how AI scores work

Every listing on mfgselect carries an AI score from 0–100. Unlike analyst ratings that optimize for enterprise consensus, mfgselect scores are calibrated for SMB reality: implementation risk, total cost of ownership, and whether a real operations team without dedicated IT can actually succeed with this software.

Scores are intentionally conservative. A score of 80 here represents a strong, confident recommendation. A score of 65 is a viable option with meaningful tradeoffs buyers should understand before committing.

the rubric — six dimensions

Each vendor is evaluated across six dimensions. Every dimension is scored 0–100, then weighted to produce the composite AI score.

Dimension	Weight	What's being measured
SMB implementability	25%	Can a business without dedicated IT successfully implement this? High: self-guided onboarding, <3 month go-live, no mandatory SI partner Mid: partner recommended, 3–6 month typical timeline, manageable risk Low: SI partner required, 6–18 month implementations, high failure rate at SMB scale SI = systems integrator — a consulting firm or implementation partner hired to configure, customize, and deploy the software. For many enterprise ERPs, engaging an SI is not optional; implementation services alone (separate from licensing) can run $50K–$400K+ and 6–18 months. A low implementability score is not necessarily a dealbreaker, but it is a significant cost and risk factor to plan around.
Purpose fit	20%	Does this software excel within its designed scope? The question is applied differently by tool type — a quoting plugin is not compared against ERP breadth; it is evaluated on whether it is best-in-class at quoting. Full ERP suites — High: purpose-built for the buyer’s vertical, native data model, domain-specific terminology Full ERP suites — Mid: solid general manufacturing fit, some vertical modules, not purpose-built Full ERP suites — Low: horizontal platform with surface-level vertical claims, key features missing or bolted on Niche MRP / ops tools — High: best-in-class for their specific market segment (job shop, food batch, apparel) — not penalized for lacking full ERP breadth Plugins & best-of-breed — High: best-in-class within their functional niche (quoting, traceability, QMS, nesting, harness design) — evaluated on functional excellence, not scope
Tech modernity	20%	What does the underlying architecture mean for your team's future? High: cloud-native, API-first, modern stack, regular release cadence, mobile-capable Mid: web-based but not cloud-native, or cloud with legacy underlying architecture Low: on-prem first, legacy core with modern UI veneer, limited or no open API
Documentation quality	15%	Can buyers and implementers learn this software without paying for training? High: comprehensive public docs, accessible API documentation, active knowledge base, community Mid: partial public documentation, basic support resources, some training required Low: documentation gated behind login or sales call, no public API docs, training is the only path
Real user sentiment	10%	What do SMB users — not enterprise analysts — say after implementation? High: consistent positive reviews from small/mid manufacturers, implementation reviews positive Mid: mixed — strong on features, complaints about implementation or support responsiveness Low: significant implementation failures at SMB scale, churn signals, support horror stories
TCO transparency	10%	Can a buyer estimate total cost before talking to sales? High: pricing published, implementation costs estimable, no material hidden fees Mid: some pricing guidance, rough ranges available through partners or community Low: “contact for pricing,” opaque implementation costs, significant surprise fees common

what the score bands mean

85–99 Exceptional for SMB — confident recommendation, clear strengths across the rubric
75–84 Strong choice — solid performer, clear strengths, manageable and known tradeoffs
65–74 Viable option — workable, but meaningful tradeoffs buyers must understand before committing
55–64 Proceed with caution — significant weaknesses in SMB context, specific fit required
<55 Not generally recommended for SMB buyers in this vertical

how scores are researched

Each score is derived from public sources, weighted in this order:

1st priority — end-user and API documentation. Publicly accessible documentation is a direct signal of vendor philosophy. Vendors who hide docs behind a sales wall are penalized. Vendors with comprehensive, searchable public knowledge bases are rewarded.
2nd priority — user reviews, filtered for SMB. G2, Capterra, and similar platforms, filtered to reviews from small and mid-size manufacturing companies. Implementation and support reviews are weighted more heavily than feature reviews. Enterprise reviews are largely excluded — they describe a different buying reality.
3rd priority — community signals. Forum activity, partner ecosystems, Reddit, LinkedIn groups, and other community sources that reveal real-world adoption patterns and pain points.
4th priority (limited weight) — vendor marketing. What vendors say about themselves is the least reliable input. It is considered for product positioning context only and does not materially influence scores.

scores are dynamic

The AI score displayed on each listing adjusts in real time based on your active filters. A vendor that is purpose-built for your specific vertical, size, and tech preference scores higher in your filtered view than in an unfiltered view. This reflects genuine contextual fit, not a different editorial judgment.

Clicking any AI score opens a Perplexity AI search pre-populated with a query about that vendor in the context of your active filters — so you can independently verify or challenge the rating.

The full per-vendor research notes — sources consulted, dimension breakdowns, and score rationale — are published at research.html.

what scores are not

Not influenced by vendor subscription status or commercial relationship
Not a measure of company revenue, market share, or analyst consensus
Not a guarantee of implementation success — your specific operation always matters
Not static — scores are updated as products, vendors, and markets change

All scores represent editorial opinion. See our full legal disclaimer for terms governing the use of this information.

challenge a score

If you believe a score is materially wrong — as a buyer, implementer, or vendor — we want to hear it. Scores improve when challenged by people with direct operational experience. Email [email protected] with the vendor name, the score you'd expect, and your reasoning.