Scoring rules · v0.2
How scoring works
We score automation tools for small teams so you can pick one. A scorecard is a weighted sheet, not a vibe. If we cannot point at a criterion, we do not invent a number.
The math, in plain English
Each tool is rated 1–5 on named criteria. Each criterion has a weight; the weights add up to 100. Contribution is rating × weight. We add those contributions up. A perfect score is 500 (5 × 100).
Make’s 410/500 means 410 out of 500 possible points on the automation card. Higher is a better fit for small teams on our criteria — not “best tool forever.”
The two cards use different criteria. Do not compare Make 410 / n8n Cloud 368 / Zapier 353 to Notion 397 / Airtable 390 / Coda 348.
n8n Self-hosted is scored on the same automation sheet so you can see the number. It is not a fourth cloud product and is not mixed into the cloud #1 / #2 / #3 rank.
Label
Not ranked yet
The table can be public; we have not issued a ranking.
Label
Approved totals
The /500 totals are approved; we have not published every 1–5 cell yet.
Label
Full scorecard
A complete /500 card with every criterion-level 1–5 cell on the page.
Status of these rules
Version 0.2, dated 7 September 2026. First scorecard totals were issued the same day as approved totals. The ops-hub card shipped 8 September 2026; weights below are per-card. Changes are listed in the changelog.
Swipe sideways for weights and max contribution.
| ID | Criterion | Weight | Max contribution | What we measure |
|---|---|---|---|---|
| C1 | Time-to-competence | 15 | 75 | How long a non-developer operator needs to ship a reliable three-step workflow with basic error handling. |
| C2 | Reliability and operability | 20 | 100 | Execution consistency, retries, logs, and the operational burden of keeping the stack healthy. |
| C3 | Ecosystem | 15 | 75 | First-party integrations and templates that small teams actually use: email, sheets, Notion, chat, payments, accounting. |
| C4 | Canada fit | 15 | 75 | CAD pricing, tax invoices, data-residency options, and privacy posture as the vendor discloses it (PIPEDA / Quebec Law 25). |
| C5 | Pricing at small-team scale | 20 | 100 | Realistic monthly spend for a 2–5 person shop, in CAD where list prices exist, with GST/HST called out separately. |
| C6 | Portability and lock-in | 10 | 50 | Export paths, self-host options, and the real cost of leaving. |
| C7 | Support and documentation | 5 | 25 | Official docs, community, and paid support that is reachable at small-team spend. |
Swipe sideways for the ops-hub sheet.
| ID | Criterion | Weight | Max contribution | What we measure |
|---|---|---|---|---|
| C1 | Time-to-first useful ops system | 12 | 60 | How quickly a small team can stand up a working ops hub: docs, a source of truth, and one live process. |
| C2 | Docs + database UX clarity | 10 | 50 | Whether writing, tables, and linked records stay legible for non-specialists on a daily ops desk. |
| C3 | Native automation depth | 12 | 60 | First-party buttons, formulas, and automations without leaving the hub. |
| C4 | External connector / webhook reach | 10 | 50 | APIs, packs, and inbound/outbound webhooks that actually reach the rest of the stack. |
| C5 | Pricing predictability at small scale | 14 | 70 | Whether a 1–10 person shop can forecast the bill without surprise seat, AI, or automation add-ons. |
| C6 | Automation reliability & observability | 8 | 40 | Run history, failure visibility, and retries on native automations. |
| C7 | Extensibility (API / code / agents) | 10 | 50 | Public API, scripting, and agent hooks a small team can actually turn on. |
| C8 | Admin & collaboration | 8 | 40 | Permissions, sharing, and guest/commenter paths at small-team scale. |
| C9 | Compliance & data controls | 8 | 40 | Retention, export, SSO/SCIM, and what the vendor actually publishes on controls. |
| C10 | Lock-in & portability | 5 | 25 | Export paths and the real cost of leaving the hub. |
| C11 | Self-host / control option | 3 | 15 | Whether an operator can run the hub themselves. Absence is scored; we do not invent a self-host lane. |
How a criterion is filled
01 · Sources
Primary sources
Vendor pricing pages, docs, DPAs, status pages, and licence texts — dated, with the URL we used.
02 · Builds
Operator builds
When we publish a 1–5 cell, it includes at least one timed build of a three-step workflow with error handling.
03 · Canada
Canada notes as evidence
CAD price if listed; otherwise USD converted on the date of capture. GST/HST treated as a separate line. Privacy claims limited to what the vendor actually publishes.
04 · Review
How we publish a cell
The /500 totals on the first card were approved 2026-09-07. Criterion-level 1–5 cells stay off the page until they are issued. We do not back-solve missing cells from a total.
Independence
- Affiliate relationships never change a score, a weight, or whether a tool is included.
- We would rather lose a commission than launder a ranking.
- Make Partner is live (
pc=stackscorecard) and marked affiliate. Other programmes are not enrolled — no invented IDs. - Vendors do not get a preview of unpublished scores in exchange for a relationship.
Full text: Affiliate disclosure.
What we will not do
- Invent testimonials, user counts, or “most teams choose X.”
- Publish a “best [category] 2026” list without a scorecard behind it.
- Score “AI magic,” “innovation,” or other unfalsifiable adjectives.
- Treat self-host operational cost as free because the licence is cheap.
- Rank a card as #1 before every criterion is scored.
- Give legal, tax, or privacy advice. Canada notes are operator notes, not counsel.
Canada fit, specifically. This criterion is our E-E-A-T edge, not a maple-leaf badge. On the first scorecard we named what we could: Zapier CAD billing, n8n Cloud in EU Frankfurt, n8n self-host under the operator’s PIPEDA posture. We do not certify compliance.
Changelog
- 2026-09-08. Ops-hub card shipped (eleven criteria, weights sum 100). Approved totals: Notion 397 / Airtable 390 / Coda 348. Coda’s connector / webhook criterion scored 4 of 5. Hands-on timed build still pending (docs only). Not comparable to the Make / n8n / Zapier totals.
- 2026-09-07. v0.2. Scale is 1–5 × weight (max 500). Approved totals issued. Make’s pricing criterion scored 4 of 5 the same day (410/500, cloud #1). n8n Self-hosted is on the same sheet, not a fourth cloud product.
- 2026-09-06. v0.1 published. Seven criteria. No scores issued. (v0.1 used a 1–10 working note; v0.2 replaces that scale.)