your range. For example, based on your internal finance calculations, you can determine that you must sell at least 100,000 items for any single SKU to recoup advertising investment, and that you generally want to have a only a handful of SKUs to avoid destocking by your retail channels.

Say, your total market is 2M units a year and you can sell 1 unit per consumer a year. That means, that any new SKU launched must reach 5% of the market.

Now, we need to model a few scenarios:

• What is the reach if you only launch 1 SKU with the best potential?
• What is the reach of the additional SKU if you only launch 2 SKU with the best potential?
• What is the reach of the third SKU if you launch 3 SKU with the best potential?
• What is the reach of the fourth SKU if you launch 4 SKU with the best potential?
• And so on.

This type of analysis is call the TURF ladder:

In this example, it is efficient to launch six SKUs because the additional share of the seventh SKU is below 5%. Adding it means you will not recoup advertising investment.

This analysis is most helpful for most FMCG/CPG situations because:

• Companies have fixed costs for produce and promote an SKU
• Companies generally favour lower complexity (i.e. a minimum number of SKUs in the range)

## FAQs

### What is Unduplicated Reach?

Unduplicated Reach of a combination of items is the percentage of people whom are activated by (or are reached by, prefer, like, hear, …) at least one of the items in that combination.

For example, if you consider offering products A and B (but not product C), and know that:

• Sally like A,
• Billy likes B,
• Zosja likes C, and
• Peter likes A,

then your Unduplicated Reach for A+B is 75% (because only Zosja doesn’t like either A or B).

### How many items can you test in TURF analysis?

Theoretically, it is possible to test a very large number of items (SKUs, products, flavours). For example, even 1000, as long as you have information about relative preferences for these 1000 SKUs across your consumers.

But because TURF is performed on survey data, it is not often realistic to ascertain relative preferences for more than 50 items for each individual. That’s why Conjointly is configured to limit TURF analysis to 50 items.

### What are the limitations of using TURF analysis for range optimisation?

TURF is commonly performed on a set of your own flavours, SKUs, or product variants. If you do not also consider your competition, this could result in optimisation for your range only.

There are two ways to avoid this pitfall:

1. Consider including competitor SKUs into your list of items. You can then run TURF analysis with the constraint of always including competitor ranges and mainly look at the total reach of your own product set.
2. Consider only surveying your loyal buyers. This way, you will have a sample who are unlikely to easily switch to competitor products.

### What if my TURF reach percentages are very low?

If you see low reach percentages for top combinations, it means that respondents’ preferences are non-homogeneous. That is, different respondents like different things.

For example, if the best two-way combination of items reaches only 10% of sample, it implies that the other 90% of sample like something else and there is not a clear winning combination.

Generally, the greater the number of items in the test, the lower the TURF reach percentage. However, what level of low threshold to use depends on your objectives.

### How is TURF analysis different from conjoint analysis?

They are different in multiple ways:

• Conjoint analysis is both a way of asking questions and an analytical framework, while TURF analysis is a type of calculation that runs on previously collected data.
• Conjoint data can be used in TURF analysis, but only to analyse combinations of levels within a specific attribute. For example, if you had an attribute of flavours with levels like `sweet`, `sour`, and `tangy`, TURF can help you find two flavours that will cater to most of the market. (You can do similar analysis through preference share simulations, but it will take into account all attributes, not just one attribute).
• Conjoint (especially when you run simulations) is more versatile than TURF and can be used to analyse mixing features in one product (i.e. combinations of levels that go well together, not combinations that will appeal to different people).

### How do I do TURF analysis in Excel?

Excel by itself is not set up to run complex analyses like TURF, but you do have options:

• Use the free and friendly Conjointly TURF Analysis Simulator and export results with nice formatting into Excel.
• Use an R package together with an Excel-R connector (such as BERT).
• Buy and install a specialised TURF package for Excel.

### Does the order of the items in the combination matter?

No, it does not matter. The Conjointly TURF simulator will sort them in the same order as in the columns in the original dataset.

### How to get the most out of open-ended questions

Open-ended questions should be part of any research project as they can gather in-depth and rich insights from your target audience. Learn more about open-ended questions and getting the most out of them for your projects.

View article