Strategy & Planning

How to run a retrospective or post-mortem as CPG Founders

Strategy & PlanningFor CPG Founders3 apps11 steps~22 min to set up

After a co-packer run goes sideways, a retailer rejects a pallet, or a Q2 launch falls flat, you know you need to do a retrospective — but it never actually happens. You're already underwater on the next production cycle. What passes for a retro is a 20-minute Slack thread that nobody saves, decisions made verbally on a Zoom call with no notes, and action items that live in your head until the same mistake happens six months later. Your institutional memory is scattered across Google Docs nobody updates, a Notion page from 2023, and one person's inbox. For CPG founders running lean, the retro is the first thing dropped and the most expensive thing to skip.

Build this on Starch → See the apps used

Strategy & PlanningFor CPG Founders3 apps11 steps~22 min to set up

Outcome

What you'll set up

A structured retro and post-mortem workflow that captures decisions, root causes, and action items from every production run, retailer review, or launch — automatically transcribed and searchable, so the same co-packer mistake never costs you twice.

A living knowledge base that stores every retro output — lot issues, deduction disputes, forecasting misses, launch learnings — tagged by SKU, co-packer, and channel, so you can actually find the lesson when you need it.

A task tracker wired to every retro's action items, with owners and due dates assigned by prompt, so 'Jake needs to update the allergen SOP by Friday' doesn't get lost in the meeting recording.

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

Meeting Notes Knowledge Management Task Manager

Data sources & config

Meeting Notes transcribes and summarizes retro calls live; Starch connects directly to Google Calendar to pull scheduled retro meetings automatically. Knowledge Management stores retro outputs and links to Notion pages via Starch's integration catalog (agent queries Notion live when an app or search needs it). Task Manager tracks follow-up action items captured from retros with priority levels and due dates.

Prompts to copy

Transcribe today's post-mortem on the March oat bar co-packer run. Summarize what went wrong with the moisture spec, what decisions we made about the SOP change, and list every action item with the person responsible and a due date.

Save this retro to our knowledge base under 'Co-packer Issues > Lot 2024-031 > Root Cause' and tag it with SKU B-042, co-packer Acme Foods, and channel Whole Foods. Flag the allergen SOP doc as stale and link the updated version once Jake uploads it.

Create a task for me to review our FBA replenishment lead times before the next co-packer scheduling call — P2 priority, due next Thursday.

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

Walkthrough

Step-by-step

1 Install Meeting Notes and connect it to Google Calendar — Starch syncs your calendar data on a schedule, so it knows when a retro is booked and is ready to transcribe the moment the call starts.

2 Run your retro or post-mortem call — co-packer debrief, retailer rejection review, Q2 launch recap, whatever it is — and let Meeting Notes transcribe in real time. Don't assign anyone to take notes.

3 After the call, prompt Starch: 'Summarize today's post-mortem. Pull out the root cause, what we decided to change, and every action item with owner and due date.' Review the output and correct anything misattributed.

4 Prompt Starch to save the retro summary to Knowledge Management: 'File this under Co-packer Issues, tag it with the SKU, lot number, co-packer name, and channel. If there's an existing SOP doc this affects, flag it as stale.' Starch queries your Notion workspace live to find the linked doc.

5 For each action item surfaced in the retro, prompt Task Manager: 'Create a task for [person] to [action], priority P[1-4], due [date].' Starch creates the task, assigns it, and sets the deadline without you clicking through a form.

6 If the retro surfaced a deduction dispute or compliance gap — say, an undisclosed short-ship that triggered a Whole Foods compliance charge — create a dedicated Knowledge Management entry: 'Save a note under Deduction Disputes > Whole Foods > March 2026 with the lot number, deduction amount, and what we're filing as a response.'

7 Before the next co-packer scheduling call, prompt Knowledge Management: 'What issues have we had with Acme Foods in the last 12 months? Pull any retros, deduction notes, or SOP changes tagged to them.' Get the full history in seconds instead of searching six Slack threads.

8 At the start of each quarter, prompt Knowledge Management: 'List every retro from the last 90 days. Group them by root cause category: co-packer, forecast, retail, regulatory. Which category has the most recurring issues?' Use this to decide where to invest process time.

9 When onboarding a new ops hire or a co-packer account manager, prompt Knowledge Management: 'Build an onboarding path for someone new to our co-packer process. Pull relevant retros, SOPs, and decisions from the last 18 months.' Turn your institutional memory into a structured ramp instead of a founder Q&A session.

10 Set a recurring calendar event called 'Weekly Retro Action Item Review' — prompt Task Manager at the start of each week: 'Show me all overdue and P1 tasks from retro action items.' Treat these as non-negotiable before new work starts.

11 When the same mistake recurs — a second moisture spec failure, another short-ship deduction — prompt Knowledge Management: 'Have we seen this before? Search for moisture spec, co-packer, and lot rejection.' If a retro surfaces, link it in the new post-mortem so you're building on the pattern instead of rediscovering it.

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

March 2026 — Acme Foods Lot Rejection Post-Mortem

Sample numbers from a real run

Co-packer re-run cost (450 cases destroyed)	8,100
Whole Foods compliance charge (short-ship penalty)	2,200
Expedited freight to hit reset window	1,650
Lost margin on two-week FBA stockout (B-042 oat bar)	4,300

In March 2026, lot 2024-031 of your B-042 oat bar came back from Acme Foods with moisture content at 14.2% against a spec ceiling of 12.5%. Whole Foods rejected the pallet at dock. You had a retro with your ops lead and the Acme QA contact on March 14th. Meeting Notes transcribed the full 38-minute call and surfaced three action items: Jake to update the pre-production moisture check SOP by March 21st, you to add a mandatory hold-and-test step to the co-packer PO template before the next run, and your broker to formally dispute the $2,200 compliance charge. Knowledge Management filed the retro under 'Co-packer Issues > Lot 2024-031 > Root Cause,' tagged to SKU B-042 and Acme Foods, and flagged the 2022 allergen SOP as stale because it hadn't been updated to reflect the new moisture spec. Three months later, when you're scheduling the June run, you prompt Knowledge Management: 'Any prior issues with Acme Foods?' — and the March retro surfaces immediately, including the SOP change and the unresolved dispute. Total documented cost of the incident: $16,250. The retro took 38 minutes. The knowledge base entry took one prompt. The action items were all closed within the due dates because Task Manager surfaced them every Monday.

Measurement

How you'll know it's working

Retro action item close rate within the agreed due date (target: >80% closed on time)

Recurrence rate of same root-cause category across co-packer runs per quarter

Time from incident to filed retro summary (target: <48 hours after the call)

Number of deduction disputes documented with supporting retro context vs. disputed blind

Founder hours per week spent answering questions that exist in the knowledge base

Comparison

What this replaces

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

Notion + Otter.ai + Asana

Three separate tools with no shared context — your Otter transcript lives in Otter, your retro doc lives in Notion, and your action items live in Asana with no link between them; you're manually copying between all three after every call.

Google Docs + Slack

Zero structure: retro notes go into a doc nobody tags or archives, action items get buried in a Slack thread, and six months later the lesson is completely unretrievable without a founder archaeology session.

Confluence + Jira

Built for software engineering teams with a dedicated ops person to maintain taxonomy and workflows — serious setup overhead for a two-person CPG brand, and you'll pay for seats you don't use.

Loom + manual notes

Loom captures the video but search is limited to transcripts, action items aren't extracted, and nothing connects to a knowledge base or task system — it's a recording, not a workflow.

On Starch RECOMMENDED

One platform — meeting notes, knowledge management, task manager 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

Can Starch pull in the actual lot data or co-packer specs to attach to a retro automatically?

If you track lot data in a spreadsheet, Airtable, or a Google Sheet, you can connect those from Starch's integration catalog and the agent queries them live when you're building out a retro entry. If your co-packer portal is web-based and doesn't have an API, Starch can automate it through your browser — no API needed — to pull the relevant production records into the retro context.

Does Starch integrate with Notion if that's where we already keep our SOPs?

Yes. Starch connects directly to Notion — it syncs your pages and databases on a schedule, so Knowledge Management can surface and link your existing SOPs when you file a retro. You can tell Starch to flag a specific Notion doc as stale and link the updated version once it's uploaded.

What if our retro isn't a Zoom call — it's a walking conversation or a Slack thread?

Meeting Notes works best with recorded calls. For informal retros, you can paste a Slack thread or typed notes directly into Starch and prompt: 'Turn this into a structured retro with root cause, decisions, and action items.' The output gets filed into Knowledge Management the same way. Starch connects directly to Slack — it can pull channel history live when your app or workflow needs it.

Is Starch SOC 2 certified? We're going into a major grocery chain and they'll ask about data security.

Starch is not SOC 2 Type II certified yet. If your retail partner or co-packer requires SOC 2 as a vendor prerequisite for data-sharing tools, that's worth knowing upfront. For retro and knowledge management workflows, this matters less than it would for something touching PII or financial transactions directly.

Can I use this to run retros after a product launch, not just a co-packer issue?

Yes — the workflow is the same structure regardless of trigger. Tell Starch: 'Run a post-mortem on the Q1 Costco launch. Capture what the sell-through rate was versus forecast, why we think we over-produced by 30%, what we'd change about the merchandising brief, and who owns the updated demand model for Q3.' Meeting Notes captures the call; Knowledge Management files it under your channel and SKU taxonomy; Task Manager tracks the follow-ups.

We lose action items constantly after retros. How does Starch actually prevent that?

Meeting Notes extracts action items from the transcript automatically and assigns them to people by name. You then prompt Task Manager to create those as tracked tasks with priorities and due dates. At the start of each week, you can prompt Task Manager to surface all overdue retro action items. The loop closes because the task exists in a system with a due date — not in someone's memory or a meeting doc nobody re-reads.