NESNet Entropy Score
NES
Net Entropy Score
A consistency layer for every brand
Averages hide instability. NES measures it.

Net Entropy Score is a brand consistency measurement framework. We score customer experience consistency from reviews, surveys, and behavioral signals using an AI-assisted production pipeline against a paper-defined methodology. NES identifies where the experience becomes inconsistent across channels, cohorts, fulfilment, onboarding, support, and repeat usage, before that inconsistency shows up in churn, CAC pressure, weak retention, or brand decay.

NES is live in market with three products: a free Scanner with 2,000+ brand reads across 1,400+ brands, NES Sentinel, free weekly brand monitoring with drift alerts and an AI (MCP) connector, and NES Signal, a monthly public-signal monitoring subscription for operators and investors tracking 100+ brands. The 3 to 4 year vision is the coherence layer an AI-saturated world can trust: a signal of what is consistent and real, that cannot be faked.

Working Paper v1.0.1  ·  Read on SSRN  ·  Framework

In market, in numbers
0+
Brands analyzed, 2,000+ reads
0+
Brands on monthly Signal monitoring
0
Products live: Scanner + Sentinel + Signal
0+
Public signals analyzed
SSRN
Published methodology, trademark filed
What We Are Building

A consistency layer for every brand.

Most companies know whether customers are satisfied. Very few know whether the experience is consistent enough to compound.

NES turns customer voice into a structured consistency score. It classifies experience signals into five bands, identifies where variance concentrates, and surfaces the operational failure points hidden beneath healthy averages.

Today, NES is delivered through paid diagnostics that combine paper-method Predicted scoring, AI-assisted review classification, and direct customer-cohort measurement. The roadmap extends the AI production layer toward lower-friction interfaces, so the same outputs become available at productized prices for SMBs and self-serve buyers:

Predicted NES score
Review-Inferred NES score
Consistency failure themes
Convergence-gap analysis
Revenue-risk estimate
Recommended intervention map

The methodology is the moat. AI-assisted production is the capability that makes the Inferred tier fast and productized; the Measured tier remains where the deepest, most defensible reading lives.

The Problem

NPS, CSAT, and star ratings average the customer base. NES isolates variance.

A brand can have strong ratings and still be leaking growth.

Averages hide the customer cohorts where the experience is breaking. Those cohorts usually show up first in reviews, complaints, refunds, failed onboarding, support friction, marketplace variance, weak repeat purchase, and inconsistent word of mouth.

Is the experience consistent enough to compound?

That is the measurement gap.

What NES Measures

Consistency of customer experience over repeated interactions.

The core question is:

How consistent is your experience with this brand each time you use it?

Responses are classified into five experience bands.

Coherent9-10Highly consistent. The experience is reliably reproduced.
Reliable7-8Mostly consistent. Minor variation does not break trust.
Variable5-6Noticeable variation. The experience lacks a stable core.
Scattered3-4Significant variation. The experience becomes unpredictable.
Disordered0-2Highly inconsistent. The experience feels chaotic or contradictory.
NES = % Coherent − % (Scattered + Disordered)

The result is a bounded score from −100 to +100. The current working paper positions this as an exploratory diagnostic framework, with statistical validation and broader cohort testing as future work.

Three NES Readings

Predicted, Review-Inferred, Measured.

NES is delivered across three tiers, each with a defined source. The framework's gold-standard reading is Measured NES from direct customer surveys; the AI-assisted production layer makes the Review-Inferred tier fast and productized.

Tier 1 · Top-down public-signal read
Predicted NES

Predicted NES uses observable signals such as reviews, ratings, marketplace data, app-store data, public sentiment, category signals, channel concentration, trust indicators, and operating-risk proxies. Computed via the 10-component formula in Appendix A of the working paper.

Best for: cold reads, public companies, competitor comparison, investor diligence, category benchmarking.

Tier 2 · Bottom-up review-corpus diagnostic
Review-Inferred NES

Public reviews are classified into NES experience bands by a transformer-based zero-shot model (BART-MNLI), identifying where consistency breaks by channel, cohort, product line, location, or failure theme.

Best for: D2C brands, SaaS review analysis, marketplace channel variance, hospitality reviews, healthcare service feedback, app-store and support-signal analysis.

Tier 3 · Direct customer-cohort measurement
Measured NES

Measured NES uses direct survey data and, where available, first-party behavioral data such as repeat purchase, refunds, returns, cancellations, churn, support load, and reactivation. The gold-standard reading.

Best for: active clients, founder-led diagnostics, cohort-level analysis, repeat-purchase categories, post-intervention measurement.

The Convergence Gap

The gap between perception and delivery is the diagnostic.

NES compares what the brand appears to be externally with what customers actually experience.

Convergence Gap

Convergence Gap = Predicted NES − Measured NES.

A positive gap (Gap > +15) means public signals may be overstating delivered experience. That is the operational risk signal.

A negative gap (Gap < −15) means actual customer experience may be stronger than public signals suggest. That is hidden strength.

A near-zero gap (|Gap| ≤ 10) means external perception and customer reality are broadly aligned. Between those bands sits a Monitor zone (10 < |Gap| ≤ 15) for modest divergence worth tracking.

The convergence gap is where the real diagnosis lives.

What NES Finds

Where customer experience variance is concentrated.

Channel inconsistency

Customers from Amazon, Flipkart, retail, direct site, app, or referral channels experience the same brand differently.

Fulfilment variance

Delivery, replacement, shipping, installation, or packaging failures cluster in specific channels.

Onboarding inconsistency

New customers do not receive the same context, education, setup, or success path as loyal customers.

Support variance

Response time, complaint resolution, refund handling, or escalation quality differs by cohort or platform.

Product-line variance

Some SKUs or service lines carry more trust friction than others.

Cohort variance

Best customers and worst customers are not simply happier or unhappier. They often experience structurally different versions of the business.

Why It Matters

Inconsistent experience creates silent leakage.

Customer experience variance usually shows up before the board-level metric breaks. It appears first as nine recurring leakage signatures.

01
Lower repeat purchase
02
Weaker retention
03
Higher support load
04
Refund friction
05
Review decay
06
CAC pressure
07
Lower referral quality
08
Channel-specific underperformance
09
Weaker cohort economics

By the time these problems become visible in revenue, the consistency drift is already embedded in the operating system. NES is designed to surface that drift earlier.

Built For

Operators, founders, and investors.

For Founders
Find where the customer experience is breaking as the business scales.

NES helps founders see whether the issue is demand, positioning, fulfilment, onboarding, product consistency, support, or channel-specific variance.

For Operators
Prioritize the customer-experience fixes most likely to compound.

NES turns scattered customer feedback into a structured operating map and identifies the fixes most likely to reduce churn, support load, refund friction, and weak repeat behavior.

For Investors
A diligence layer for hidden operational variance.

A company can look strong from ratings, revenue growth, or brand perception while still carrying channel-level or cohort-level experience risk. NES surfaces it.

Early Traction

Validation across categories, in market.

NES is live in market with three products and early recurring revenue: a free Scanner with 2,000+ brand reads, NES Sentinel for free weekly monitoring with drift alerts, and NES Signal, a monthly public-signal subscription tracking 100+ brands. Paid diagnostics validate the model and calibrate the engine.

2,000+ reads across 1,400+ brands in the NES dataset
100+ brands now on monthly Signal monitoring, recurring
2 products live in market: the free Scanner and NES Signal
Live Measured customer-cohort engagement validating the model
14,000+ public signals analyzed to date
Diagnostic coverage across D2C, FMCG, wellness, hospitality, SaaS, healthcare, retail
Working paper published on SSRN; trademark filed in India, Class 35
Automated public-signal scoring pipeline running in production

Two products in market, early revenue across the full ladder, and a compounding dataset. Each scan and each monthly read makes the next read sharper and the standard harder to displace.

How It Works

From customer voice to consistency score.

01
Ingest
Reviews, ratings, survey responses, support themes, marketplace data, app-store reviews, and available behavioral signals.
02
Classify
Customer signals are mapped into the five NES bands: Coherent, Reliable, Variable, Scattered, and Disordered.
03
Score
NES is computed as a bounded −100 to +100 score using the framework's net-percentage formula.
04
Diagnose
Variance is mapped by channel, cohort, product line, location, onboarding path, fulfilment layer, and support theme.
05
Intervene
NES identifies the operational fixes most likely to reduce inconsistency and improve compounding behavior.
Early Access and Diagnostic Engagements

Paid diagnostics validate the model and calibrate the engine.

Paid diagnostics validate the scoring model, collect category patterns, and build the calibration base for the productization roadmap. Engagement models available today:

Full Brand Consistency Review
$49
Website read · instant report

A read on how your website, voice, and customer promise hold together. Detailed PDF with specific on-page friction and copy rewrites you can ship the same week.

  • Auto-generated, evidence-based report
  • Brand Consistency, Message Clarity, Identity Confusion, Trust Leakage, Diagnostic Confidence
  • Specific on-page friction with copy rewrites
  • Delivered instantly, emailed as a PDF
Run a scan
Measured Engagement
From $15,000
Customisable scope · Enterprise & multi-brand

For brands at scale that want the Measured NES tier with direct customer-cohort surveys, behavioral analysis, and ongoing measurement. Scope and pricing built around your industry, business model, and brand portfolio.

  • Direct customer-cohort survey via v4.0 NES instrument
  • Measured NES with statistical confidence intervals
  • Channel-split analysis across portfolio
  • Quarterly remeasurement option with leadership readout
  • Dedicated framework analyst
Talk to us
Public Score Index

Public-signal NES across 13 brands.

NES maintains a public score index of selected brands using public-signal methodology. The score index is not a claim of internal access or endorsement. It is a directional read based on public information, review signals, category context, and the diagnostic toolkit. Use it to see how the framework reads different categories and operating models.

13 brands · 5 categories · Updated May 2026
Retail, B2B SaaS, supplements, beverages, and healthcare. Each scored via the 10-component formula.

Athletic Brewing, Graza, Thorne, AlayaCare, Olipop, Huel, Eyewa, Walmart, ClickUp, Wellbeing Nutrition, Alleanza Healthcare, Kapiva, HealthKart. Component breakdowns auditable to Appendix A.

View score index

Public-signal analysis only. No internal company data accessed. Directional NES readings, not investment advice or statements of company performance. Brand names used for analytical identification only. NES is not affiliated with, endorsed by, or authorized by the companies analyzed unless explicitly stated. Full disclosures on the Score Index page.

Free Calculator

Estimate what inconsistency may be costing you.

The calculator is designed for founders, operators, marketers, and investors who want a quick read on consistency-driven leakage. It produces a directional NES read plus a 24-month revenue and investment projection, calibrated to the NES v1.0 framework.

Free · Five-question diagnostic · 24-month projection
See what your inconsistency is costing you, and what NES recovers.

No login required. The calculator is calibrated to the v1.0 framework and the 10-component diagnostic toolkit; numbers are scenario-bounded estimates.

Run calculator
Methodology

The published methodology, v1.0.1.

The NES working paper introduces the framework, score construction, convergence-gap logic, retrospective illustrations, prospective cohort observations, and limitations.

The framework is currently exploratory. Statistical validation, inter-rater reliability testing, multi-item instrument expansion, and out-of-sample prediction testing remain future work.

SSRN Working Paper · April 2026, revised May 2026

Net Entropy Score (NES): Measuring Brand Experience Consistency and Its Diagnostic Implications

Pankaj Upadhyay · Independent Researcher · Founder, Impossible Marketing

The paper introduces the Net Entropy Score framework, distinguishes Measured NES from Predicted NES, proposes the convergence gap as a diagnostic indicator, and reports illustrative application across twelve retrospective anchor observations on ten brand cases, and a five-case prospective cohort. The paper is positioned as exploratory; statistical validation, out-of-sample prediction testing, and broader cohort generalizability remain open. Limitations are explicitly stated.

Pages17
References9
Abstract ID6667158
TrademarkIndia · 14178358

Find where the experience is breaking.

If your ratings look healthy but retention, referrals, support load, or channel performance feel inconsistent, NES can identify where the variance is concentrated.

The Author

Pankaj Upadhyay

Pankaj Upadhyay is the creator of NES (Net Entropy Score). He developed the BART-MNLI zero-shot classification pipeline that produces the Inferred tier at scale, and authored the v1.0 working paper that defines the underlying methodology. The framework draws on nine years of operating depth across consumer brands in DTC, FMCG, alt-wellness, B2B SaaS, and luxury hospitality, plus the same nine years of pattern recognition across categories where the convergence-gap signature kept surfacing without a quantified read.

He writes The Impossible Brief on Substack, founded Impossible Marketing in 2017, and is based in Dubai.