# Divya Menon, Senior UX Researcher at Groww (350 employees) — read of ResearchScope, June 22 2026

> 8 years doing qual research, currently running 6-8 interviews a week across Hindi and English speaking users in Tier 2 Indian cities. My team has 3 researchers. Two of our Dovetail seats go unused every month.

## How I got here

Someone in the UX Research India Slack forwarded it with a one-liner: "finally something with Indic language support." I clicked expecting to be annoyed. The forwarding context matters here because I was already primed to look for something specific, not just browse.

## What I clicked first

"Record in Hindi, Tamil, Telugu, Kannada. Auto-transcription and English synthesis built in." That stopped me. That is the first time I have seen that written plainly on a research tool homepage. Every other tool treats Indic languages as an afterthought or a future roadmap item. I did not click anything, I just read that line twice.

The "$25 per interview" hit immediately after and I did the math fast. I do maybe 25 interviews a month. That is $625. We currently pay roughly $400/month in Dovetail seats and another $150 in Rev.com for transcription of Hindi sessions (which Rev gets wrong constantly). So the comparison is not crazy.

## Where I paused

The "Synthesis in Hours" section. "Extract themes, quotes, sentiment, patterns automatically. Build findings in hours, not weeks." I have seen this exact sentence, with almost these exact words, on Dovetail, Looppanel, Notably, and EnjoyHQ. It is not false but it is also not a differentiator anymore. Every single tool says this. What I wanted was: what does the synthesis actually look like? What is the output? Is it a structured affinity map? A tagged quote library? A summary paragraph? I clicked around and could not find an actual screenshot of the output in context.

## What I distrusted

Two things.

First: "Honest disclosure: we don't have live customers on this idea yet." I respect the transparency, genuinely. But it also means the "$25/interview" is a pricing hypothesis, not a validated number. The Indic transcription accuracy claim is also unvalidated. Indic ASR is notoriously bad, especially for code-switched speech (Hindi-English mixing, which is like 80% of how my participants speak). Nobody proves that claim with a demo or a WER metric.

Second: the Fermi math is visible on the page and it says "Year-1 take-home: -$22,850." That is the estimated revenue for the OPERATOR building this product, not for me as a user. But it is right there on the homepage next to the product features. It made me feel like I was looking at a startup pitch deck that forgot to hide the internal slide. The "Wishdeal Factory" scoring system is interesting but also very strange to put on a product page rather than a company blog.

## What would convince me

A 3-minute screen recording of a real Hindi interview going through the pipeline: raw audio, transcription output with the code-switching visible, and the English synthesis side by side. Not a polished demo. A rough one where the tool handles "toh basically kya hua tha" and I can see what it does with that.

Also: one researcher, named, saying "I switched from Dovetail and here is what changed." Not a stats claim. A name and a context I can verify on LinkedIn.

## What I'd ask in an email reply

1. What ASR engine are you using for Hindi and Tamil, and what is your word error rate on code-switched speech? Do you have a test set you can share or point me to?

2. When you say "synthesis in hours," what is the actual artifact? Can you show me an export from a real (or anonymized) session so I understand what I am paying $25 to receive?

3. The page mentions "Researcher-First Design" but the only integrations listed are Miro, Figma, Notion. No Dovetail, no Confluence, no Google Slides. Is that intentional or on the roadmap?

## Verdict: curious-enough-to-reply

The Indic language claim is the only thing on this page that I have not seen before, and it is exactly the problem I spend the most time working around. If the transcription actually works on real field interviews, this is worth a pilot even if everything else is rough.

---
*Memo by skeptic persona, generated 2026-06-22. Studio breaks own self-grading loop.*