TURF Analysis: What It Is, How It Works and Example

TURF (Total Unduplicated Reach and Frequency) analysis finds the mix of products, flavors or features that reaches the most different people, counting each person once. It uses a clear rule for what counts as “reached.” Follow the worked example to see how overlapping preferences and launch constraints change the recommended range.

A useful report names a range you can supply, shows whom each addition reaches and explains what could change the choice. One winning percentage can hide supply limits and overlapping audiences.

What the analysis needs

Keep each person’s responses across all the options, not just the total for each one. Options may be flavors or services, with a fixed number of places in the range. Displayr's TURF guide explains this portfolio task and the person-by-option records it uses.

Choose the reach rule before comparing ranges. In the example below, a person qualifies if they would consider at least one selected flavor at the stated offer. Liking both still counts as one person. This measures stated consideration, not actual transactions.

The method suits a team asking, “Which two offers belong together?” To decide which benefits deserve development, start with our MaxDiff guide. Benefit priority and range coverage may use related evidence, but they need different calculations.

Example: choosing a second sparkling-water flavor

This hypothetical example follows sparkling-water brand Rill. It has two launch slots and needs an available flavor to join its existing Lime. The next most popular flavor may appeal to the same people, leaving others without an option they would consider.

Below, the two individually strongest flavors reach six of twelve people together. Another available pair reaches nine. A third pair reaches ten but cannot be supplied at launch. Keep the response records and business limits beside those results.

Each Rill candidate is one 330 mL can offered at USD 1.80 before tax. The can format, displayed price and shelf context are kept the same; the flavor label changes. The candidates are A Lime, B Lemon, C Berry and D Ginger.

Lime must stay as the approved existing flavor. Ginger cannot launch because its flavoring supply is not confirmed for the required volume. Those are business conditions, not signs that people dislike Ginger.

Choose which available flavor should join Lime. Show every pair so the team can see what the constraints exclude and when to revisit them. Reporting only the highest reach before applying those rules would not answer the launch question.

Twelve response records show the overlap

The twelve rows below were written by hand for teaching. They represent adult category shoppers, with complete answers and weights of one. They are not results from a survey, simulation, taste test or transaction.

A 1 records willingness to consider that flavor at the stated offer. A 0 records rejection. Read the four values in each row as A / B / C / D: Lime, Lemon, Berry and Ginger.

Six people consider Lime, five consider Lemon and four each consider Berry and Ginger. Those totals put Lime and Lemon first individually. But P01 through P05 already appear in Lime’s audience. Lemon adds no one new to Lime.

Berry overlaps with Lime for P06 and adds P07, P08 and P09. Its smaller individual total adds more new people to the range. Only the combined records show that overlap; four separate flavor totals cannot.

Compare every pair using the same twelve people

For each row, check whether any selected flavor has a 1. If so, count that person once. Divide the number reached by all twelve people in the study.

For Lime and Berry, Lime covers P01 through P06. Berry adds P07 through P09, while P06 is already covered. The result is nine distinct people out of twelve, or 75%.

Keep Lime and add Berry under these launch rules. That pair reaches nine people, compared with six for the only other available pair. Lime and Ginger would reach the most people without the supply restriction, but the business cannot currently offer it.

Keep P12 in the total even though no flavor reaches them. Dropping that row would turn the same nine people into 9/11, or 81.8%, without reaching anyone new. Define who belongs in the study before checking which offers they accept.

Frequency shows how many options people accept

Here, frequency counts accepted options within the selected range. The table also shows average accepted options per person reached. Neither number tells you how often anyone shops or how many cans they buy.

Lime and Lemon get eleven positive entries from six people, not eleven people. Five of the six accept both flavors. Lime and Berry spread ten positive entries across nine people instead.

If you want more people to have at least one acceptable option, compare unique reach. If several acceptable options per person also matter, name that objective and examine it separately. Repeated willingness to consider an offer does not show repeat purchases.

Changing supply or range rules can change the choice

If Ginger’s supply is confirmed and Lime must stay, Lime and Ginger reach the most people in these records. They add P10 and P11, whom Lime and Berry miss, but lose P09. Reach rises from nine to ten, a net gain of one person. Who is covered changes as well as how many.

If Lime may be dropped while Ginger remains unavailable, Lime and Berry tie with Lemon and Berry at nine. Both cover P01 through P09. Reach cannot choose between them. Operations or another agreed objective may decide the tie, but show that reasoning apart from the reach calculation.

These twelve invented records make the overlap easy to see. They cannot estimate population demand or sampling uncertainty. For a paid study, ask how stable the leading ranges are under its sampling or simulation approach, other reasonable reach rules and important buyer groups. Plan the evidence for the cost of choosing the wrong range. Do not borrow a universal sample size.

Different reach rules answer different questions

The yes/no consideration rule above is one option. Sawtooth's TURF documentation also describes probability-weighted, first-choice and threshold methods for its MaxDiff analysis. Read a probability result using that model’s assumptions. It is not the same as counting rows in this yes/no table.

Ask the analyst to keep the question, response scale, any cutoff or conversion from utility scores and the reach rule. If ratings become yes/no values, agree the cutoff before choosing a range. Changing “would consider” to “first choice” changes the question the analysis answers.

Keep missing answers apart from rejection. This example has no missing cells. For real data, agree how to handle incomplete answers or use a suitable model, then report its assumptions and the number included. A blank is not automatically zero. If audience weights are justified, report them and calculate coverage against the weighted total for that same audience.

Ask for a report you can check

A useful study returns more than a ranked list. For Rill, check the full offers, target audience, person-by-option responses or simulated inputs you can check, reach rule and two launch constraints. Ask for the leading available ranges and an explanation of why the apparent winner was excluded.

Report where responses came from. Human stated consideration, simulated consideration and purchase history are different evidence. For simulated responses, check the audience data and human comparisons that support this task. Our guide to synthetic and human respondents explains how to choose and combine these sources.

For a larger range, ask whether the analyst checks every allowed combination or uses a search method. Check what that lets them claim about the result. Keep close alternatives and ties so a supply change can be evaluated later using the original decision rule.

Then explain whom an addition brings into the range. Berry adds three people Lime misses, but the teaching table cannot explain their reasons. A real study could ask about the occasions or expectations behind that difference. Keep the initial answers; do not change them after seeing the preferred result.

Reach leaves taste and substitution unanswered

TURF coverage cannot tell you whether Berry takes purchases from Lime, brings shoppers from a competing brand, or creates a new occasion. When the decision depends on what a range change replaces in an actual shelf choice, our shopper research guide follows that choice and substitution question.

If the offers differ in size, features and price, the business may instead need a conjoint study to study those trade-offs. Keeping the price fixed in Rill’s example avoids answering a different product-combination question with the same reach table.

Test taste and physical use with the product. Ask operations to confirm supply readiness. During rollout, keep availability, distribution, promotion and actual sales records to study trial and repeat buying. The nine positive rows in the recommended pair do not measure those outcomes.

Plan a comparison the business can use

For Rill, the research brief starts with the four offers, target buyers, current range and supply limits. Sapien builds simulated audiences from data, compares alternatives and helps teams understand differences across groups. We can show how buyer groups react to each range and investigate their reasons.

If you need unduplicated reach, agree the person-by-option records, calculation and report when planning the study. Include human product testing and commercial checks where the decision needs them. Design the research around a range you can actually launch.

Book a demo with the offers you can supply and the audience coverage you need to improve. For this range, the useful comparison is why Berry adds buyers whom Lime misses, and whether the difference is worth testing with the product and in market.