How AI Works Service Research ↳ Protein Powder ↳ Mattress About Get in Touch
Category: Home & Sleep AI Surface: ChatGPT Date: Feb 19–20, 2026 Regions: North America + Thailand Query set: Broad + constrained queries Variable tested: Regional context

How AI Structures
"Best" Mattress Recommendations

A controlled experiment mapping how ChatGPT selects and frames mattress brands across broad and constrained "best" queries — run across two regional contexts. The findings reveal a fundamentally different selection model from commodity categories.

Core Finding
In one sentence

In the mattress category, AI does not crown a single winner. Instead, it distributes visibility across structured segments defined by region, price tier, and functional intent — making brand eligibility conditional and role-dependent, not reputational.

This stands in direct contrast to commodity categories like protein powder, where a single consensus anchor (Optimum Nutrition) dominated 4 of 5 queries. In mattresses, no brand appeared across the majority of queries. Dominance was fragmented — and that fragmentation is not a flaw in the data. It reflects category structure.

Regional Variation

The same query. Completely different brands.

This experiment was run in two network environments. Identical queries returned materially different brands depending on inferred geography — but the underlying structural logic remained consistent. This reveals an important finding: AI applies a regional eligibility filter before ranking logic is applied.

Thai IP Context (THB pricing)

Brands surfaced

  • Ashley Sleep Essentials
  • Novilla Hybrid
  • Sealy Dreamlife
  • Zinus Cooling Gel Hybrid
  • Siena Memory Foam
  • PlushBeds Cooling Gel
North American Network (USD pricing)

Brands surfaced

  • Helix Midnight
  • Nectar Premier
  • Puffy Lux Hybrid
  • Saatva Classic
  • Tempur-Pedic Adapt
  • Purple Restore

Brand pools changed completely. Category segmentation structure did not. This means AI doesn't have brand loyalty — it has regional eligibility gates.

Brand Frequency

No single brand dominated

Unlike protein powder where Optimum Nutrition appeared in 4/5 queries, no mattress brand achieved cross-query dominance. This is not instability — it is category-level behaviour. High-consideration, high-personalization purchases produce distributed authority, not consensus anchors.

Top brands by query appearances (North America)
Helix Midnight
2/5
2
Nectar Premier
2/5
2
Purple Restore
2/5
2
Tempur-Pedic
2/5
2
Siena Memory Foam
2/5
2
All others
1

Notable absences: Casper, Tuft & Needle, Leesa, Brooklyn Bedding, Avocado Green, Sleep Number, and Stearns & Foster did not appear in any query — despite strong brand awareness and market presence.

Pattern Analysis

What we found

Five structural patterns emerged consistently across queries and regions. These are not tactics — they are signals AI appears to reward in a high-consideration category.

Pattern 01

Segmentation over singular dominance

ChatGPT consistently divided mattress results into structured, labeled tiers — "Best Overall," "Premium Picks," "Budget & Value," "Cooling Specialists," "Hybrid Comfort." Even broad queries did not produce a single crowned brand. Visibility was distributed across buckets.

Segmentation appeared in every query — broad and constrained — and remained stable across both regional contexts even when brands changed completely.

Pattern 02

Regional context overrides brand dominance

Geography influenced which brands were eligible before ranking logic was applied. Brand pools changed entirely based on inferred location. But category structure — the segmentation logic — remained consistent across regions. Regional eligibility is a structural gate, applied before quality or authority is evaluated.

"Best mattress for side sleepers" · Thai IP: Novilla, Zinus, Sealy · North America: Helix, Nectar, Puffy, Purple. Same structure. Different brands.

Pattern 03

No universal category anchor emerges

Unlike protein powder (where Optimum appeared in 4/5 queries), no mattress brand appeared across the majority of queries. Even dominant DTC brands like Helix, Nectar, and Purple appeared only within relevant contexts. Mattresses are high-price and highly personal — AI appears structurally unwilling to default to a single universal winner.

Maximum brand frequency: 2/5 queries. Protein powder comparison: Optimum Nutrition 4/5. The difference reflects category structure, not data noise.

Pattern 04

Price-tier alignment is enforced

Budget queries surfaced exclusively lower-priced brands. Premium framing appeared only in higher-tier contexts. Mid-tier brands with ambiguous positioning showed reduced cross-query appearance. AI appears to enforce price congruence between query intent and brand positioning.

"Best affordable mattress" → Siena, Zinus, Ashley, Today Mattress. "Best cooling mattress" → Purple Restore, Tempur-ProBreeze. No overlap.

Pattern 05

Feature-to-intent matching drives inclusion

Queries with functional constraints triggered material-specific language. Brand selection aligned with construction narrative. Brands without explicit feature-to-problem alignment did not surface in constrained contexts.

Back pain → zoned coils, lumbar support · Side sleepers → pressure relief, medium firmness · Cooling → gel foam, PCM cooling tech, airflow, heat dissipation

Structural Model

The layered logic AI applies

Brand visibility in the mattress category appears governed by a sequence of filters applied before any brand is selected. Brands that fail a filter at any stage are excluded — regardless of quality, awareness, or market share.

Filter 01

Regional eligibility

Is this brand available and recognised in the inferred geography? If not, excluded before ranking begins.

Filter 02

Price-tier alignment

Does the brand's pricing signal match the query's intent? Budget brands excluded from premium queries. Premium brands excluded from budget queries.

Filter 03

Functional intent match

Does the brand's construction narrative align with the query's functional goal? Missing alignment = exclusion from constrained queries.

Filter 04

Tier segmentation

Which bucket does the brand fit? AI distributes authority across labelled segments rather than crowning a single winner.

Filter 05

Brand selection

Within the qualified pool, brands are selected. Only brands that passed all prior filters are considered.

Cross-Category Contrast

Protein powder vs. mattresses

Comparing these two experiments reveals something important: AI does not have one universal selection behaviour. It has category-dependent selection logic. The difference between these two categories illustrates how AI adapts its recommendation model based on the nature of the purchase.

Protein Powder — Commodity

Consensus anchor logic

One brand (Optimum Nutrition) dominated 4/5 queries. Broad queries favor widely trusted, general-purpose brands. Constraints applied hard filters but anchor dominance was strong.

Mattresses — High-Consideration

Stratified eligibility logic

No brand appeared across the majority of queries. Visibility was distributed across tiers. Eligibility was conditional — driven by region, price, and functional fit.

The broader implication

Commodity categories may reward consensus anchors. High-consideration categories may favour stratified eligibility models. This means the strategy for improving AI visibility is not universal — it must be tailored to the structural behaviour of your specific category.

What This Means for Brands

Signal gaps that determine visibility

Tier identity

Define your bucket

AI segments results into labeled tiers. Brands without a clear tier identity — budget, mid-range, premium, or specialist — may fail to qualify within any recommendation bucket.

Regional presence

Geographic eligibility

AI filters regionally before ranking. Brands with weak signals in a target geography may be excluded entirely, regardless of product quality or global awareness.

Price-tier consistency

Avoid price ambiguity

Oscillating between discount and luxury framing creates eligibility gaps in both query types. Pick a lane and reinforce it consistently.

Feature-to-intent alignment

Match construction to query

AI matches material-level claims to user intent. If your brand doesn't explicitly signal zoned coils, pressure relief, or cooling tech, it won't qualify for those queries.

Closing Observation
The takeaway

In high-consideration categories, AI does not appear to crown singular category leaders. Instead, it distributes visibility across structured segments defined by region, price tier, and functional alignment. Brand eligibility is conditional and role-dependent. Structural clarity — not broad reputation alone — appears to determine inclusion.

Run Your Own Experiment

Find out where your brand stands.

Every category behaves differently. We'll run a controlled experiment on your brand and tell you exactly what AI sees — and what it doesn't.