Mental Availability Diagnostic Kit

A CMO's questionnaire + analysis worksheet, adapted from Jenni Romaniuk's CBM template. All example data below is illustrative — not from any real study. Companion material to Lesson 02: How Markets Actually Work.

Most marketers guess what their brand means to buyers. The Ehrenberg-Bass method measures it — one Category Entry Point at a time, one deviation from the category baseline at a time.

This is the tool the marketing science community uses to answer one question: when a category buyer thinks about buying, does our brand come to mind — and for which situations? The answer is not "yes" or "no." It is a matrix of associations that maps every brand in the category against every buying situation. That matrix is what you build here.

The kit has three parts. The questionnaire template that you deploy in the market. The analysis workbook logic that turns raw responses into a distinctiveness heatmap. And the interpretation heuristics that translate the heatmap into an investment decision.

01Why this exists

Byron Sharp and Jenni Romaniuk built the case that brands grow not by persuading customers to prefer them, but by being mentally available in more buying situations. Two brands with the same market share look identical on the P&L. In the buyer's mind they are not. One is thought of for eight situations. The other is thought of for two. The brand thought of for eight has structural room to grow. The brand thought of for two is at its ceiling.

You cannot see this in Nielsen data. You cannot see this in your own sales data. You can only see it by asking category buyers directly, one Category Entry Point (CEP) at a time, and comparing your brand's associations against the category benchmark.

The questionnaire below is the standard instrument the Ehrenberg-Bass Institute uses. The analysis logic is the same chi-square deviation model the academic literature uses. What is new here is the packaging: a single-file kit any CMO can hand to a market research agency, or run themselves on Prolific for €400.

02Build your questionnaire

Fill in your category, your brand list, and your CEP list. The template renders below. Print it. Send it to your research vendor.

The 7 questions that generate CEPs (Romaniuk, Better Brand Health, 2023):
  1. What are the main occasions for using the product?
  2. Who are the primary users of the product?
  3. What are the main benefits sought from the product?
  4. What are the main alternatives to the product?
  5. What are the main locations for using the product?
  6. What are the main times for using the product?
  7. What are the main complementary products?
Interview 10–15 category buyers with these seven questions before you finalize the CEP list. Do not sit at your desk and invent them.

03How the analysis works

The questionnaire produces a big grid of ticks — every respondent, every CEP, every brand. That grid has four canonical outputs. Each answers a different strategic question.

The four Mental Availability metrics

Mental Penetration
The share of respondents with at least one CEP association with your brand. Answers: How many buyers think of us in any situation?
Mental Market Share
Your brand's share of all CEP associations across all brands. Answers: How present are we in the buyer's mental inventory relative to the category?
Network Size
Average number of CEP associations per respondent — among those with at least one. Answers: When we come to mind, how broadly are we linked?
Share of Mind
Your brand's CEP share within the mental network of buyers who mentioned a focus brand. Answers: Who do we share mental space with?

Mental Penetration is the reach figure. Network Size is the depth figure. Mental Market Share is the competitive figure. Share of Mind is the overlap figure. Track all four. Growing brands typically show rising Mental Penetration and Network Size in parallel — one without the other is a red flag.

The three-matrix workbook

Once you have the raw response grid, every CEP block becomes three stacked matrices — one tab per CEP block:

MatrixWhat it isWhat it tells you
ACTUAL Observed number of ticks for each CEP × Brand cell. Row totals = how "popular" a CEP is. Column totals = how "big" a brand is mentally. The raw scorecard. Big brands win most rows on absolute count — that is not distinctiveness, that is size.
EXPECTED What the counts would be if brand size and CEP size were independent. Calculated by chi-square logic. The category baseline. Every brand gets its "fair share" of every CEP proportional to its overall mental size.
DEVIATION Actual minus Expected — the mental advantage or disadvantage for each CEP × Brand cell. The distinctiveness map. Positive = brand over-owns this CEP. Negative = under-owns. Zero = category average.

Formula 1 — Expected counts

Expected[CEP, Brand] = (Row Total × Column Total) / Grand Total
// Row Total = total ticks for this CEP across all brands
// Column Total = total ticks for this brand across all CEPs
// Grand Total = total ticks in the entire matrix

This is the standard chi-square "expected under independence" formula. It tells you what each cell would be if there were no brand–CEP association at all — just proportional sharing.

Formula 2 — Deviation (three ways to express it)

Raw Deviation = Actual − Expected
// Sign-only signal. Positive = over-associated, negative = under.
// Useful for a raw heatmap, harder to compare across CEP blocks of different sizes.

Deviation Index = (Actual / Expected) × 100
// Romaniuk's preferred display. 100 = category average.
// 120+ = distinctively over-owned. 80− = under-owned.
// Comparable across CEPs and across studies — this is the number to report.

Standardized Residual = (Actual − Expected) / √Expected
// Statistical test. |value| > 2 ≈ statistically significant at p<0.05.
// Use when you need to defend "this deviation is real, not noise."
Which formula to use? Report the Deviation Index (Actual / Expected × 100) to your executive team — the "100 = average" framing is intuitive. Use the Standardized Residual internally to filter out noise before you present anything. The raw deviation is fine for a heatmap but hard to compare across CEP blocks with different totals.

Worked example — Meadowline's "tastes good" mental score

Illustrative numbers built for teaching. Imagine an oat-drink category study with 10 brands and ~1,100 respondents. Read the arithmetic from the top:

StepValueWhat it means
Actual ticks: Meadowline × "tastes good"195195 respondents said Meadowline tastes good
Row total: all brands × "tastes good"1,000"Tastes good" was a busy CEP — 1,000 total ticks
Column total: Meadowline × all attributes2,100Meadowline is the biggest brand in mental terms
Grand total (this CEP block)11,000All CEP × Brand ticks combined
Expected = 1,000 × 2,100 / 11,000= 190.9If tastiness were split proportional to brand size, Meadowline would get ~191 ticks
Raw Deviation = 195 − 190.9= +4.1Almost exactly at baseline — no distinctiveness signal
Deviation Index = 195 / 190.9 × 100= 102Meadowline is not distinctively tasty. It is category-average, which for the biggest brand is the danger signal.
Compare to NordicOat: Column total 1,800 (smaller mental footprint) → Expected on "tastes good" = 1,000 × 1,800 / 11,000 = 163.6. Actual = 177 → Deviation Index = 108. NordicOat over-owns tastiness relative to its size — a distinctive asset worth amplifying.
And Cloudmilk on a different CEP: Column total 1,500 (small barista-focused brand). On the "For coffee" CEP (row total 850, grand total 10,200), Expected = 850 × 1,500 / 10,200 = 125.0. Actual = 165 → Deviation Index = 132. Cloudmilk over-owns "For coffee" by 32 percentage points. That is the strongest distinctive asset in this fictional dataset — the entire brand strategy should protect it.

Now scale that logic across every CEP × Brand cell in the study. You get a full distinctiveness heatmap. Every green cell is an asset your brand owns. Every red cell is a gap. Every column of mostly-white cells is a brand that is mentally big but distinctively nothing — the most dangerous position in the category.

Live example — an illustrative deviation heatmap

Below: fictional deviation indices built to demonstrate the four patterns you should learn to spot. Eight CEPs × six brands. Green = over-indexed (distinctive asset). Red = under-indexed (mental gap). White = category-average.

Under-indexedOver-indexed · Scale: 60 ← 100 → 140

How to read this heatmap — four patterns to spot:

  1. Cloudmilk owns coffee. Two green cells stacked ("For coffee" 132, "For latte art" 125) — a coherent barista positioning that is measurably locked into buyers' minds. Cloudmilk's entire brand plan should defend and amplify these two cells.
  2. NordicOat owns "modern" (122), SunHarvest owns "sustainable" (126), Purely owns "healthy" (121). Three challengers, each with a single distinctive asset. Narrow but ownable — the ownable CEP is the only reason each brand grows.
  3. Meadowline is category-average across the board. No cell above 108, no cell below 92. This is the Kraft Heinz problem — mentally big, distinctively nothing. Meadowline grows and shrinks with category tailwind. The moment a distinctive challenger scales, Meadowline loses share and cannot explain why.
  4. The "With family" row belongs to no one. Highest index is 104. No brand claims the family occasion. This is category white space — the point the next section covers in detail.

FarmFirst (the sixth column) has no green cells and multiple red cells. That is a brand with weak mental structure — either it needs multi-year investment to build associations, or the strategy needs to change.

04How to run it

Sample size

Minimum n = 300 category buyers per market. Below 300, individual cell counts (especially for smaller brands × niche CEPs) become too small to trust the chi-square math. Romaniuk's practical rule: at least 20 expected ticks per brand-CEP cell you actually care about. If your smallest brand of interest has a mental market share of 3%, you need ~700 respondents to get 20 expected ticks in most cells. Target n = 500 for a well-established category. Target n = 1,000 if you want statistically defensible sub-group cuts (heavy vs light buyers, region, demographic).

Survey platforms

Three price bands. Prolific (~€3–4 per completed 15-min response, best sample quality, EU + UK + US). Bilendi / Cint / Toluna (~€2–3, larger panels, more targeting depth, DACH strong). Norstat / Respondi (traditional panel providers, best for regulated categories). For a 500-respondent DACH oat-drink study, budget €1,500–2,500 in fielding. Add €800 if you use a fieldwork agency to script the survey.

Fielding checklist

Cleaning the raw data

  1. Remove speeders (< 40% of median completion time).
  2. Remove straight-liners (same response pattern on 5+ consecutive CEP screens).
  3. Remove respondents who ticked every brand for every CEP (careless clicking).
  4. Remove respondents who ticked zero brands across all CEPs (survey abandon or bot).
  5. Cross-check screener: keep only respondents whose brand purchase history in the screener is internally consistent.

Building the pivot (Excel / Google Sheets)

  1. Tab 1 — RAW: one row per respondent, one column per CEP × Brand combination, 1 = ticked / 0 = not. This is the export from your survey platform.
  2. Tab 2 — ACTUAL_CEP_1: pivot the sensory attributes block into a CEP (rows) × Brand (columns) matrix using SUMPRODUCT. Add row totals, column totals, grand total in the last row / last column.
  3. Tab 3 — EXPECTED_CEP_1: same layout. Every cell = (Row Total × Column Total) / Grand Total. This is a single formula copied across the matrix.
  4. Tab 4 — DEVIATION_CEP_1: same layout. Every cell = (ACTUAL cell / EXPECTED cell) × 100. Apply conditional formatting: red-white-green diverging scale, midpoint at 100.
  5. Repeat tabs 2–4 for CEP_2 (occasions) and CEP_3 (context).
  6. Tab 11 — METRICS: compute Mental Penetration, Network Size, Mental Market Share, Share of Mind per brand from the RAW tab. See the formulas below.
Mental Penetration[Brand] = COUNTIF(sum-per-respondent for Brand > 0) / total respondents

Network Size[Brand] = AVERAGE(sum-per-respondent for Brand | sum-per-respondent for Brand > 0)
// Filter to respondents with at least one association, then average

Mental Market Share[Brand] = SUM(Brand column across all CEPs) / SUM(entire matrix)

Share of Mind[Brand X | Focus Brand Y] = SUM(Brand X ticks among respondents who ticked Focus Brand Y) / SUM(all ticks among those respondents)

Interpretation heuristics

Deviation IndexMeaningWhat to do
> 120 Distinctively over-associated. Your brand owns this CEP in the buyer's mind. Amplify. Make this CEP a codified distinctive asset — put it in the tagline, the pack, the media. Byron Sharp's mental-availability compounding logic applies most strongly here.
80–120 Category-average. No distinctiveness — neither an asset nor a gap. Watch. These CEPs are the middle ground where a small campaign push can create ownership if the category structure allows. Do not spend big to build here unless the CEP is strategically core.
< 80 Under-owned. Buyers do not think of your brand for this CEP. Decide: ignore (this CEP is off-strategy for your brand) or invest (this CEP is strategic and you accept a 3–5 year build). Do not attempt to close a <80 gap with tactical activity — the memory structure has to be built.
Whole CEP row where no brand indexes > 120 Category white space. No brand owns this CEP. Land grab — but weigh the potential first. An unowned CEP is only an opportunity if the CEP itself matters. Check the row total (how many respondents ticked any brand for this CEP) and cross-reference against category buying frequency for that occasion. A big-volume CEP with no owner is the single most valuable finding a study produces — a first mover with a consistent multi-year signal can own it. A low-volume CEP with no owner is usually just an irrelevant CEP for the category, not an opportunity. Rank white-space CEPs by row total (potential), then invest in the top two or three.
Whole brand column with mostly 100 ± 15 Mentally big, distinctively nothing. The Kraft Heinz problem. Big Mental Market Share, low Network Size, no ownable CEP. The brand grows with category tailwind and shrinks the moment a distinctive challenger enters. Fix by picking two CEPs and investing consistently for 24+ months.

05What to hand back to the business

The study produces four deliverables the executive team should see, in this order:

  1. Mental Market Share ranking — one bar chart, one number per brand. This is the "how big are we mentally" scoreboard. Compare against volumetric market share. A brand mentally bigger than it is commercially is compressed — it will grow with distribution / physical availability. A brand commercially bigger than it is mentally is running on borrowed time.
  2. Network Size ranking — how broadly each brand is linked. Growing brands show rising Network Size year-over-year. Declining brands show it collapsing to one or two CEPs before they show any P&L symptom.
  3. The distinctiveness heatmap — the CEP × Brand deviation matrix. This is where the strategic conversation actually happens. Not "are we in the buyer's mind" but which situations are we in the buyer's mind for.
  4. Two owned CEPs, two build-target CEPs, one white-space bet — the output of the analysis. Give the CEO five bets, not fifty findings. This is the framing that turns a research report into a marketing plan.
Mental availability is the most measurable and least measured thing in marketing. Every CFO can tell you distribution weighting to two decimal places. Almost no executive team can tell you their brand's deviation index on their top five CEPs. Fix that.

The brand that knows its mental map plans its next three years around a real category picture. The brand that guesses runs the same campaign it ran last year, wonders why it did not work, and blames the media agency. The difference is one questionnaire and one workbook. Now you have both.