Analyze TURF

blog-image

Why it is dangerous to look at only a single optimal solution from a TURF analysis, and how the Stat & More approach, supported by an online demo tool, makes it possible to consciously select truly relevant portfolios.



Going beyond the “best” TURF solution

TURF analysis (Total Unduplicated Reach and Frequency) is a powerful tool for optimizing a range of products, services, or messages, by identifying combinations that maximize target coverage while limiting overlaps. However, relying solely on the single “optimal” solution suggested by the algorithm is often insufficient – and even risky – for making a sound business decision.

In this article, we explain why it is essential to consider a set of candidate solutions rather than a single winning combination, and how the demo tool developed by Stat & More helps marketing and research teams make informed trade‑offs.

Reminder: what is a TURF analysis for?

A TURF analysis aims to answer a seemingly simple question: “Which combination of items (products, services, features, messages, etc.) reaches the largest number of people in the target, while avoiding offers that cannibalize each other?”.

In practice, the algorithm tests many possible combinations and computes, for each of them:

  • The reach (often denoted rchX): the share of the population that finds at least one element in the combination attractive.
  • The frequency (frqX): the average number of elements chosen per person within the combination (useful for understanding how rich or dense the offer is).

The usual objective is then to identify, for a given range size (for example 4, 5, or 6 products), the combination that maximizes reach.


The limitation of a single “optimal” solution

In practice, most TURF deliverables focus on a single solution per range size: the “best” combination of 3 products, then of 4, 5, etc. This type of output is simple and reassuring, but it has several drawbacks.

1. The risk of a local optimum

Many operational implementations rely on iterative “greedy” algorithms that, at each step, add the item that brings in the most new customers, or conversely successively remove the worst performers (“forward TURF” or “backward TURF”). Yet these strategies do not guarantee the global optimum: they may lead to sub‑optimal combinations simply because some synergies between items are never explored.

A solution may appear “better” in the neighborhood of the combinations explored, while another combination, further away in the search space, would actually reach more people. This is precisely the distinction between a local optimum and a global optimum in combinatorial optimization.

2. A purely “research‑driven” view of performance

Even when the algorithm is exhaustive (that is, it evaluates all possible combinations for a given range size), the solution that maximizes reach is not always the best one for the company. Indeed, the algorithm does not know:

  • The margins per product or service.
  • The production constraints, logistics, or supply constraints.
  • The brand image stakes or strategic positioning issues.

Two combinations can deliver almost identical coverage (for example 95.4% of the target each), yet have very different economic implications: in one scenario, overall margin may be significantly higher or the range much easier to produce. Focusing only on the “top reach” solution therefore means overlooking essential profitability and feasibility levers.

3. A misleading impression of a single truth

Presenting a single “optimal” combination can give stakeholders the illusion that there is only one right answer, even though survey data, by design, only capture part of business reality. In many cases, several solutions are statistically equivalent in terms of reach and frequency and must be differentiated using internal criteria.


Why explore all the best solutions?

To truly deliver on its promise, a TURF analysis must highlight a set of candidate solutions rather than impose a single scenario.

1. Identifying truly equivalent statistical solutions

By computing, when possible, all combinations for a given range size, one often observes:

  • A small group of top‑ranked combinations, very close to each other in reach and frequency.
  • Combinations that are slightly less effective in terms of reach but still very competitive.

For example, one solution A may achieve 95.4% reach with 2.1 items selected on average, while another solution B also achieves 95.4% reach with 2.0 items on average, by replacing only one item. From a research standpoint, these two solutions are statistically equivalent and should be considered as two serious scenarios.

2. Comparing research performance and internal constraints

Once this panel of winning solutions has been identified, the key is to confront it with:

  • Internal margin and cost data.
  • Industrial or logistics capacity thresholds.
  • Brand priorities and competitive differentiation goals.

In some cases, accepting a slight drop in reach (for example moving from 95.4% to 94.8%) allows substantial gains in margin, industrial simplicity, or strategic consistency. It is precisely in this trade‑off space that TURF becomes a true decision‑support tool, rather than a purely academic exercise.

3. Co‑constructing the decision with the client

At Stat & More, we believe that TURF reporting should be a co‑construction effort with the end client. Rather than delivering a single “ideal” range, we present:

  • Several candidate solutions for each range size deemed relevant.
  • The associated metrics (reach, frequency, possible cannibalization, etc.).
  • A discussion framework to integrate internal data (price, margins, constraints, strategy).

The “best” solution is no longer the one that comes out first from the algorithm, but the one the company chooses consciously, having understood the associated statistical and economic trade‑offs.


A TURF demonstrator to explore scenarios

To make these trade‑offs more concrete and accessible to non‑specialist teams, Stat & More has developed a TURF demo website that lets users directly manipulate candidate solutions and visualize their potential impacts.

Below, you can directly interact with the online demo hosted by Stat & More:

This demonstrator shows how to move from a single “best combination” to a reasoned exploration of several scenarios, each of which can be assessed in light of your own internal data.

A support for informed decision‑making

In practice, this type of tool allows teams to:

  • Quickly visualize how reach and frequency change when the range composition is slightly modified.
  • Identify combinations that are almost equivalent in terms of research performance but more attractive in margin or portfolio simplicity.
  • Fuel trade‑off workshops involving marketing, finance, production, and top management, based on quantified and comparable scenarios.

Rather than passively accepting the solution proposed by the algorithm, decision‑makers can test, challenge, and compare – and therefore make a truly informed decision.


The Stat & More approach: an enriched TURF deliverable

The Stat & More approach is in line with the articles detailing our methodology on TURF and on the MaxDiff + TURF combination, available on our blog.

1. From statistical robustness…

From a statistical and computational standpoint, we implement:

  • Pre‑filtering procedures to focus the analysis on relevant offers and reduce noise.
  • Well‑controlled heuristic algorithms when the number of items exceeds the practical limits of an exhaustive calculation, with multiple starting points to limit the risk of local optima.
  • Targeted exhaustive computations on the most strategic range sizes, especially when business stakes are high.

The goal is to provide a set of candidate solutions that is both statistically sound and operationally usable.

2. …to the integration of business constraints

At the reporting stage, we go beyond standard deliverables by systematically crossing TURF results with:

  • The cost and margin information provided by our clients.
  • Production constraints (capacity, industrial complexity, supply chain).
  • Brand positioning and differentiation issues.

We present several winning solutions per range size, with their research indicators, and then help teams score these scenarios against their own internal criteria. It is this 360° approach, combining statistical excellence and a deep understanding of business stakes, that turns our TURF analyses into a real strategic steering tool, rather than just a “model output”.


Want to go further with your own data?

If you already use MaxDiff surveys, multiple‑choice questions, or internal data on subscription or purchase intent, you probably have the raw material needed to derive real value from an advanced TURF analysis.

At Stat & More, we support our clients end‑to‑end:

  • From data structuring to implementing TURF algorithms tailored to the size and complexity of your portfolio.
  • From generating detailed reports to providing interactive tools, such as our TURF demonstrator, to facilitate internal trade‑offs.
  • From analyzing research performance to producing operational recommendations that take into account your margin, production capacity, and brand strategy constraints.

To learn how this approach can be applied to your own products, services, or media plans, we invite you to visit our website and contact us:

Would you like to discuss a concrete case, test different range solutions, or deploy a simulation tool tailored to your teams? Contact us today to discuss your challenges and jointly build the optimal solution(s) for your company.

To dive deeper into the TURF methodology and its applications, see our previous article on the Stat & More blog.

... #### REFERENCES: ✔ (2024, November 24) Stat & More, Benoit Riou: MaxDiff and TURF https://statandmore.com/fr/blog/sam-maxdiff_turf/
✔ (2022, July 25). Wikipedia project contributors: TURF analysis https://en.wikipedia.org/wiki/TURF_analysis
✔ (2025, May 15). Wikipedia project contributors: Cannibalization (marketing) https://en.wikipedia.org/wiki/Cannibalization_(marketing)
✔ Laramore, J. & SAS Institute. (n.d.). Optimizing Product Assortment with Total Unduplicated Reach and Frequency Analysis in SAS/OR. Paper SAS2981-2019. https://support.sas.com/resources/papers/proceedings19/2981-2019.pdf
✔ Rossi, A. (2023, August 11). Chapter 2 TURF | R Tools for Market Research. https://bookdown.org/rossialessio095/R_Market_Research/turf.html
✔ Serra, D. & Department of Economics and Business, UPF. (n.d.). A new model for designing a product line using TURF analysis. In Research Papers in Economics [Journal-article]. https://www.researchgate.net/publication/46468284_A_New_Model_for_Designing_a_Product_Line_Using_TURF_Analysis
✔ (2025, May 18) Wikipedia project contributors. Greedy algorithm https://fr.wikipedia.org/wiki/Algorithme_glouton
✔ (2025a, January 2) Wikipedia project contributors. Data clustering https://fr.wikipedia.org/wiki/Partitionnement_de_donn%C3%A9es