Customer Health Score (No CS Platform)

A working customer health score system for mid-market SaaS teams who don't have Gainsight, ChurnZero, or Vitally — and don't want to spend $50k/year to get one. Uses the tools you already have: Mixpanel, Zendesk, Stripe, Google Sheets, Slack.

---

What this replaces

The problem: You're a $5-30M ARR SaaS team. Your CSMs feel blind. You can't justify $50-100k/year for Gainsight or ChurnZero. So you have no health score. Or worse: you have a spreadsheet someone updates by hand once a month. This template: A daily-refreshing weighted health score per account, dropped into a Google Sheet you can chart, sorted by risk. Slack alerts when any account's score drops sharply. Built in n8n. Total cost: ~$0/month beyond the APIs you already pay for.

---

Prerequisites

• n8n instance (self-hosted or n8n Cloud)
• Mixpanel project with Service Account credentials (or Amplitude — see alternates)
• Zendesk account with API access
• Stripe account with read API key (for MRR data)
• Google Sheets — one spreadsheet, will be created/maintained automatically
• Slack workspace with bot OAuth token
• Optional: NPS tool API (Delighted, AskNicely, SurveyMonkey)

---

Setup Steps

1. Create the Google Sheet

Create a new Google Sheet called Customer Health Dashboard. Add these columns to Sheet1:

account_id | account_name | mrr | usage_score | support_score | nps_score | retention_score | health_score | health_level | last_updated | score_delta_7d

Copy the spreadsheet ID (from the URL) — you'll need it in step 5.

2. Import the workflow

Workflows → Import from File → workflow.json

3. Configure credentials

Service

Type

Notes

Mixpanel	HTTP Header Auth	Authorization: Basic
Zendesk	API token	Subdomain + email + token
Stripe	API key	Restricted key with customers:read and subscriptions:read
Google Sheets	OAuth2	Scope: spreadsheets
Slack	OAuth2 / Bot Token	chat:write + chat:write.public

4. Define your account universe

The workflow uses Stripe Customers as the source of truth for "who is a customer." If your data lives elsewhere:

• CRM-based: replace the "Fetch Customers" Stripe node with a Salesforce/HubSpot query
• Auth provider-based: replace with a query against your auth DB

5. Set the Google Sheets ID

• Open the "Append/Update Health Row" node
• Set documentId to your spreadsheet ID from step 1

6. Tune the scoring weights

• Open the "Calculate Health Score" Code node
• Default weights:

- Usage (30%): daily/weekly active days, key feature adoption

- Support (25%): ticket count + escalation rate (inverse)

- NPS (20%): latest score per account

- Retention/MRR (25%): tenure + MRR growth direction

Adjust the weights based on what predicts churn in your business. If you're early stage, weight usage higher. If you're enterprise-heavy, weight engagement and exec sponsor activity.

7. Set the Slack alert threshold

• Open the "Score Dropped >15 Points?" IF node
• Default: triggers if score_delta_7d <= -15
• Channel: edit the "Post Health Alert" node, replace #cs-alerts with your channel

8. Activate the workflow

• Toggle to Active
• Default schedule: daily at 5 AM
• First run may take 5-15 min for >500 accounts (Mixpanel JQL is slow at scale)

---

Scoring methodology

Each component is normalized to 0-100, then weighted:

Usage score (30%)

• 80-100: Active daily, multiple users, multiple features
• 60-79: Active 2-3 days/week, core features
• 40-59: Active weekly, single feature
• 20-39: Active monthly
• 0-19: No activity in last 30 days

Support score (25%)

• 80-100: 0-1 tickets last 30 days, no escalations
• 60-79: 2-3 tickets, no escalations
• 40-59: 4+ tickets OR 1 escalation
• 20-39: Multiple escalations
• 0-19: Active escalation in last 7 days

NPS score (20%)

• 80-100: NPS 9-10 (Promoter)
• 60-79: NPS 7-8 (Passive, recent)
• 40-59: No recent NPS response
• 20-39: NPS 5-6 (mild detractor)
• 0-19: NPS 0-4 (Detractor)

Retention/MRR score (25%)

• 80-100: MRR growing, tenure >1yr
• 60-79: MRR flat, tenure >6mo
• 40-59: MRR flat, new customer
• 20-39: MRR shrinking
• 0-19: MRR cut by half OR contract paused

Health levels

• 80-100: 🟢 Healthy — promoter/expansion candidate
• 60-79: 🟡 Watch — monitor, possible play
• 40-59: 🟠 At Risk — proactive intervention
• 0-39: 🔴 Critical — escalate to Head of CS

---

What the Slack alert looks like

🚨 Health Score Drop: Acme Corp
Previous: 78 (Watch)  →  Current: 51 (At Risk)
Delta: -27 points in 7 days

Top drivers:
• Usage score: 70 → 35 (key feature adoption dropped)
• Support score: 85 → 60 (3 new tickets, 1 escalation)
• NPS score: unchanged
• MRR/retention: unchanged

CSM: jane@yourcompany.com
View account: [Salesforce link]
View health trend: [Google Sheets link]

---

Alternate analytics tools

Amplitude instead of Mixpanel:

• Replace the Mixpanel JQL node with an Amplitude Dashboard REST API call
• POST https://amplitude.com/api/3/segmentation with the Charts API
• Same account_id event property required

PostHog instead of Mixpanel:

• Use the PostHog Query API
• Same event property model

Heap instead of Mixpanel:

• Heap doesn't expose JQL-equivalent — you'll need to dump events to your warehouse first

---

Cost estimate

Per day (full account refresh):

• Mixpanel JQL: free under standard plan (1 query/sec rate limit)
• Zendesk API: free under standard plan
• Stripe API: free
• Google Sheets API: free (60 writes/min — large customer bases may need to batch)
• Slack: free

Total cost: $0/month beyond the APIs you already pay for.

Compare to:

• Gainsight: $50,000-$200,000/year + implementation
• ChurnZero: $35,000-$120,000/year + implementation
• Vitally: $25,000-$80,000/year

---

Troubleshooting

"Some accounts return blank usage score" — make sure every event in Mixpanel includes an account_id property. If you're tracking by distinct_id (user ID), you'll need to add account context to your tracking calls first. "Google Sheets is slow / hitting rate limits" — switch the "Append/Update Health Row" node to batch mode (one Sheets API call per 100 accounts). "Health scores all cluster around 50" — your weights aren't differentiated enough. Try widening the bands: instead of 5 buckets per component, try 3 (good/ok/bad) with bigger point gaps. "NPS data is sparse" — if most customers don't respond to NPS, set the default to a neutral 60 (not 0). Otherwise you'll punish accounts for not responding.

---

Roadmap

• v1.1 — Add CSM assignment column, alert routes to the assigned CSM directly via Slack DM
• v1.2 — Save daily snapshots to a "history" tab so you can chart trends over time
• v2.0 — Trigger a Slack workflow when score drops to "Critical" for >7 days

---

Built by OnboardSuccess — AI for Customer Success, without the vendor spin.