Trust
Methodology
Every tool on NorthStark is scored against the same 25-criteria rubric below, and every fact — pricing, features, sentiment quotes — comes from a real source: official pricing pages, vendor documentation, or independent reviews on G2, Capterra, Trustpilot, Reddit, and Hacker News. We don't fabricate quotes or invent numbers. Where something isn't publicly documented, or a tool hasn't been scored against a given criterion yet, the page says so instead of guessing.
The 25-criteria rubric
Every tool is judged against the same 25 criteria, grouped into six categories below. Each criterion is scored 0-5 with a note explaining the score, so a number is never presented without the reasoning behind it. We're re-scoring tools against this rubric in batches — a tool page will show whatever has been scored so far, clearly labeled, rather than padding out the rest with guesses.
Channel Coverage
Number of native channels
Email, live chat, WhatsApp, SMS, Instagram DM, Facebook Messenger, voice.
True omnichannel unification
Do all channels land in one thread per customer, or do agents juggle separate inboxes per channel.
AI Capability
AI availability
Is there a genuine AI agent, or just canned "smart replies."
AI resolution accuracy
Measurable rate of correctly resolved tickets without escalation — the number vendors most inflate.
AI training method
Does it learn from your help center/docs automatically, or require manual intent-building.
Hallucination controls
Can it cite sources, restrict itself to approved knowledge, or does it improvise.
Automation depth
Simple if/then rules vs. multi-step agentic workflows (routing + tagging + resolution + follow-up).
Sentiment/intent detection
Can it flag frustrated customers or urgent issues automatically.
Inbox & Agent Experience
Unified inbox
Single pane across all channels, one customer view.
Live chat option alongside AI
Seamless AI-to-human handoff without losing context.
Internal collaboration tools
Internal notes, @mentions, assignment rules.
Canned responses/macros
Reusable templates for agents.
Native Integrations
Shopify
Order lookups, refunds, order status inside chat.
WooCommerce
Native WooCommerce integration depth.
WordPress
As a website widget, not just a plugin embed.
Wix
Native Wix integration depth.
Squarespace
Native Squarespace integration depth.
Webflow
Native Webflow integration depth.
Make.com / Zapier / n8n
No-code workflow triggers.
CRM integrations
HubSpot, Salesforce, Pipedrive sync.
Commerce & Business Logic
Order & payment data access
Can the bot actually see order status/tracking, not just talk about it.
Multi-language support
Auto-detect and respond in the customer's language.
Reporting & Scale
Analytics depth
Resolution time, CSAT, deflection rate, agent performance dashboards.
Pricing model transparency
Per-seat vs. per-resolution vs. per-conversation — this materially changes cost at scale.
API/webhook extensibility
Can you build custom logic on top, or are you locked into the vendor's workflow builder.
Legacy 5-point scorecard
Tools we haven't re-scored against the 25-criteria rubric yet still carry an older, simpler 5-dimension scorecard. It's being phased out as tools are migrated to the fuller rubric above, but it's documented here for as long as it's still live on any tool page.
Ease of Setup
How fast a team can go from signup to a working, customer-facing deployment — based on documented onboarding flows and what independent reviewers report about time-to-launch.
AI Quality
How well the AI actually resolves conversations without hallucinating or losing context, based on independent review data (G2, Trustpilot, Capterra) where available, and product design signals (guardrails against stale data, escalation logic) where it isn't.
Omnichannel Support
How many real support channels (web chat, WhatsApp, email, voice, social DMs) the tool covers natively, versus bolted on or missing entirely.
Pricing Value
Whether the published (or documented) pricing is predictable and fair for what you get — usage-based AI billing that can spike unexpectedly scores lower here, even if the base price looks cheap.
Vendor Support Quality
How responsive and reliable the vendor itself is, based on independent review volume and sentiment, not vendor marketing claims.
Freshness & "last verified"
Every tool, comparison, and alternatives page carries a "last verified" date — the last time someone actually re-checked pricing and facts against the live source, not just when the page was first published. Pricing and product details change often in this category; a stale review is worse than no review.
Who writes this
NorthStark is written and scored by Maxwell Timothy, who works in customer support tooling. Every tool is scored against the same rubric, and no tool pays for placement, a better score, or inclusion.