Strategy & Planning

How to synthesize customer research interviews as Fitness Studio Founders

Strategy & PlanningFor Fitness Studio Founders3 apps10 steps~20 min to set up

After every member survey, focus group, or informal chat with regulars, you have a folder of voice memos, scribbled notes, and half-transcribed Google Docs that never get turned into anything actionable. You know your 6am CrossFit crowd wants longer cool-downs and your Wednesday yoga members want earlier start times — but that insight is buried in a recording on your phone from three weeks ago. You don't have a UX researcher on staff. Synthesizing even five interviews takes half a Sunday, and by the time you find the pattern, you've already set next quarter's class schedule.

Build this on Starch → See the apps used

Strategy & PlanningFor Fitness Studio Founders3 apps10 steps~20 min to set up

Outcome

What you'll set up

A structured synthesis of every member or prospect interview, automatically organized by theme (class timing, instructor preference, pricing friction, churn reasons) so you can see patterns across ten conversations instead of reading each one in isolation

A living knowledge base that stores every round of customer research so next time you run a survey or member focus group, you can compare new feedback against what members said six months ago

A ready-to-present summary — slide deck or written brief — that turns raw interview notes into a clear recommendation you can bring to your instructors or an investor

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 Presentation Agent

Data sources & config

Meeting Notes connects to Google Calendar (Starch syncs your Google Calendar data on a schedule) to pull interview events and match transcripts to sessions. Interview notes, past survey exports, and synthesis documents are stored in Knowledge Management, which connects to Notion — Starch syncs your Notion data on a schedule — so your research library is always searchable. Mindbody and MarianaTek member data (attendance history, cancellation dates) is pulled through browser automation — no API needed — to give your research context like visit frequency and membership tenure when you're segmenting interview responses.

Prompts to copy

Transcribe this recording of my member exit interview. Pull out every complaint, every compliment, and any specific class or instructor they mentioned. Format it as: theme → quote → my takeaway.

I've uploaded notes from eight member interviews I ran this month. Find every theme that came up in three or more conversations, rank them by how often they appeared, and flag anything that contradicts what members told us in our January survey.

Build me a 6-slide summary of our Q2 member research: what members love, the top three friction points, what we're going to change as a result, and one slide showing the split between long-term members and people who cancelled in the last 90 days.

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

Walkthrough

Step-by-step

1 After each member interview or focus group, open Meeting Notes and paste the transcript or upload the audio file. Tell Starch: 'This is an exit interview with a member who cancelled after 14 months. Pull out every reason they gave, every class they mentioned, and any competitor they named.'

2 Starch transcribes and structures the output — themes, direct quotes, and your own follow-up questions — and saves it automatically to your Knowledge Management library under a folder you define, like 'Member Research / Q2 2026.'

3 After you've run five or more interviews, go to Knowledge Management and prompt: 'Read every interview in Member Research / Q2 2026. List every theme that came up more than twice, with the quotes that support each one. Note whether the theme came from active members, churned members, or both.'

4 Starch scans across all stored interviews and returns a ranked theme list. You'll see things like 'class timing' appearing in six of eight interviews, with the exact member language — 'the 6pm slot is always too crowded' — attached.

5 Pull in attendance and cancellation context: tell Starch to use browser automation to pull 90-day attendance data from your Mindbody or MarianaTek account and match it against the members you interviewed, so you know whether the people complaining about crowded evening classes are actually your highest-frequency visitors.

6 If you ran a previous round of member research, ask Knowledge Management: 'Compare this month's themes to what members said in our January survey. What's new? What's the same? What reversed?' Starch searches your entire research history and returns a structured comparison.

7 Write up your synthesis: tell Starch 'Summarize our Q2 research into a one-page brief. Lead with the single biggest finding, then list three things we're going to change and why, with the member quotes that drove each decision.'

8 Hand that brief to Presentation Agent: 'Turn this research summary into a 6-slide deck I can share with my instructors. Slide 1: what we heard. Slides 2-4: the three changes we're making. Slide 5: what we're not changing and why. Slide 6: what we're going to measure to know if it worked.' Note: Presentation Agent is currently in development — request beta access to get notified when it launches.

9 Store the final synthesis document back in Knowledge Management so it becomes the baseline for your next research round. Tag it with the quarter and the member segment you focused on.

10 Set a recurring reminder in Calendar Management to run a lightweight five-question member check-in every quarter, so your research library grows continuously instead of being a one-time project you forget about.

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

Willow & Steel Pilates — April 2026 Churn Research Sprint

Sample numbers from a real run

Interviews conducted	9
Churned members interviewed	6
Active members interviewed	3
Distinct themes surfaced by Starch	11
Themes appearing in 4+ interviews	4
Time from first upload to final brief	3

Mara runs a 12-instructor Pilates studio in Austin with 340 active members on monthly plans. She'd lost 28 members in Q1 and had a hunch it was pricing — her Reformer membership went from $189 to $219 in January. She ran nine exit and stay interviews over two weeks, uploading each audio file to Meeting Notes as she finished. Starch surfaced 11 distinct themes across the nine conversations. Pricing came up in four, but so did 'not enough weekend morning slots' — which appeared in five interviews, including three from her most tenured active members. When she asked Starch to cross-reference interview subjects with 90-day attendance pulled from MarianaTek through browser automation, she found that the four members who cited weekend access were all in the top 15% by visit frequency. The pricing narrative was partially right, but the constraint driving churn among her best members was a scheduling gap, not rate sensitivity. Her final six-slide brief — built in about 40 minutes once the synthesis was done — recommended adding two 8am Saturday Reformer sessions before any rollback on pricing. She presented it at her April instructor meeting with the member quotes on screen. First new Saturday session filled in six days.

Measurement

How you'll know it's working

Number of member interviews synthesized per quarter (target: at least 8, covering both churned and active members)

Themes-to-action rate: how many research themes resulted in an actual class schedule, pricing, or instructor change within 60 days

Research reuse rate: how often current synthesis references findings from a prior quarter rather than starting blind

Churn rate by member segment in the 90 days following a research-driven change

Instructor retention correlation: whether members who cite a specific instructor in interviews have materially different churn rates than the studio average

Comparison

What this replaces

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

Otter.ai + manual Google Docs synthesis

Otter transcribes accurately but does nothing with the content — you still spend two hours per interview round manually reading, tagging, and comparing, and last quarter's notes live in a Doc nobody opens.

Typeform or Google Forms for member surveys

Surveys give you quantitative signal but miss the nuance — you don't find out that 'scheduling' means 'specifically 8am Saturday' until someone tells you in a conversation, and survey tools don't synthesize open-text responses across rounds.

Notion as a research repository

Notion stores everything but searches poorly across documents and has no synthesis layer — you can find a specific interview if you know where it lives, but you can't ask 'what did churned members say about instructor quality across all of 2025.'

Hiring a part-time ops person to do research synthesis

Consistent but expensive — a 10-hour-per-month contractor at $40/hr is $4,800/year to do work that compounds in value only if the findings are stored and searchable, which they usually aren't.

On Starch RECOMMENDED

One platform — meeting notes, knowledge management, presentation 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

My interviews are just voice memos on my phone. Can Starch actually work with those?

Yes. Upload the audio file to Meeting Notes and tell Starch what the interview was about and who you were talking to — churned member, long-term regular, prospective member who didn't join. Starch transcribes and structures it. The more context you give upfront ('this person cancelled after 8 months, they used to come three times a week'), the more useful the theme extraction.

Does Starch connect to Mindbody or MarianaTek to pull member data automatically?

Mindbody and MarianaTek don't have open APIs for independent studios, so Starch automates them through your browser — no API needed. You'll need to connect your account once, and then Starch can pull attendance history, membership tenure, and cancellation dates on a schedule so you have that context when you're interpreting interview findings.

What if I want to compare this quarter's research to what members said six months ago?

That's exactly what Knowledge Management is built for. Every synthesis document and interview extract you store becomes part of a searchable library. Ask Starch: 'What did churned members say about class scheduling in Q4 2025 versus what they're saying now?' and it searches across everything you've stored and returns a structured comparison.

Is my member interview data secure? These are real conversations with real customers.

Honest answer: Starch is not SOC 2 Type II certified yet. If your studio operates under any specific data retention or privacy obligations — for example, if you're in the EU or collect health-adjacent information — review whether Starch's current compliance posture meets your requirements before storing sensitive member conversations. For most independent fitness studios in the US, this isn't a blocker, but it's worth knowing.

The Presentation Agent sounds useful. Is it available now?

Not yet — Presentation Agent is currently in development. You can request beta access to get notified when it launches. In the meantime, you can have Starch write the synthesis as a structured document in Knowledge Management, and export or paste it into Google Slides manually. The research and synthesis work — the part that takes the most time — is fully available today.

I only talk to five or six members a quarter. Is that enough for Starch to find patterns?

Five interviews is enough to start. Starch will surface what appeared in multiple conversations and be honest about what appeared only once. The more useful habit is consistency: five interviews every quarter, stored in the same library, means by the end of the year you have 20 conversations to search across — and themes that are real tend to keep showing up.