Customer Support

How to score customer health as CPG Founders

Customer SupportFor CPG Founders2 apps12 steps~24 min to set up

You're running a CPG brand with a small team and your customer health visibility is a patchwork of Shopify order counts, Klaviyo open rates, and gut feeling. When a wholesale account goes quiet or an Amazon Subscribe & Save subscriber churns, you find out weeks later — usually when the reorder doesn't come. You don't have a CS ops team building health scores in Gainsight. You have a spreadsheet with last-order dates that nobody updates, a support inbox with no tagging system, and a mental model of 'active vs. at-risk' that lives entirely in your head. You need a system that watches your accounts and surfaces the ones slipping before they're gone.

Build this on Starch → See the apps used

Customer SupportFor CPG Founders2 apps12 steps~24 min to set up

Outcome

What you'll set up

A live customer health score per account — DTC, wholesale, and Amazon — built from order recency, order frequency, support ticket volume, and email engagement, so you always know who's healthy, who's drifting, and who needs a call this week

Automated at-risk alerts that fire when a wholesale buyer misses their usual reorder window or an Amazon subscriber cancels, giving you time to intervene before the relationship is truly lost

A single CRM view that ties health scores to account records — so when you look up a buyer at UNFI or a key DTC VIP, you see their health trend, last touchpoint, open support issues, and what to do next

The Starch recipe

Apps, data, and prompts

The combination of Starch apps, the data sources they pull from, and the prompts you use to drive them.

Apps used

CRM Customer Support Agent

Data sources & config

Starch connects to Shopify from its integration catalog — the agent queries order history live when your health scoring app runs. Starch syncs your Gmail data on a schedule so email thread history flows into CRM account records automatically. Connect Zendesk and Klaviyo from Starch's integration catalog; the agent queries ticket counts and campaign engagement live. Starch syncs your Slack data on a schedule to route at-risk alerts to your ops channel.

Prompts to copy

Build me a customer health scoring app for my CPG brand. Score each account on four signals: days since last order (from Shopify), order frequency trend over the last 90 days (from Shopify), number of open or unresolved support tickets in the last 30 days (from Zendesk), and email engagement rate over the last 4 campaigns (from Klaviyo). Roll these into a 0-100 health score. Flag anyone below 40 as at-risk and send me a Slack message with their name, score, and last order date every Monday morning.

Add a health score column to my CRM. For each company record, pull their current health score from the scoring app and show their 30-day trend — improving, stable, or declining. When a score drops below 50, automatically create a follow-up task assigned to me with the note 'health score dropped — check in this week.'

Run these in Starch → or paste them into your favorite agent

Walkthrough

Step-by-step

1 Connect Shopify from Starch's integration catalog. This gives your health scoring app access to order history per customer — date of last order, order count over 30/60/90 days, average order value, and subscription status for Subscribe & Save or ReCharge customers.

2 Connect Zendesk (or Help Scout, or whichever support tool you use) from Starch's integration catalog so the agent can count open, unresolved, and recently escalated tickets per account when it calculates health scores.

3 Connect Klaviyo from Starch's integration catalog so the app can pull email open and click rates per contact over your last four campaign sends — engagement decay is one of the earliest signals a DTC account is drifting.

4 Starch syncs your Gmail data on a schedule. Wire it to your CRM app so every email thread with a buyer or distributor rep is automatically logged against the right company record — no manual BCC-to-CRM required.

5 Open the CRM app from Starch's App Store and describe your account structure: 'I have three account types — DTC customers, wholesale buyers (UNFI, KeHE, regional distributors), and Amazon accounts. Add a health score field and a last-contacted field to each. I want to see them in a single pipeline view sorted by health score ascending so the most at-risk are always at the top.'

6 Build the health scoring app by typing your scoring logic in plain language. Define the four signals, their weights (e.g., order recency 40%, frequency trend 30%, support burden 20%, email engagement 10%), and the thresholds that define healthy, at-risk, and churned. Starch assembles the calculation and runs it on your connected data.

7 Set up the at-risk alert automation: 'Every Monday at 7 AM, run the health scoring app, find every account whose score dropped more than 10 points in the last 7 days or whose score is below 40, and send me a Slack message listing them with their score, their last order date, and whether they have any open support tickets.' Starch queries your live data and fires the message without you touching it.

8 Add a reorder-window alert specifically for wholesale accounts: 'Look at each wholesale buyer's historical reorder cadence. If a buyer who typically orders every 45 days hasn't placed an order in 60 days, create a task in my CRM to call them and flag them as at-risk.' This is especially important for seasonal SKUs where a missed reorder window can mean a whole quarter of lost distribution.

9 Wire the health score back into your CRM account records. When you pull up a wholesale buyer's record, you should see their current score, their 30-day trend line, their last three orders with dates and amounts, and any open support tickets — all in one place without switching tabs.

10 Customer Support Agent — coming soon — will handle tier-one inbound tickets automatically (shipping status, return policy, subscription changes) and route anything that requires human judgment to your inbox with full account context already attached, including the customer's health score so you can prioritize your responses.

11 Once the system has run for 60 days, use Starch to ask: 'Which customers went from healthy to churned in the last quarter? What did their health score trend look like in the 30 days before they stopped ordering? Were there common signals — support tickets, email drop-off, order frequency decline?' This gives you a CPG-specific churn leading indicator you can act on earlier next cycle.

12 Publish your health scoring app to a shared view your co-founder or ops hire can see. Everyone on the team should be working from the same at-risk list — not a different version of the spreadsheet.

See this running on Starch

Connect your tools, describe what you want, and the agent builds it. Closed beta is free.

Try it on Starch →

Worked example

Q1 2026 wholesale at-risk review — regional grocery chain account

Sample numbers from a real run

Days since last order (was 38-day cadence)	67
Order frequency trend (orders in last 90 days vs. prior 90)	-2
Open support tickets (deduction dispute, unresolved 19 days)	1
Klaviyo email engagement rate, last 4 sends	8
Calculated health score (out of 100)	31

Your health scoring app flags Sunridge Market — a 12-store regional chain that's been one of your top wholesale accounts since launch — as at-risk with a score of 31 on Monday morning. The breakdown: they haven't ordered in 67 days against their usual 38-day cadence, they've placed two fewer orders over the last 90 days compared to the prior period, there's an open deduction dispute in Zendesk that's been sitting unresolved for 19 days (your broker filed it but nobody followed up), and their email engagement has fallen to 8% open rate — down from 31% six months ago. Without this system, you'd have noticed the missing order eventually, but probably not the deduction dispute quietly poisoning the relationship. You call their category manager that Tuesday, resolve the $840 deduction dispute on the spot, and learn they've been frustrated by a shelf-placement issue your sales rep hadn't surfaced. You reorder that week. Health score climbs back to 74 by the end of the month.

Measurement

How you'll know it's working

Percentage of wholesale accounts with health score above 60 (distribution health rate)

Average days between at-risk flag and reorder or relationship close — measures how fast you intervene

Deduction dispute resolution time (days from filing to credit), since unresolved deductions are a major health drag on distributor accounts

DTC subscriber churn rate month-over-month, tracked at the cohort level by acquisition channel

Support ticket volume per 100 orders — a rising ratio signals product or fulfillment issues before they show up in churn

Comparison

What this replaces

The other ways teams handle this today, and how the Starch version compares.

Gainsight or Totango

Built for SaaS CS teams managing hundreds of software accounts; the data model, pricing, and implementation complexity assume a dedicated CS ops person and a six-figure budget — not a two-person CPG brand.

Klaviyo + Shopify reports (manual)

You already have the raw data here, but assembling a cross-channel health score by hand every week isn't sustainable once you're managing 30+ wholesale accounts alongside DTC.

HubSpot CRM with custom scoring

HubSpot can score contacts but the setup requires significant admin work, and it doesn't natively pull in support ticket data or order history without expensive add-ons and a developer.

Spreadsheet (Google Sheets + manual export)

Works until it doesn't — data is always stale, the person who built the formula leaves, and you can't send automated alerts from a spreadsheet without duct-taping Zapier to it.

On Starch RECOMMENDED

One platform — crm, customer support agent all running on connected data. Setup in plain English; numbers stay current via scheduled syncs and live agent queries.

Try it on Starch →

FAQ

Frequently asked questions

Does Starch work with Shopify? I run DTC through Shopify and wholesale through Faire and KeHE Connect.

Yes — connect Shopify from Starch's integration catalog and the agent queries your order data live when the health scoring app runs. Faire and KeHE Connect don't have formal API connectors, but Starch can automate them through your browser — no API needed. That means you can pull order history from any portal a human can log into.

My support tickets are split between email (Gmail) and Zendesk. Can Starch pull from both?

Yes. Starch syncs your Gmail data on a schedule, so inbound support emails are accessible directly. Connect Zendesk from Starch's integration catalog for structured ticket data — open/closed status, resolution time, ticket category. Your health score can weight both.

Can this handle Amazon accounts, not just DTC and wholesale?

Yes. The Amazon Seller Dashboard app connects to your Seller Central data, and your health scoring app can pull order volume and subscription status per ASIN or customer segment. Subscribe & Save cancellation rate is a useful health signal you can wire in.

Is Starch SOC 2 certified? I'm nervous about connecting my Shopify and Stripe data.

Starch is not currently SOC 2 Type II certified — that's worth knowing before connecting sensitive financial data. SOC 2 certification is on the roadmap. If that's a hard requirement for your business today, it's an honest reason to wait.

What about the Customer Support Agent you mentioned? When is that available?

Customer Support Agent is currently in development — you can request beta access to get notified when it launches. It's not available today. In the meantime, the CRM app handles inbound email thread logging and you can build a triage automation manually using Gmail, Zendesk, and Slack.

I don't have consistent historical data — we switched from one 3PL to another 18 months ago and our order history is fragmented. Will the health scores be garbage?

For accounts that existed before the data gap, the scores will be less reliable at first — Starch is working from whatever history you can connect. A practical approach: start scoring from today, let 60 days of clean data accumulate, and then the trend signals become meaningful. You can also describe your incomplete data to Starch and it'll tell you which signals are calculable vs. which ones need more history.