Strategy & Planning

How to synthesize customer research interviews as Small Customer Success Teams

Strategy & PlanningFor Small Customer Success Teams3 apps10 steps~20 min to set up

After every customer call, someone on your three-person team is supposed to synthesize what was said — common complaints, feature requests, churn signals, expansion hints — and turn it into something actionable. In practice, the notes live in a Notion doc nobody updates, the Loom recording sits unwatched, and the insight that Customer X mentioned three competitors in one call never makes it to the product team. You're running 30+ kickoff, QBR, and check-in calls per quarter across 250 accounts. There's no research ops function. No insights repo. The synthesis is you, on a Friday afternoon, trying to remember what six different customers said about the same onboarding friction.

Build this on Starch → See the apps used

Strategy & PlanningFor Small Customer Success Teams3 apps10 steps~20 min to set up

Outcome

What you'll set up

A structured interview synthesis app that pulls transcript data, tags themes by account health and segment, and surfaces recurring patterns across every call your team runs

An auto-generated insights digest that connects what customers are saying to what HubSpot shows about their deal stage, renewal date, and support volume — so you stop triangulating across three tabs

A searchable archive of every customer conversation your team has ever had, organized by theme, account, and date — so when a prospect asks 'do your customers struggle with X?', you have a real answer in 30 seconds

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 Growth Analyst Knowledge Management

Data sources & config

Meeting Notes captures and transcribes calls. Starch syncs your HubSpot data on a schedule — contacts, companies, deals, and owner assignments — so every synthesized insight can be linked back to account health and renewal stage. Intercom is connected from Starch's integration catalog and queried live when your synthesis app needs to cross-reference open tickets or conversation history. Notion is synced on a schedule so your existing docs and templates stay part of the knowledge base. Gmail is synced on a schedule and surfaces relevant email threads alongside call data when you're building the full picture for an account.

Prompts to copy

Transcribe this customer call, extract the top 3 pain points mentioned, any competitor names that came up, any expansion signals (asking about additional seats, new features, or other teams using the product), and flag anything that sounds like churn risk. Tag by account name and save to my CS research archive.

Across all customer calls from the last 90 days in my research archive, identify the top 5 recurring themes, which account segments they're most common in, and whether any theme correlates with accounts that have low HubSpot engagement scores or open Intercom tickets.

Build me a customer insights wiki that organizes call summaries by theme (onboarding friction, feature gaps, expansion signals, competitive mentions) and lets me search across all of it. Pull in the meeting notes from my CS archive and sync with HubSpot contact and deal data so each entry is linked to the account record.

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

Walkthrough

Step-by-step

1 Connect HubSpot — Starch syncs your contacts, companies, deals, and deal stages on a schedule so every customer insight is automatically tied to the right account record and renewal date.

2 Connect Intercom from Starch's integration catalog so the synthesis app can pull open ticket volume and recent conversation history for each account when you're assessing whether a theme represents a real churn signal or an isolated complaint.

3 Set up the Meeting Notes app and point it at your CS call recordings. Tell Starch: 'Transcribe each call, extract pain points, competitor mentions, expansion signals, and churn risk flags, then tag by account name and save to my customer research archive.'

4 Build a customer insights knowledge base using the Knowledge Management app. Prompt Starch: 'Organize my call summaries into a searchable wiki by theme — onboarding friction, feature requests, competitive mentions, expansion signals. Link each entry to the HubSpot account record.'

5 Run your first cross-call synthesis by telling Starch: 'Across all calls from the last 60 days, what are the three most common onboarding complaints, which customer segments raise them most, and how many of those accounts have a renewal in the next 90 days?' Starch pulls from your archive and cross-references HubSpot renewal data.

6 Set up a weekly digest automation — prompt: 'Every Monday, scan new call notes from the past 7 days, surface any new themes or accounts that mentioned a competitor or asked about a feature we don't have, and Slack me a summary.' This runs automatically so nothing falls through on busy weeks.

7 Wire the expansion signal tracker — tell Starch: 'Flag any call where a customer asked about adding seats, onboarding another team, or integrating with a new tool. Show me those accounts alongside their HubSpot deal stage and current MRR.' This becomes your weekly expansion pipeline review.

8 Build the QBR prep surface — prompt: 'For Account X's upcoming QBR, pull every call summary from the last 12 months, open Intercom tickets, HubSpot activity, and any expansion signals flagged, and draft a one-page account narrative I can use to prep.' Run this for each account in your QBR queue.

9 Share the insights wiki with your product team — because the knowledge base is searchable and tagged by theme, they can filter for 'feature gap: reporting' across 50 accounts without asking you to reparse notes.

10 Review the archive quarterly and tell Starch: 'Which themes from Q1 are still appearing in Q2 calls? Which ones have disappeared — and which accounts that raised them have since churned or expanded?' This becomes your CS retrospective.

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 Customer Research Synthesis — 47 Calls, 3-Person Team

Sample numbers from a real run

Calls synthesized	47
Distinct themes identified	11
Expansion signals surfaced	8
Accounts flagged as churn risk	5
Hours saved vs. manual synthesis	14

In Q1, your team ran 47 calls — kickoffs, 30-day check-ins, and QBRs — across a mix of mid-market and SMB accounts. Before Starch, synthesizing those into anything usable meant one team member spending a half-day each month pulling Notion notes and Loom links together into a Slack message that three people read. With Meeting Notes running, every call was transcribed and tagged automatically. Starch surfaced a pattern none of you had explicitly noticed: 9 accounts in the 51-200 employee segment all mentioned the same onboarding step — the first data integration setup — as where they lost momentum. Five of those 9 had Intercom tickets from week 2 on the same issue. When you cross-referenced with HubSpot, 3 of them had renewal dates inside 60 days. That's a save. The expansion signal tracker flagged 8 accounts that asked about adding a second team — your AE didn't know about 6 of them. The knowledge base now has 47 call summaries organized by theme, searchable in seconds, and linked to live HubSpot records. When your product team asked 'do customers actually care about bulk export?' you searched 'export' across the archive and had 12 relevant quotes and account names in under a minute.

Measurement

How you'll know it's working

Time from call completion to synthesized insight in the shared archive (target: same day, not same week)

Expansion signals identified per quarter that weren't in the CRM before synthesis

Accounts flagged as churn risk with documented supporting call evidence (not gut feel)

Recurring themes surfaced across more than 10% of accounts — tracked quarter over quarter to measure whether product changes actually resolve them

QBR prep time per account (target: under 30 minutes with the archive built out)

Comparison

What this replaces

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

Gong or Chorus

Strong call intelligence for sales teams, but built around rep coaching and pipeline forecasting — not CS research synthesis or cross-call theme analysis across a 250-account base; also priced per seat at a scale that doesn't fit a 3-person team.

Notion + manual tagging

Free and flexible, but the tagging only happens if someone does it — and on a 3-person team, 'someone' usually means nobody. No cross-call synthesis, no connection to HubSpot or Intercom data.

Gainsight / ChurnZero / Catalyst

Purpose-built for CS health scoring and playbooks, but six-figure contracts, a CS-ops person to configure, and months to implement — none of which a 3-person team has.

ChatGPT with copy-pasted transcripts

Works for a single call but doesn't persist across calls, doesn't connect to your HubSpot or Intercom data, and creates no searchable archive — so every synthesis starts from scratch.

On Starch RECOMMENDED

One platform — meeting notes, growth analyst, knowledge management 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

We don't have a dedicated note-taker or research ops function. How does this actually work for a team of three?

That's exactly who this is built for. Meeting Notes handles the transcription and initial extraction automatically — you describe the tagging logic once ('flag churn signals, expansion signals, competitor mentions') and it applies that to every call. The synthesis app does the cross-call pattern matching. Your team's job is to review the output and act on it, not to produce it.

Our calls happen in Zoom and we already use Notion for notes. Does Starch work with what we have?

Yes. Starch syncs your Notion data on a schedule, so existing docs and templates are part of the knowledge base from day one. For Zoom call recordings, the Meeting Notes app processes the transcript — you can point it at the recording. Zoom is reachable from Starch's integration catalog.

Can Starch pull in Intercom ticket data alongside call notes to give a fuller picture of each account?

Yes. Connect Intercom from Starch's integration catalog and the agent queries it live when your synthesis app needs ticket history for a given account. So when a call note says 'they're frustrated with onboarding,' you can see in the same surface whether there's an open Intercom thread on the same issue.

Is the customer data stored securely? We're sharing call recordings that contain customer conversations.

Honest answer: Starch is not SOC 2 Type II certified today. If your accounts have contractual requirements around call recording storage or data residency, check with your legal team before routing transcripts through any third-party tool, including this one. That's a real limit worth naming.

What if our HubSpot deal data is inconsistent — some accounts don't have renewal dates filled in, some have wrong CSM assignments?

Starch syncs what's in HubSpot. If the underlying data is patchy, the cross-references will be too. The synthesis app can still surface call-level patterns independent of HubSpot, but the account-level enrichment (renewal date, deal stage, MRR) is only as good as what's in your CRM. Fixing the HubSpot hygiene is worth doing regardless — Starch just makes the gaps more visible faster.

Can I share the insights archive with our product team without giving them access to everything?

You can share specific views or exports from the knowledge base. The full Starch workspace sharing model is worth confirming with the Starch team for your specific access-control needs — but the common pattern is building a shared read-only view of the themes wiki that non-CS users can search without seeing raw call notes.