Google Sheets + Koji: Run AI Interviews from a Spreadsheet and Auto-Log Every Insight Back

Answer first: For a huge number of teams, the real CRM is a Google Sheet. Koji meets you there. The integration runs both ways: (1) a row becomes an interview — add a contact (or paste a list) to a Sheet and Koji sends each person a personalized AI-moderated interview link, and (2) every finished interview becomes a row — Koji appends the respondent, top themes, sentiment, quality score, key structured answers, and a transcript URL to a live results tab. Build it with zero code using the Google Sheets modules in Zapier or Make, or with a short Apps Script that calls Koji's headless API and receives its webhooks. If you just want a one-shot batch, Koji's built-in CSV import does it in five minutes. This is the fastest way for a founder or ops lead with no data warehouse to run continuous research.

Why Google Sheets is the right home for lightweight research ops

Not every team has Segment, a warehouse, or a RevOps engineer. What they do have is a spreadsheet: the churned-this-week list, the beta signup tab, the conference-badge scan export, the trial accounts a founder is watching. A Sheet is the universal, zero-onboarding database — and it is exactly where a scrappy research motion should live.

The problem is that spreadsheets are inert. They hold contacts but cannot talk to them, and they hold no memory of why anyone did anything. Koji turns the Sheet into a two-way research surface: the left columns are who to talk to, and the right columns fill themselves in with what those people said. Because Koji's AI moderator runs the interviews (voice or text, adaptive follow-ups, automatic analysis), you get interview-grade depth without booking a single call — the spreadsheet equivalent of hiring a research team that never sleeps.

What flows in each direction

Sheet → Koji (a row triggers an interview)

• You add or paste a row with at least an email (and any personalization fields — name, plan, last action).
• An automation watches the sheet and calls Koji's Start Interview endpoint with respondent_email and optional custom fields.
• Koji sends each person a personalized link and moderates the interview end to end.

Koji → Sheet (an interview appends a results row)

• On analysis_ready, Koji's webhook fires.
• Your automation appends a row to a Results tab: respondent, date, sentiment, top themes, each structured answer, quality score, and a transcript URL.
• The tab becomes a living dataset you can pivot, filter, and chart with plain spreadsheet formulas.

Three ways to build it

1. No-code with Zapier or Make (about 20 minutes). Use the Google Sheets "New Spreadsheet Row" trigger to fire Koji's "Start Interview" action, and a Koji-webhook trigger to run Google Sheets' "Create Row" action into your Results tab. See Connect Koji to Zapier or Connect Koji to Make. Best for most teams.

2. Apps Script + headless API (about 45 minutes). A short Google Apps Script reads new rows and calls Koji's headless API; a companion doPost(e) web app receives Koji webhooks and appends results. Fully self-contained inside your Google account, no third-party automation tool.

3. CSV import for a one-shot batch (5 minutes). Download the Sheet as CSV and use Koji's built-in participant CSV import to launch interviews immediately, then export results back to a Sheet when they complete. Perfect for a first test before you automate.

Turn structured answers into clean columns

The magic that makes a Koji Results tab genuinely analyzable is structured questions. Each Koji study can mix six question types — open_ended, scale, single_choice, multiple_choice, ranking, and yes_no — and each returns a typed value. That means a scale question becomes a numeric column you can average, a single_choice becomes a categorical column you can pivot, and the open_ended themes come back as text you can filter. Design them with the structured questions guide and your spreadsheet does the aggregation for free.

Example: a weekly churn-interview spreadsheet

1. Every Monday, export "cancelled last week" from billing into a Contacts tab.
2. A Zapier zap sends each a Koji interview: one single_choice cancellation reason, two open_ended probes, one scale "would you return if we fixed this?"
3. As interviews finish, rows append to the Results tab with reason, sentiment, and quotes.
4. A pivot table ranks cancellation reasons by week — a live churn dashboard built entirely in Sheets. Pair it with churn interview questions for the script.

Identity and privacy

Match on email (respondent_email) so results land next to the right contact, or pass a row ID as respondent_id. For sensitive studies, run in anonymous mode: Koji collects no email, and you log only the aggregate insight columns — no personal identifiers in the sheet.

Plans and limits

The headless API and webhooks (needed for the fully automated, two-way setup) are on the Interviews plan (EUR79/month, 79 credits) and Enterprise. On the Insights plan (EUR29/month) you can still use CSV import to launch from a Sheet and export results back manually. Each completed interview consumes one credit.

What a Results row looks like

A well-designed Results tab is self-documenting. For each completed interview Koji can append: respondent (or an anonymous ID), completion date, overall sentiment, the top themes as text, one column per structured question (a number for scale, a category for single_choice, a list for multiple_choice or ranking), the interview quality score, and a transcript URL. From there, a pivot table or a QUERY() formula turns the tab into a dashboard — average NPS by segment, a count of each cancellation reason, or a filtered list of every frustrated respondent worth a personal follow-up.

Common pitfalls (and how to avoid them)

• One giant tab for everything. Keep a Contacts tab (inputs) separate from a Results tab (outputs) so triggers and appends never collide.
• No dedup key. Add an email or row-ID column and match on it so re-runs update the right person instead of creating duplicates.
• Forgetting rate limits. If you paste thousands of rows at once, batch them; do not fire thousands of interview starts in a single second.
• Storing sensitive answers in a shared sheet. For sensitive topics, run Koji in anonymous mode, log only aggregate columns, and restrict sharing on the file.

Beyond one sheet: when to graduate

A spreadsheet is the perfect place to start a research motion and often good enough to keep it. But when your Results tab crosses a few thousand rows, when several teammates need role-based access, or when you want insights joined to product behavior, it is time to graduate the same automation to a CRM or CDP. The good news: the Koji side does not change. The identical Start Interview call and analysis_ready webhook that powered your sheet also power a Segment or HubSpot pipeline — you just point the output at a new destination. Start in Sheets, scale when the spreadsheet starts to strain, and never rebuild your research logic from scratch.

A five-minute first test before you automate

You do not need any automation tool to prove the value. Put a list of emails in a single column, export the tab as a CSV, and use Koji's participant CSV import to launch interviews to everyone at once. Watch the real-time report fill in as responses arrive, then export the results back into a fresh Sheet tab when the study wraps. If the transcripts and themes are useful — and they almost always are — you then invest the 20 minutes to make it continuous with Zapier or Make. Test manually first, automate second.

Formulas that turn results into a dashboard

Once results land in a tab, plain spreadsheet formulas do the analysis no separate BI tool required. A COUNTIF over the single_choice cancellation-reason column gives you a ranked reason breakdown. An AVERAGEIF over a scale column segmented by plan tells you which customers are least satisfied. A QUERY() statement filtering sentiment = "frustrated" produces a live call list of at-risk customers for the CS team. Because Koji's structured questions return clean typed values rather than free text blobs, these formulas just work, and the dashboard updates itself every time a new interview appends a row.

Related Resources

• Importing Participants via CSV
• Exporting Research Data
• Connect Koji to Zapier
• Structured Questions in AI Interviews
• Churn Interview Questions
• Real-Time Research Insights