How Smartreview AI™ measures software trust — and why you can rely on it. Methodology version 1.0, published July 11, 2026.
Proprietary Rights and Infringement Notice. The Trust-IQ Score™ methodology — including its scoring architecture, dual-pillar model, control registers, question banks, evidence-grading model, weightings, thresholds, and evaluation processes — is the exclusive intellectual property of Smartreview AI Pvt Ltd and is protected under applicable patent law (including patent filings made by Smartreview AI Pvt Ltd), trademark, copyright, and trade secret laws of India and other jurisdictions. "Smartreview AI" and "Trust-IQ Score" are trademarks of Smartreview AI Pvt Ltd.
This document is published for transparency. It grants no license, right, or permission of any kind, express or implied. Without our prior written authorization, you must not: (a) copy, reproduce, adapt, translate, or create derivative works of this methodology; (b) use it, in whole or in part, to build, train, benchmark, or improve any competing or similar scoring, rating, review, audit, or certification product or dataset; (c) reverse-engineer, probe, or attempt to derive the underlying models, weightings, thresholds, or detection techniques; (d) use our trademarks, badges, or scores except as expressly licensed; or (e) ingest this document or our data into any machine-learning or AI training pipeline.
Any infringement, misappropriation, or unauthorized use will be pursued to the fullest extent permitted by law, including claims for injunctive relief, damages, and recovery of costs and attorneys' fees, together with suspension or termination of any account, listing, certification, or partnership, and referral to relevant authorities. Report suspected infringement to contact@smartreviewai.io.
Most software ratings ask you to trust a number without telling you where it came from. We take the opposite view: a trust score you cannot examine is not a trust score. This document explains, in plain language, how the Trust-IQ Score™ is constructed, what evidence counts, what never counts, and the safeguards that keep the score independent of money.
This is a public document, written for both software buyers (Customers) and the organizations we evaluate (Customers/Partners). It describes our methodology at the level of structure, principles, and safeguards. The detailed control texts, question banks, model parameters, and fraud-detection techniques are deliberately not published — disclosing them would teach bad actors how to game the score, which would harm the buyers this system exists to protect.
The Trust-IQ Score™ is a 0–100 composite trust assessment of a business software product and the organization behind it, built from two equally weighted, independently computed pillars:
Trust-IQ Score™ = Integrity AI (max 50) + Auditor AI (max 50). Neither pillar can compensate for the other: a beloved product from an unauditable company, or a well-audited company whose users are unhappy, will show it. The two pillars are computed by separate engines from separate evidence and are combined only at the final step.
The two pillars move on different clocks. The Auditor AI pillar is established through a formal audit conducted once each year (and reassessed when material events warrant it). The Integrity AI pillar is alive: it moves throughout the year as verified users contribute new experiences and as older experience ages out of the score.
The published Trust-IQ Score™ therefore varies throughout the certification period. Smartreview AI does not control this movement: the Verified User Trust component is determined solely by verified user input processed through this published methodology, and we cannot set, steer, freeze, or adjust it — for any organization, paying or not. Our only interventions are integrity interventions: removing input that fails verification, authenticity, or conflict-of-interest checks. This is by design; a score no one can steer, including us, is the point.
Integrity AI captures the real-world experience of using the product, across five domains of 10 points each:
| Domain | What it measures |
|---|---|
| Product Capability | Does the product do what it claims, for the jobs it is bought for, and does it keep improving? |
| Adoption Experience | How quickly and easily do real teams get to value — implementation, learning curve, day-to-day usability? |
| Technical Operation | Performance, reliability, and integration behavior as actually experienced in production — not as advertised. |
| Administration & Support Experience | What it is like to govern, secure, and administer the product, and how support behaves when something breaks. |
| Commercial Truth | Whether the commercial relationship matches what was sold — pricing behavior, renewal conduct, and whether the promised outcome was realized. |
Not every submitted opinion carries weight. Each response is weighted by four factors, all disclosed here in principle:
Auditor AI evaluates the organization behind the product against a defined register of mandatory controls across five domains of 10 points each:
| Domain | What it measures |
|---|---|
| Software Development Life Cycle | How software is built, tested, secured, and released — the engineering discipline behind the product. |
| Product Management | Whether the organization keeps its product word: release practices, documentation currency, versioning and deprecation conduct, accessibility. |
| Organization Structure | Whether the company is what it says it is: corporate transparency, named accountability for security and privacy, governance of people and subprocessors, continuity planning. |
| Security & Privacy | Security posture, certifications validated for scope and legitimacy, data protection, vulnerability and incident management, privacy program — including the organization’s position on using customer data for AI training. |
| Customer Support | The support commitments an organization makes and operates: service levels, availability and escalation, status transparency, and fair access to support. |
Every control is scored against evidence, and evidence is not all equal. We grade evidence by authority:
Evidence also expires: each type of evidence has a defined freshness window, after which it must be renewed or the associated credit decays.
The majority of Auditor AI controls are verifiable from independent, public, or directly observable sources without any cooperation from the evaluated organization. This means an assessment cannot be improved by withholding information, and it means we evaluate organizations that never engaged with us against the same standard as those that did. Participation lets an organization supply additional evidence and context; it never buys leniency.
The Trust-IQ audit is conducted once each year. On successful completion of the annual audit and satisfaction of the published certification criteria — including minimum overall and domain-level requirements and the integrity gates described above — a Trust-IQ certificate and badge are issued once a year, valid for twelve months from issuance.
An annual certificate does not mean a frozen score. Because the Verified User Trust pillar moves continuously (see Section 02), the published Trust-IQ Score™ varies throughout the certification period; the badge attests that the organization passed the annual audit and remains within the certification criteria, while the live score on smartreviewai.io is always the current, authoritative number. Certification remains subject to continuous monitoring for the full period and is suspended or revoked mid-cycle if criteria are no longer met, if material adverse events occur, or if manipulation is detected. Recertification requires a fresh audit each year — certification never renews automatically.
Certification is voluntary and never guaranteed. Every legitimate badge links to a live verification page on smartreviewai.io showing current status; a badge without a working verification link should not be trusted.
The single most important safeguard in this methodology is architectural, not contractual. The Integrity Firewall is a technical and organizational separation that guarantees:
Annual audit, living score. The Auditor AI pillar is re-audited annually and reassessed on material events; the Integrity AI pillar recomputes continuously as verified user input arrives and ages. The published Trust-IQ Score™ is always the current assessment, not a historical award.
Human review of adverse findings. No adverse finding about an identifiable organization is published on automated output alone. Adverse findings require human adjudication, are anchored to preserved evidence, and absence of evidence is always published as "no record found in [named source] as of [date]" — never dressed up as either innocence or guilt.
Right of reply on facts. Any evaluated organization may notify us of a factual inaccuracy. Where we agree an error of fact occurred, we correct it and recompute. What is not open to negotiation is the opinion: the methodology's conclusions, weightings, and the score itself cannot be argued, purchased, or pressured upward.
Versioned methodology. This methodology is versioned. Material changes are published with a new version number and effective date, so scores are always interpretable against the rules that produced them.
Some widely used trust signals are excluded from the Trust-IQ Score™ on purpose, because they measure marketing budgets, company size, or sentiment — not trustworthiness:
Publishing this exclusion list is itself part of the methodology: you should know not only what moves a score, but what cannot.
A Trust-IQ Score™ is Smartreview AI's independent, good-faith, evidence-based assessment, produced by the published methodology from the data available at the time of computation. It is designed to be the most rigorous single input to a software trust decision available anywhere.
It is not a warranty, a guarantee of security or fitness, an endorsement of any transaction, or a substitute for your own procurement, legal, and security diligence. No methodology can see everything: evidence can be incomplete, organizations change, and a score is a continuously maintained opinion — held to a published standard — not a promise about the future. We say this plainly because overstating what a score means would be its own form of dishonesty.
This methodology is owned and operated by Smartreview AI Pvt Ltd. Questions about the methodology, correction requests, and reports of suspected review manipulation or score gaming are welcome:
Smartreview AI Pvt Ltd · www.smartreviewai.io · contact@smartreviewai.io · Support: support@smartreviewai.io
© 2026 Smartreview AI Pvt Ltd. "Smartreview AI" and "Trust-IQ Score" are trademarks of Smartreview AI Pvt Ltd. This document describes methodology structure and principles; detailed control texts, question banks, weights, thresholds, and detection techniques are proprietary and confidential.