Customer Support

How to build a customer knowledge base as DTC Brand Founders

Customer SupportFor DTC Brand Founders2 apps12 steps~24 min to set up

Your return policy lives in a Google Doc nobody can find. Your shipping FAQs are in a Notion page that's six months out of date. Your customer support runs through a shared Gmail inbox where three people are stepping on each other's replies. When a customer emails asking why their order hasn't arrived, someone on your team manually pulls up Shopify, checks the tracking number, copies the carrier link, and pastes it into a reply — twelve times a day. When you run a new bundle or change your exchange policy, that update doesn't make it into the doc, and your next hire answers every question from memory. You're a two- or three-person team handling support volume that a ten-person team would struggle with, and every hour you spend answering 'where is my order' is an hour you're not spending on the growth work that actually moves CAC.

Build this on Starch → See the apps used

Customer SupportFor DTC Brand Founders2 apps12 steps~24 min to set up

Outcome

What you'll set up

A single, searchable knowledge base that holds your shipping policy, return and exchange rules, product FAQs, and brand guidelines — all in one place, always current, accessible to every person on your team

An AI-powered Email Triage setup that routes incoming customer questions to the right answer or the right person, so your shared inbox stops being a game of hot potato

A foundation for the Customer Support Agent (coming soon) — your knowledge base becomes the source of truth the AI will use to answer tickets 24/7 without a dedicated support hire

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

Knowledge Management

Data sources & config

Starch connects directly to Notion (scheduled sync) to pull your existing docs and detect when pages go stale. Gmail is connected via scheduled sync so the Email Triage app can read incoming messages, categorize by topic, and draft replies. Shopify is connected from Starch's integration catalog — the agent queries it live when an automation needs order or tracking data to fill out a customer reply. Any carrier tracking pages or return portal sites Starch can't reach by API are automated through your browser — no API needed.

Prompts to copy

Build me a customer knowledge base with sections for shipping timelines, return and exchange policy, product care instructions, bundle FAQs, and wholesale inquiry responses. Pull in our Notion pages as a starting point and flag any docs that haven't been updated in 60 days.

Set up email triage for my Gmail inbox. Prioritize messages from customers with open Shopify orders. Auto-draft replies to common questions about shipping status and return requests using the knowledge base as the source. Flag anything that looks like a chargeback risk or a frustrated repeat customer for me to handle personally.

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

Walkthrough

Step-by-step

1 Connect Notion to Starch via scheduled sync. Starch ingests your existing SOPs, return policies, and shipping FAQs — whatever you already have, even if it's messy.

2 Tell Starch to organize your content into a structured knowledge base with sections that match how your customers actually ask questions: shipping, returns, product info, account and order issues.

3 Starch flags every doc that hasn't been edited in 60+ days and surfaces them in a 'needs review' queue so you can update your holiday shipping cutoffs, new bundle details, or policy changes without hunting for the original file.

4 Connect Gmail to Starch via scheduled sync. Point the Email Triage app at your support inbox.

5 Tell Starch to categorize inbound customer emails by topic — shipping status, refund requests, product questions, press and wholesale — and draft a reply for each common type using your knowledge base as the source.

6 For shipping status questions, Starch queries Shopify live from its integration catalog to pull the order and tracking number, then populates the reply draft automatically. Your team reviews and sends in one click.

7 Set a rule: any email flagged as a potential chargeback, a customer who has ordered more than three times, or a tone that reads as escalating frustration gets routed to you directly with full thread context.

8 Publish the knowledge base internally so every teammate — including seasonal hires — can search for answers without asking you. Starch's AI search finds the right doc even if they don't use the exact words in the title.

9 Set up a weekly automation: every Monday, Starch scans last week's support email threads, identifies any question that came up more than twice without a knowledge base match, and adds a draft article to your 'content gaps' queue for you to fill in.

10 As you add new products, run promotions, or update policies, paste the new info into Starch and it auto-categorizes and routes it into the right knowledge base section — no manual filing.

11 When the Customer Support Agent launches (currently in development — request beta access), your knowledge base becomes the source of truth it uses to answer tickets automatically. The setup you do today is the foundation for 24/7 AI support later.

12 Run a monthly check: Starch shows you which knowledge base articles got referenced most in email replies, so you know what's pulling weight and what's missing — your documentation roadmap, built from actual support data.

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

Post-BFCM Support Surge, December 2025

Sample numbers from a real run

Inbound support emails (first 72 hrs after BFCM)	340
Emails auto-drafted by Email Triage using knowledge base	218
Emails that required manual reply (escalations, chargebacks)	52
Average founder time per support email before Starch (minutes)	6
Average founder time per support email after Starch (minutes)	1.5
Hours saved in first 72 hours post-BFCM	19

Your brand ran a 30% sitewide sale on Black Friday. By Sunday morning you had 340 support emails, mostly variations of 'where is my order,' 'can I change my address,' and 'your site said free shipping but I got charged.' Before Starch, you and one other person would have spent the better part of Monday manually checking Shopify, pulling tracking numbers, and copy-pasting your return policy URL over and over. With the knowledge base and Email Triage set up, Starch drafted replies to 218 of those 340 emails automatically — pulling live order data from Shopify for tracking questions, referencing the updated free shipping threshold from the knowledge base for billing questions, and using your return policy doc for exchange requests. You reviewed and sent each draft in under two minutes. The remaining 52 emails — the ones with address correction requests that needed Shopify edits, the customer who got a damaged item and was already threatening a chargeback, and a wholesale inquiry from a boutique — were flagged for you with full context. You cleared the entire backlog by Tuesday afternoon instead of Friday. The knowledge base also caught that three customers asked about a bundle configuration your site page didn't clearly explain; Starch added it to the content gaps queue, you wrote the FAQ in ten minutes, and it's already live for next time.

Measurement

How you'll know it's working

First reply time to customer support emails (hours)

Percentage of support emails resolved without founder involvement

Number of repeat questions that have no knowledge base article yet (content gap count)

Refund and chargeback rate as a percentage of orders (tracked weekly)

Team onboarding time to handle support independently (days, not weeks)

Comparison

What this replaces

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

Gorgias + Notion (separate tools)

Gorgias handles ticket routing well, but your Notion knowledge base and your support inbox stay disconnected — your team still has to open two tabs and cross-reference manually, and updates to policy docs don't automatically improve reply quality.

Zendesk

Zendesk is built for larger support teams with dedicated agents and SLA workflows — the setup overhead and monthly cost don't make sense for a two-person DTC team handling a few hundred tickets a month.

A shared Gmail inbox with labels

Free and fast to start, but it doesn't draft replies, doesn't search your docs, and scales directly with your headcount — every new ticket still requires a human to read, look something up, and type.

ChatGPT for drafting replies

Works for one-off drafts but has no access to your actual Shopify order data, your current return policy, or your brand voice guidelines — you're still doing the lookup work yourself and pasting context in every time.

On Starch RECOMMENDED

One platform — knowledge management, founder inbox 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 have a built-in customer support ticketing system right now?

Not yet as a dedicated tool — the Customer Support Agent is currently in development and you can request beta access to be notified when it launches. What you can build today is the knowledge base and email triage layer that the future support agent will use as its source of truth. Getting that set up now means you're ready to flip the switch when AI-automated ticket resolution goes live.

We use Shopify. Can Starch actually pull order and tracking data to fill in customer reply drafts?

Yes. Shopify is available through Starch's integration catalog — the agent queries it live when an automation or email draft needs order status, tracking numbers, or customer purchase history. You connect it once from the integration catalog and the Email Triage app can reference it every time a 'where is my order' message comes in.

We already have some SOPs and FAQs in Notion. Do I have to rebuild everything from scratch?

No. Starch syncs your Notion pages on a schedule and imports them into the knowledge base as a starting point. It organizes, flags stale content, and builds search on top of what you already have. You're not starting from zero — you're cleaning up and centralizing what already exists.

What happens if we change our return policy or add a new product? Does the knowledge base update automatically?

If the update lives in Notion, Starch picks it up on the next scheduled sync. If you're adding something new — a new bundle, a holiday cutoff date — you paste it into Starch and tell it where to file it. The weekly content-gap automation also surfaces questions your team is getting that don't have an article yet, so your documentation roadmap stays tied to what customers are actually asking.

Is Starch SOC 2 certified? We're cautious about connecting customer data.

Starch is not SOC 2 Type II certified today — that's worth knowing upfront. If your compliance requirements make SOC 2 a hard requirement right now, that's a real constraint. For most early DTC brands handling customer email and order data, the tradeoff is worth evaluating against the operational cost of your current manual process.

Can Starch automate replies and actually send them, or does a human still have to approve?

Both modes are available depending on how you set it up. You can run it in draft mode — Starch writes the reply and a human reviews and sends — which is the right starting point while you're calibrating the knowledge base. Once you trust the drafts on common question types, you can set specific categories (like order tracking confirmations) to send automatically and keep escalations in the human review queue.