Zero Acre Farms Consumer Research · May 2026 N = 714 U.S. adults Chipotle pre-pitch read

How U.S. consumers respond to a seed oil‑free Chipotle

A four-pillar read on brand-KPI alignment, frequency lift, message and menu demand, and the consumer-side risks of switching cooking oils.

Sample size
714 completed surveys
Frequent / Infrequent
399 / 315
Gen Z (18–27)
263 respondents (37%)
Field period
April–May 2026
Topline

The four headline findings

Each number below reflects the consumer-side movement attributable to a 100% seed oil-free Chipotle, net of pre-exposure baseline.

+10pp
KPI lift on "better-for-you food." Share rating Chipotle Good/Extremely well rises from 66% to 76%.
58%
would visit more often once Chipotle is 100% seed oil-free — 26% significantly, 32% slightly.
44%
prefer the new oils over Chipotle's current canola and high-oleic sunflower blend, vs. 10% who prefer current.
50%
more likely to eat at Chipotle after hearing the phased 100% seed oil-free commitment; 10% less likely.
  • Every Food-with-Integrity attribute lifts. "Better-for-you food" (+10pp), "environmentally responsible sourcing" (+11pp), and "ingredient transparency" (+6pp) all move post-exposure. "Real food" (+3pp) is constrained by a high baseline.
  • The lift is concentrated in the audience Chipotle most needs to grow. Frequent customers already rate Chipotle near ceiling; the seed oil-free switch moves the dial most among Infrequent customers (+11 to +21pp across the four KPIs) and consistently more among Gen Z than older cohorts.
  • "Avocado oil" and "Clean-fried" lead the message stack. Avocado oil tops at 62% top-2 appeal; Clean-fried follows at 59%. The newer "Fera fruit oil" and "Regenerative" claims trail the field at 47–48% — a familiarity gap, not a credibility one.
  • "Seed oil-free" and "Processed without hexane or solvents" land identically. Both score 53% top-2 appeal with means of 3.48 and 3.46. The data does not support the hypothesis that seed oil-free outperforms solvent-free language at the consumer level.
  • The 53-ingredient list is a low-risk number to change. 83% of consumers say a change to 54 or 52 wouldn't affect their opinion; only 5% would significantly change their opinion.
  • Risk is contained. 28.6% of respondents express any concern about Fera fruit oil at the initial reveal; only 3.9% of the full sample say that concern would actually change whether they eat at Chipotle. Concerns about olive pomace land lower still.
Sample

Who we heard from

714 U.S. adults who eat at restaurants at least occasionally, split into Frequent (visits Chipotle at least a few times a month, or every few months) and Infrequent (every year or two, or never).

Chipotle visit frequency

Screener S3 · n=714

Age distribution

Screener S1 · n=714
Age bucket × Chipotle frequency
AgeFrequentInfrequentTotal% of sample
18-27 (Gen Z)161102 26337%
28-4314740 18726%
44-596974 14320%
60+2299 12117%
Total399 315714 100%

Sample skews younger and more Frequent than the U.S. adult population. When reading subsegment lifts below, treat aggregate stats as descriptive of this sample, not as population-projectable estimates. Margin of error at 95% confidence is ±3.7pp for aggregate measures; subsegment MoEs are larger and noted with each table.

Pillar 1 Brand alignment

A seed oil-free Chipotle lifts every Food-with-Integrity attribute — most where Chipotle has the most room to grow

We asked the same four-attribute matrix before and after a 100% seed oil-free hypothetical. The pre/post deltas below are the core Pillar 1 evidence.

Brand-KPI ratings: pre-exposure vs. post-exposure

Q3.1 × Q10.1 · % rating Good or Extremely well (top-2 box)

Q3.1: "How well does Chipotle do each of the following today?" — Q10.1: same matrix, after "Now imagine Chipotle is 100% seed oil-free and uses higher-quality cooking oils."

How to read the significance markers in the Pillar 1 tables below. Each pre/post KPI lift was tested with McNemar's paired-binary test on the same respondents who answered both Q3.1 and Q10.1. Look for these markers next to each lift value:

*** p<0.001 (highly significant) ** p<0.01 * p<0.05 no marker = not statistically distinguishable from zero change
KPI lift by Chipotle frequency · % rating Good or Extremely well
Attribute All
n=714
Frequent
n=399
Infrequent
n=315
Real food 76% → 78% +2.8pp 93% → 89% -3.9pp* 54% → 65% +11.2pp***
Ingredient transparency 71% → 77% +5.6pp** 87% → 87% +0.7pp 52% → 63% +11.8pp***
Better-for-you food 66% → 76% +9.5pp*** 84% → 86% +1.4pp 44% → 63% +19.5pp***
Environmentally responsible sourcing 61% → 72% +10.8pp*** 80% → 83% +2.5pp 37% → 58% +21.1pp***
Read

Among Frequent customers, baseline scores are already 80–93% top-2 box; the oil change holds them at ceiling but barely moves the dial — and produces a small but statistically significant dip on "real food" (-3.9pp, p<0.05), where the baseline is highest. The action is in the Infrequent segment, where every KPI lifts double-digits, all highly significant: +11pp on real food, +12pp on ingredient transparency, +20pp on better-for-you food, +21pp on environmentally responsible sourcing (all p<0.001). This is consistent with a "better oil brings lapsed customers back" thesis.

KPI lift by age bucket · % rating Good or Extremely well
Attribute 18-27
n=263
28-43
n=187
44-59
n=143
60+
n=121
Real food 74% → 79% +5.1pp 87% → 83% -3.8pp 78% → 79% +1.2pp 61% → 70% +9.0pp*
Ingredient transparency 68% → 76% +8.3pp* 80% → 80% +0.1pp 76% → 80% +4.0pp 59% → 69% +9.9pp*
Better-for-you food 62% → 73% +11.5pp** 81% → 83% +1.7pp 72% → 78% +5.6pp 46% → 68% +21.5pp***
Environmentally responsible sourcing 59% → 72% +13.0pp*** 74% → 78% +4.5pp 66% → 73% +7.7pp* 40% → 59% +19.0pp***

The age cuts mirror the Frequent/Infrequent pattern: 60+ start with the lowest baselines and show the largest absolute lifts. Gen Z lifts consistently and meaningfully across all four KPIs — note that "real food" actually moves +5.1pp among Gen Z vs. only +1.3pp among older respondents.

Pillar 1 · supporting Personal importance ranking

Seed oil-free ranks last of six food-integrity attributes — but the gap to non-GMO is only 5 points

Of the six Food-with-Integrity attributes tested, seed oil-free cooking is the least personally important to consumers. The gap to non-GMO is close to the margin of error; the gap to the top of the list (no artificial flavors) is about 18 points.

"How much does each of the following matter to you personally?"

Q5.1 · % rating 4 or 5 (top-2 box) · n=709
Food-integrity attribute importance × segment cuts · % rating 4 or 5
Attribute All Frequent Infrequent Gen Z Other ages
No artificial flavors/colors/preservatives 62% 72% 49% 57% 65%
No antibiotics or added hormones 61% 69% 52% 58% 63%
Organic ingredients 57% 67% 43% 62% 53%
Local & sustainable sourcing 56% 65% 45% 57% 56%
Non-GMO ingredients 49% 58% 38% 46% 51%
Seed oil-free cooking 43% 53% 32% 42% 44%
Read

The full ranking from most to least personally important: no artificial flavors/colors/preservatives (62%), no antibiotics or added hormones (61%), organic ingredients (57%), local & sustainable sourcing (56%), non-GMO ingredients (49%), and seed oil-free cooking last at 43%. Two readings of this matter for the pitch: (a) seed oil-free has not yet reached the cultural-awareness baseline of the long-established food-integrity attributes that Chipotle has built around for years, so room to grow exists; (b) the 5-point gap to non-GMO is close to the ±3.7pp margin of error — within striking distance, especially given non-GMO has been in the consumer vocabulary for two decades. The Frequent segment rates every attribute higher than the Infrequent segment by 20–30pp (including seed oil-free), so Chipotle's existing customer base over-indexes on food-integrity values broadly.

Note: this question measures personal importance independent of Chipotle. Pre-exposure brand performance (Pillar 1 above) and post-exposure brand performance are separate measures — the gap between "this attribute matters to me" and "Chipotle delivers on this attribute" is where the oil change creates value.

Pillar 1 · supporting 53-ingredient probe

Most consumers wouldn't notice — or care — if the ingredient count changed

Chipotle's marketing organization is sensitive to the 53-ingredient figure on the website. The data suggests a 54 or 52 figure during a phased rollout would be effectively invisible to most consumers.

"If Chipotle's ingredient count changed to 54 or 52, how much would you care?"

Q4.2 · single-select · n=706
Ingredient-count sensitivity by segment · % distribution
Response All Frequent Infrequent Gen Z Other ages
Wouldn't notice or care 53% 44% 64% 47% 56%
Notice but no opinion change 30% 36% 24% 30% 31%
Slight opinion change 12% 14% 9% 17% 9%
Significant opinion change 5% 6% 3% 6% 4%
Read

Only 5% of consumers say a change in the ingredient count would significantly change their opinion of Chipotle. The marketing team's protectiveness of "53" is not a consumer-side risk; treat it as an internal-comms consideration rather than a brand-equity one.

Pillar 1 · supporting Net preference vs. status quo

When forced to choose, 44% prefer the new oils to Chipotle's current canola and high-oleic sunflower

Asked head-to-head, no segment prefers the status quo. The "no preference" pool (46%) is large but neutral — not a defense of current oils.

"If you had to choose between Chipotle's current oils and the new oils discussed, which would you prefer?"

Q13.7 · single-select · n=709
Net preference by segment · % distribution
Response All Frequent Infrequent Gen Z Other ages
Strongly prefer new oils 23% 27% 18% 23% 23%
Slightly prefer new oils 21% 24% 16% 20% 21%
No preference 46% 38% 56% 48% 45%
Slightly prefer current oils 5% 6% 5% 5% 6%
Strongly prefer current oils 5% 5% 5% 4% 5%
Net: prefer new (top-2) 44% 51% 34% 43% 44%
Net: prefer current (bottom-2) 10% 11% 10% 9% 11%
Read

Across every cut, the share preferring the new oils outweighs the share preferring current oils by 3:1 or more. The largest preference-for-new comes from Frequent customers (60%) and Gen Z (50%). Even among Infrequent customers, where engagement is lower across the board, preference-for-new (28%) still beats preference-for-current (5%).

Pillar 2 Loyalty & frequency

A 100% seed oil-free Chipotle would drive measurable frequency lift — sharpest among the highest-LTV segment

Two questions feed this pillar: Q10.2 asked respondents whether they'd go more often (3-point scale); Q11.2 followed with how the commitment affected likelihood to eat at Chipotle (5-point scale).

"Would you expect to go to Chipotle more or less often?"

Q10.2 · 3-point scale · n=708

Asked after the full 100% seed oil-free framing in Topic 10.

Frequency lift by segment · % distribution
Response All
n=714
Frequent
n=399
Infrequent
n=315
Gen Z
n=263
Other ages
n=451
Significantly more often 26% 37% 11% 32% 22%
Slightly more often 32% 38% 25% 34% 31%
No change 42% 25% 64% 34% 47%
Any frequency lift (top-2) 58% 75% 36% 66% 53%
Frequency lift by 4-bucket age · % saying "more often"
Age Significantly more Slightly more No change Any lift (top-2)
18-27 (Gen Z) n=263 32% 34% 34% 66%
28-43 n=184 32% 35% 33% 67%
44-59 n=142 21% 35% 44% 56%
60+ n=119 9% 19% 72% 28%
Read

Headline: 26% would go significantly more often, and another 32% slightly more — combined 58% any-lift. Frequency lift is sharply differentiated by current behavior: 75% of Frequent customers would visit more often (deepening existing loyalty) vs. 36% of Infrequent customers (reactivation of lapsed/light users). The Gen Z any-lift figure (66%) outpaces the rest of the sample by 13pp.

Likelihood after the phased 100% seed oil-free commitment

Q11.2 · 5-point scale · n=711

"Imagine Chipotle publicly committed to going 100% seed oil-free across their menu, with a phased rollout. How does that commitment affect how you feel about eating at Chipotle?"

Commitment likelihood by segment · % distribution
Response All Frequent Infrequent Gen Z Other ages
Much more likely 26% 36% 14% 27% 26%
Slightly more likely 24% 29% 17% 25% 23%
No change 40% 27% 57% 35% 43%
Slightly less likely 4% 5% 2% 6% 3%
Much less likely 6% 3% 10% 7% 5%
Net: more likely 50% 65% 31% 52% 49%
Net: less likely 10% 8% 12% 13% 8%
Read

The phased-commitment framing earns a 5:1 net-positive response aggregate (50% more likely vs. 10% less likely). The Frequent segment is net-positive by even larger margins (65% more likely vs. 8% less likely). A small negative tail exists in every segment but never exceeds 13% — within the typical noise floor for any brand-change announcement.

Pillar 3 Messaging & menu demand

Avocado oil and Clean-fried lead the message stack; seed oil-free and solvent-free land identically

Eight message territories were rated on a 5-point appeal scale. Ranked below by share rating each "Appealing" or "Very appealing" (top-2 box).

Message appeal ranking

Q7.1 · % rating Appealing or Very appealing · n=710

"How appealing is each of these ways Chipotle could talk about a new oil?" 5-point scale from Not appealing → Very appealing.

Message appeal × segment cuts · % rating 4 or 5
Message All Frequent Infrequent Gen Z Other ages
Made with Avocado oil 62% 76% 45% 58% 65%
Made with Organic oil 61% 74% 44% 64% 59%
Clean-fried 59% 67% 50% 61% 58%
Seed oil-free 53% 63% 40% 56% 51%
Processed without hexane or solvents 53% 63% 40% 53% 53%
No canola oil 51% 61% 39% 52% 50%
Made with Organic Fera fruit oil 48% 61% 32% 51% 46%
Made with Regenerative Organic Certified oil 47% 58% 32% 50% 45%
Read · "Seed oil-free" vs "Processed without hexane or solvents"

Both messages score 53% top-2 appeal. Mean scores are 3.48 (seed oil-free) vs. 3.46 (without hexane). Across every segment cut — Frequent, Infrequent, Gen Z, other ages — the two messages are statistically indistinguishable. This contradicts the hypothesis that consumers care less about solvent-free framing than seed-oil-free framing. From a consumer-appeal standpoint, they're interchangeable.

Read · Familiarity gap, not credibility gap

The two lowest-ranked messages are "Made with Organic Fera fruit oil" (48%) and "Made with Regenerative Organic Certified oil" (47%). These trail the field by 15pp behind "Avocado oil" — but the gap closes among Gen Z (where Fera scores 51% vs. 46% among other ages). This is a familiarity / awareness gap, addressable through claim education, not a fundamental skepticism gap.

Pillar 3 · supporting Willingness to pay

Roughly 6 in 10 would pay something extra for any of the three tested concepts

Three menu concepts were tested with identical 5-tier WTP ladders. The "any premium" signal — the share willing to pay any amount over base — is the primary read; the dollar tiers add granularity.

WTP across the three menu concepts

Q8.2 / Q8.4 / Q9.2 · single-select 5-tier ladders

SOF Bowl: a seed oil-free bowl built à la carte (n=710). Clean Protein: a Chipotle Clean Protein Menu item (n=711). Nachos: Clean-fried Nachos vs. nachos elsewhere (n=710).

"Any premium" willingness (any amount above base price) by segment
Concept All Frequent Infrequent Gen Z Other ages
Seed oil-free bowl (Q8.2) 55% 71% 35% 60% 53%
Clean Protein Menu item (Q8.4) 61% 75% 43% 65% 58%
Clean-fried Nachos (Q9.2) 58% 70% 44% 61% 57%
WTP at "$1 or more above base" tier · share willing to pay $1+ premium
Concept All Frequent Infrequent Gen Z Other ages
Seed oil-free bowl 41% 56% 23% 46% 39%
Clean Protein Menu item 48% 62% 31% 55% 44%
Clean-fried Nachos 42% 55% 28% 46% 41%
Read

The "any premium" signal lands in a narrow band of 56–61% across the three concepts — meaningfully above 50% and consistent across segments. At the higher "$1 or more" tier, Clean Protein Menu items command the highest willingness (48%), followed by Clean-fried Nachos (42%), and SOF bowls (41%). Frequent customers are 25–35pp more willing to pay $1+ than Infrequent ones on every concept (55–62% vs. 23–31%). For the broader QSR pitch, the topline "more than half of consumers will pay something extra for clean-fried" stands up at the aggregate and replicates across cuts.

Detailed five-tier WTP distributions across all three concepts (No / same price · A few cents · Up to $1 · $1–$3 · More than $3) available on request — the patterns are consistent with the summary above. Note: WTP responses are intent-based and should be treated as relative comparisons across concepts rather than projectable price-tolerance estimates.

Pillar 4 Risk & consumer reaction

Risk is contained: most concerns about the new oils are mild, and only a small fraction would change actual behavior

Two paired questions per oil: severity of any concern raised (1–5), and a behavioral yes/no — would that concern actually change whether they'd eat at Chipotle? Mitigation framings tested separately in Topic 13.

Fera fruit oil — % expressing any concern at first reveal

Q6.2 conditional · n triggered: 204 of 714

28.6%

of full sample raised any concern after the Fera fruit oil introduction. The remaining 71.4% had no concern to register.

Among concerned, would it change eating at Chipotle?

Q6.3 conditional Yes/No · n=116 who reached Q6.3

24.1%

of those who expressed concern say it would change whether they eat at Chipotle. That's 3.9% of the full sample. The behavioral floor is small.

What are those concerns about?

The Q6.2 severity question doesn't capture what the concern was. To answer that, we coded the open-ended Q6.1 responses ("What's your honest first reaction? Anything that gives you pause?") from the 197 concerned respondents who answered both. Themes are non-exclusive — one response can touch several.

Concern themes · % of 197 concerned respondents whose open-ended response touched each theme
Theme % of concerned Implied % of full sample What it sounds like in their own words
Taste / flavor uncertainty 37% 10% "Curious how it would affect the taste" · "I would have to taste it to see if I notice a difference"
Is the new oil actually healthier? 19% 5% "I'd be worried about it being less healthy" · "Unclear health benefit"
Unfamiliarity / "never heard of Fera" 16% 5% "I'd be hesitant because I've never heard of fera" · "Unfamiliar oil raises questions"
The phrase "fruit oil" itself 12% 3% "You can't fix food in fruit oil" · "I'd be hesitant because it's fruit oil"
Price / cost concern 9% 3% "Want to know the taste, cost, and availability before forming a strong opinion"
Environmental / sourcing 8% 2% Mentions of farming, organic, pesticides, palm sourcing
Skepticism / marketing distrust 6% 2% "You'd still be liars and frauds because it doesn't matter what oil you use AT ALL"
Read

The concerns are mostly familiar product-launch friction, not category-existential pushback. Taste uncertainty (37%) is the dominant theme — addressable through culinary proof points and the kind of chef testimonial wall Zero Acre already deploys on the foodservice page. Health credibility (19%) is the second-largest theme: consumers want evidence the new oil is actually better, not just marketing. Unfamiliarity (16%) and the "fruit oil" descriptor itself (12%) are language / awareness gaps that resolve with exposure. Outright skepticism that this is just marketing spin is rare (6% of concerned, 2% of full sample). No respondent flagged "palm" as a concern in the open-ended at this stage — that question is probed separately in Topic 13.

Fera concern severity distribution · among those who expressed any concern (Q6.2)
Severity % of concerned Implied % of full sample
1 — Mild 32% 9.2%
2 — Slight 24% 6.9%
3 — Moderate 23% 6.6%
4 — Significant 12% 3.4%
5 — Severe 9% 2.6%
Concern rate × segment — share of segment expressing any concern about Fera at first reveal
SegmentBase nConcerned n% concernedConcerned → would change eating
All 714 204 28.6% 24.1%
Frequent 396 120 30.3% 30.3%
Infrequent 314 84 26.8% 16.0%
Gen Z 262 60 22.9% 22.9%
Other ages 448 144 32.1% 24.7%

Mitigation effectiveness — does deeper context resolve the concern?

In Topic 13 we walked respondents who'd already expressed concern through additional context for each oil, then re-asked whether their opinion shifted.

Fera mitigation: "35% less saturated fat than typical palm, deforestation-free"

Q13.3 conditional · n=37
Response%
Yes, fully resolves my concern11%
Yes, somewhat27%
No change40%
Still skeptical22%

Among the small group reaching this question, 38% say the mitigation framing resolves their concern fully or somewhat.

Olive pomace mitigation: "cleaner version, processed with 80% fewer chemicals"

Q13.5 conditional · n=76
Response%
Yes, fully resolves my concern24%
Yes, somewhat30%
No change34%
Still skeptical12%

Among the small group reaching this question, 54% say the mitigation framing resolves their concern fully or somewhat.

Read

Three observations on risk: (1) only 28.6% of consumers raise any Fera concern at first reveal; (2) of those who do, 76% say it wouldn't change their behavior — leaving just 3.9% of the full sample with concerns severe enough to change eating; (3) for those who do reach the mitigation probe, deeper context resolves concern for roughly half. The risk is real but contained, and the mitigation framings address it for a meaningful share.

Gen Z deep dive The dream-headline question

For Gen Z, dropping seed oils is a credible category-level behavior driver

Q12.2 explicitly removed Chipotle from the frame and asked respondents to imagine fast-casual restaurants generally moving away from seed oils. The Gen Z–vs–other-ages gap is meaningful.

Gen Z (18–27) · n=262
47%
would actually change behavior — go more often or try places they currently avoid — if fast-casual restaurants stopped using seed oils.
Other ages (28+) · n=448
38%
say the same. The behavioral-conversion gap between Gen Z and older cohorts is 9pp — meaningful, repeatable across cuts.
"If fast-casual restaurants stopped using seed oils, how would that change your behavior?" by age bucket
Response 18-27 (Gen Z)
n=262
28-43
n=186
44-59
n=143
60+
n=119
I'd go more often 35% 33% 23% 15%
I'd try places I currently avoid 12% 12% 18% 8%
Change opinion not behavior 20% 19% 13% 16%
Wouldn't matter to me 33% 36% 46% 61%
Any behavior change (top-2) 47% 45% 41% 23%
"Wouldn't matter to me" 33% 36% 46% 61%
Headline read

A 9pp gap on behavioral intent and a 13pp gap on indifference both run in Gen Z's favor: Gen Z's 33% "wouldn't matter to me" is meaningfully lower than the 46% registered among other ages. The dream-headline framing — "Want to win Gen Z? Drop seed oils." — is supported by the data, especially when paired with the consistent Gen Z out-performance on the KPI lift (next table) and frequency lift measures.

Gen Z out-lifts older cohorts on every Food-with-Integrity KPI

The Pillar 1 lift data, sliced by Gen Z vs. other ages. Gen Z shows larger pre/post movement on all four attributes.

KPI lift · Gen Z vs. other ages · % rating Good or Extremely well
Attribute Gen Z
n=263
Other ages
n=451
Gen Z over-lift
Real food 74% → 79% +5.1pp 77% → 78% +1.3pp +3.8pp
Ingredient transparency 68% → 76% +8.3pp 73% → 77% +4.0pp +4.3pp
Better-for-you food 62% → 73% +11.5pp 69% → 77% +8.3pp +3.2pp
Environmentally responsible sourcing 59% → 72% +13.0pp 62% → 71% +9.5pp +3.5pp

Frequency lift among Gen Z: 66% would go more often if Chipotle went 100% seed oil-free (vs. 53% among other ages — a 13pp gap). Commitment likelihood among Gen Z: 52% more likely vs. 49% among other ages.

Bonus · category-level probe

Seed oils vs. organic: a side-by-side category test

Q12.1 and Q12.2 ran the same behavioral question with two different category-wide changes. They are useful as a benchmark — does dropping seed oils move consumer behavior more than going more organic?

If fast-casual restaurants used more organic ingredients...

Q12.1 · n=706
Response%
I'd go more often34%
I'd try places I currently avoid15%
Change opinion not behavior16%
Wouldn't matter to me35%

Any behavior change (top-2): 49%

If fast-casual restaurants stopped using seed oils...

Q12.2 · n=710
Response%
I'd go more often29%
I'd try places I currently avoid12%
Change opinion not behavior18%
Wouldn't matter to me41%

Any behavior change (top-2): 41%

Read

The two category-level questions return broadly comparable signals: 49% any-behavior-change for "more organic" vs. 41% for "stopped using seed oils." The "seed oil" message is at parity with the organic message in driving stated behavior change — a useful equivalence given organic has decades of consumer-recognition tailwind that seed oil-free does not yet have.

Limitations & open questions

Where the data is weakest, and what we'd want to know more about

Every survey has limits. We've worked through the dataset honestly and want to flag the places where the findings deserve hedging, additional context, or follow-up before they enter the Chipotle conversation as settled facts.

  1. 01

    Sample composition is not nationally representative

    37% of the sample is Gen Z (vs. ~17% of U.S. adults); 55% are Frequent Chipotle customers (likely <30% in the U.S. QSR-using population). Aggregate stats over-state both Gen Z enthusiasm and Frequent-customer ceiling effects relative to a nationally projectable sample. Use the segment cuts (Gen Z vs. Other ages, Frequent vs. Infrequent) for any projection-style argument; treat the aggregate numbers as descriptive of this sample.

  2. 02

    Frequents dip slightly on "real food" — and the dip is statistically significant

    Among Frequent Chipotle customers, the post-exposure "real food" rating moves from 93% to 89% top-2 box (-3.9pp, p<0.05). The absolute movement is small and probably reflects ceiling-effect noise, but the result is not zero. A plausible second reading: a small share of loyalists read "we're switching oils" as an implicit admission that the previous oils weren't quite "real food" — which raises a question rather than reinforces a belief. Worth understanding through follow-up if Chipotle's research team probes it.

  3. 03

    "Seed oil-free" does not outperform "without hexane / solvents" at the consumer level

    Both messages score 53% top-2 appeal, with means within 0.02 of each other (3.48 vs. 3.46). The hypothesis going in — that "seed oil-free" framing would beat solvent-free framing — is not supported. The pitch reframes this as comms flexibility, which is honest. But there's a sharper read worth carrying: the "seed oil" concept may not be carrying unique consumer-side weight beyond what "cleanly processed / solvent-free" already conveys. If true, the long-run category-positioning bet around the term "seed oil" rests more on cultural momentum than on the term itself being descriptively differentiated to consumers.

  4. 04

    Stated intent inflates real behavior — apply the standard discount

    The 58% "would visit more often" and 56–61% "would pay something extra" measures are stated-intent, not observed behavior. The standard QSR discount for intent-to-behavior translation is roughly 30–60%. The directional signal here is robust; the projectable magnitude is not. Don't pitch the 58% as "incremental visits"; pitch it as the relative comparison against the <10% who would visit less often.

  5. 05

    No control / no holdout — we can't fully isolate the seed-oil effect

    We didn't test "imagine Chipotle made a different positive change" (a salad bar, a regional sourcing initiative, a new menu item). Some share of the KPI lift is likely attributable to the framing of any positive food-quality change, not to seed oils specifically. The Q12.1 organic-comparison probe (49% any-behavior-change for organic vs. 41% for seed oils) provides a partial calibration and suggests the two attribute changes drive similar consumer movement — useful context, but not a clean control.

  6. 06

    Seed oil-free is the lowest-ranked of six food-integrity attributes tested

    In Q5.1 (personal importance), seed oil-free cooking lands at 43% — last among the six attributes tested. The gap to non-GMO (49%) is within margin of error; the gap to "no artificial flavors/colors/preservatives" (62%) is substantial. We've framed this as a trajectory story (and it likely is one), but Chipotle's research team will see the bottom-of-list result. Be ready to address it head-on with the cultural-momentum and category-vocabulary-age framing rather than relying on the close gap to non-GMO alone.

  7. 07

    "No preference" pool on Q13.7 is 46% — a large and ambiguous middle

    We frame the 44% prefer-new vs. 10% prefer-current finding as a 4:1 win. That holds among the engaged. But 46% of respondents say "no preference," and that pool could reflect either (a) genuine openness to either oil — which is what the pitch implies — or (b) low engagement on the topic and a coin-flip default. The data doesn't let us cleanly separate the two interpretations. The 4:1 ratio is real among those with views, but is a smaller share of the total sample than headline framing suggests.

Bottom line

None of the seven limitations above invalidates the core direction of the findings. Every measure runs net-positive, and the asymmetry between upside (50%+ become more likely) and downside (3.9% behavioral risk) is robust to all reasonable interpretations. But the limitations are real, and Chipotle's research team will notice some of them on their own. Better to surface them in the pitch than to be caught flat-footed in Q&A.

Methodology

Sample, fieldwork, the conversational survey design, and analytical notes

This study uses a conversational, AI-moderated survey method that differs from traditional fixed-form panels in several material ways. We've spelled out what is standard and what is non-standard below so the methodology can be explained fluently in the Chipotle conversation.

How the conversational survey method works

Research instrument
AI-moderated chat conversations, not a multi-page form. Each respondent is routed to a one-on-one chat with an LLM-driven interviewer that asks the planned survey questions in conversational form, listens to the response, and probes intelligently when the answer is ambiguous, incomplete, or interesting. The conversation guide is authored ahead of time (it's the same fixed protocol every respondent moves through), but the delivery is dynamic.
"Continuous" fielding
Respondents enter the survey on a rolling basis through Zero Acre's panel infrastructure rather than in discrete weekly waves. Once the field period opens, completes accumulate continuously until the project hits its sample target or the field window closes. The "continuous" label refers to fielding flow, not to interview length — each individual interview takes 12–18 minutes and runs to completion in a single session. There is no longitudinal "panel returns next week" component.
What's standard about it
The conversation guide is fixed; the structured-response options (5-point scales, single-select buckets) are pre-defined; the screening logic is rigorous; the sample is drawn from a verified online panel; quality controls catch low-effort respondents. The same evidence-grade considerations as a Pew, YouGov, or Qualtrics-fielded study.
What's non-standard about it
Three things differ from a traditional fixed-form panel. (1) The AI interviewer can ask clarifying follow-ups when an answer is unclear — which means open-ended responses are typically richer and more probed than a panel form would produce, but with some variability in exactly how a given respondent was probed. (2) Structured responses are elicited conversationally ("on a scale of very poor to extremely well, how would you rate…") rather than ticked from a radio list, so the AI parses the response into the structured option after the fact. (3) Conditional logic flows from the AI's read of the response, not from form branching — e.g., the Fera concern severity probe (Q6.2) fires only if the AI reads the open-ended Q6.1 response as expressing concern.
What this means for the data
Strengths: open-ended responses are substantially richer than a typical panel (see the 2,838-record verbatim library); respondents engage more deeply because the format feels human; survey fatigue is lower across a 12–18 minute interview. Trade-offs: response variability is slightly higher than a fixed form because not every respondent gets exactly the same follow-up probing; the structured-question parse rate (97–98% here) is high but not the 99.9% of a radio button. The trade-offs are well-suited to a study like this one where both quantitative segmentation and qualitative themes matter.

Sample & fieldwork

Sample size
714 completed conversations across two project batches (proj863 + proj873). Twelve respondents appeared in both batches with identical screening responses; those are deduplicated to a single row in the analysis. The exceeding-target sample (~200 over the 500-target package) resulted from supplementing proj863 with proj873 to balance the Frequent / Infrequent split toward 50/50.
Field period
April–May 2026. Continuous flow within the window; no waved fielding.
Audience & screening
U.S. adults age 18+ who eat at restaurants at least occasionally (S1 age, S2 restaurant-frequency screens). No restrictions on diet, household income, or geography beyond U.S. residence. Chipotle visit frequency (S3) is captured for downstream segmentation but is not a screening criterion.
Frequency buckets
Frequent (n=399): respondents who selected "At least a couple times a month" or "Once every few months" on S3 Chipotle visit frequency. Infrequent (n=315): "Once every year or two" or "Never." Final 55/45 split is within the ±5pp tolerance of a 50/50 design target.
Age buckets
18–27 (Gen Z, n=263, 37% of sample) · 28–43 (n=187, 26%) · 44–59 (n=143, 20%) · 60+ (n=121, 17%). Sample tilts younger than the U.S. adult population — see Limitation #1.
Age × frequency interaction
Strong confound: 60+ respondents are 83% Infrequent; 28–43 respondents are 79% Frequent. Read age cuts and frequency cuts as overlapping rather than independent. Where a single dimension is reported (e.g. "Gen Z lift"), the directional finding holds in the matched-segment cut as well — but absolute pp values are not directly additive across the two dimensions.

Analytical pipeline

Data extraction
Each session arrives as a log of assistant_message / user_message turn pairs. A purpose-built parser walks the conversation, anchors each survey question by the AI's prompt wording, and extracts the user's response from the following turn. For matrix questions, the parser handles both line-by-line responses ("Real food: Good. Ingredient transparency: Fair.") and blanket responses ("All matter a lot").
Question types fielded
Three structured-question formats: matrix ratings on 5-point qualitative scales (Q3.1 / Q10.1 KPI ratings; Q5.1 food-integrity importance; Q7.1 message appeal), single-select options (WTP ladders, frequency, commitment, category-level behavior, ingredient count, net preference), and conditional severity / yes-no probes (Q6.2 / Q6.3 / Q13.2 / Q13.3 / Q13.5). Open-ended follow-ups captured throughout — 2,838 of those are now compiled in the standalone verbatim library.
Parse rate
Structured response capture: 97–98% across the 20+ quantitative questions analyzed. The ~2% loss is split between respondents who gave genuinely uninterpretable answers (e.g., "ok" to a 5-point scale) and a small parser miss rate for edge-case phrasings. Conditional questions (concern severity, mitigation effectiveness) fire only for respondents who expressed an underlying concern; lower response counts there are intentional, not a parse miss.
KPI lift basis
Top-2 box ("Good" + "Extremely well" on the 5-point performance scale) is the primary headline measure. Statistical significance on each pre/post lift is tested with McNemar's paired-binary test on the same-respondent top-2 box change (since each respondent answered both Q3.1 and Q10.1). Asterisks in Pillar 1 tables: * p<0.05, ** p<0.01, *** p<0.001. Cells without asterisks are not statistically distinguishable from zero change.
Margins of error
At 95% confidence for simple proportion estimates: aggregate (n=714) ±3.7pp · Frequent (n=399) ±4.9pp · Infrequent (n=315) ±5.5pp · Gen Z (n=263) ±6.0pp · 60+ (n=121) ±8.9pp. For paired pre/post lifts the relevant test is McNemar's (see above); MoEs above are for single-point proportions.
Rounding conventions
Frequencies and percentages may not sum to 100 due to largest-remainder rounding applied to ensure pie/distribution charts hit exactly 100. Pre/post lift figures are paired-respondent where the same respondent answered both rounds.
Reproducibility
Raw conversation logs, the structured parse, the segmentation joins, and the significance calculations are all retained internally and available on request. The verbatim library released alongside this report covers the four open-ended questions on Fera and pomace; additional open-ended fields (Q3.2, Q4.1, Q5.2, Q8.1, etc.) can be released the same way if needed.

Zero Acre Farms · Consumer Research · May 2026

Zero Acre Farms · Internal research · 714 completed surveys · April–May 2026 fieldwork

0