F.01
Pantry Replenishment
Stockpiling cadence, off-cycle purchases, substitution frequency, and bulk-channel migration.
9 signalsThe outlier is not noise. It is the signal.
We treat deviation as a first-class measurement — a defensible philosophy of how consumer behavior actually moves, scored weekly across forty-one distinct signals and nearly ten million Americans.
By the research team at Abnormis, Inc.
Brooklyn · London · São Paulo
Chapter I
The Outlier Index, in four methodological pillars.
Mainstream trackers are designed to converge on the median consumer because that is where purchasing volume lives. The Outlier Index is engineered to do the opposite: locate the consumers whose behavior consistently deviates, weight the signals that distinguish them, and surface the deviation eleven months before the median catches up. What follows is the full architecture — the four pillars any independent reviewer can audit before booking a diagnostic call.
-
01
Signal Taxonomy
Forty-one behavioral signals, each defined against a published operationalization document. A signal is in scope only if it can be observed at scale, replicated across the panel, and tied to a category-level purchase or media decision. Anything that cannot pass those three tests is excluded, regardless of intellectual interest.
- 41 scored signals
- 6 behavioral families
- Refreshed every Monday, 06:00 UTC
-
02
Weighting Protocol
Weights are not editorial. They are derived empirically from historical lead-lag correlations between each signal and category-defining brand outcomes — repeat purchase, channel migration, brand switching. The protocol is re-fitted every quarter against the prior twelve months of outcome data, and any weight whose predictive contribution falls below threshold is retired rather than carried forward.
- Empirical, not editorial
- Quarterly re-fit
- No legacy carry of weak signals
-
03
Panel Calibration
The Index does not infer deviation from a generic population. It is scored against a 480,000-consumer high-deviation panel recruited on behavioral screening criteria — longitudinal, consented, and weighted to mirror the U.S. census on age, income, region, and household composition. The calibration layer is what distinguishes measured deviation from sampling noise.
- 480,000 opted-in panelists
- Continuous longitudinal enrollment
- Census-weighted scoring overlay
-
04
Validation & Audit
Every Index release is reproducible. Backtests against the prior thirty-six months of brand outcomes are published internally; backtests against the ESOMAR 28 standard are audited quarterly by an independent third party. The methodology is also citation-reviewed in HBR, The Economist, and the Journal of Consumer Research — a discipline that keeps the instrument honest.
- Quarterly ESOMAR 28 audit
- Independent third-party review
- Public methodology citations
Coverage Audit
Six behavioral families. Forty-one signals. One weekly score.
The Index does not claim to measure everything. It claims to measure what moves categories. Below is the family-level taxonomy — the six groupings a strategy head can audit against their own category model before commissioning work.
F.02
Channel Switching
Cross-channel basket construction, D2C trial velocity, and marketplace re-routing.
8 signals
F.03
Identity-Linked Choice
Values-driven purchase, cause-association affinity, and self-expressive brand selection.
7 signals
F.04
Discovery & Triage
Source mix, social-platform referral, peer-influence receptivity, and shelf-vs-stream decision speed.
7 signals
F.05
Price Elasticity
Promotional response asymmetry, coupon-stack behavior, and discretionary spend posture.
6 signals
F.06
Loyalty Erosion
Switching latency, defection intent, and category disengagement early-warning signals.
4 signals
Where the data comes from
A 480,000-consumer panel recruited for deviation, not representativeness.
The Outlier Index requires a sample that produces statistically meaningful deviation by construction. The Abnormis Panel is recruited on behavioral screening criteria — not on demographic convenience — and longitudinally maintained with explicit opt-in consent, quarterly re-engagement, and an independent privacy audit.
Panelists are enrolled across all fifty U.S. states and re-screened annually for behavioral continuity. The census-weighted scoring overlay that runs on top of the raw panel is what allows Index outputs to be projected back onto the 9.2 million high-deviation U.S. consumers the Index is ultimately scored against each week.
See the panel in the 2024 reportPredictive lead time, validated
0
months
The Outlier Index identifies category-disrupting behaviors an average of eleven months before they appear in mainstream consumer trackers — the lead time that separates an early strategic move from a defensive one.
Validated against backtests across the 2019–2024 consumer panel, audited under the ESOMAR 28 standard. Source: Abnormis State of the Outlier 2024.
Governance & Audit
Three independent standards the methodology is held to.
Procurement and insights governance leads should be able to clear this page in one pass. The three standards below are the ones our enterprise clients most frequently request — and the ones the Abnormis research team has held continuously since 2021.
-
01
ESOMAR 28 — Panel & Methodology Audit
An independent auditor reviews panel recruitment, consent posture, signal operationalization, and weighting protocol every quarter. The most recent audit letter is available on request under NDA. The Abnormis panel has been continuously certified under ESOMAR 28 since 2018.
-
02
ISO 27001 — Information Security
The Abnormis information security management system is certified to ISO 27001, covering data handling, access controls, encryption posture, and incident response. Certification is renewed annually by an accredited certification body.
-
03
SOC 2 Type II — Operating Controls
Abnormis has maintained continuous SOC 2 Type II compliance since 2021. The most recent report covers Security, Availability, and Confidentiality trust service criteria, with a full observation window of twelve months.