What we learned from our first 180 conversations with Indian elders

Most of what gets said about elderly parents is guesswork. For once, we could replace some of it with data.

Over about three months, Alfaaz recorded 180 conversations with 20 Indian elders across 16 families, all of them answering gentle questions over WhatsApp, in their own language and at their own pace. This is what that early data shows: how elders answer, what they actually talk about, and which questions get them to open up. The sample is small and these are first months, so treat them as honest observations rather than laws. But they confirm something most families already sense, and a few of them genuinely surprised us.

Elders answer by voice, never by typing

Of the 751 elder replies in our data, every one came as a voice note. Not a single elder has typed their answer. Together those recordings add up to about 3.2 hours of voice.

That number says something simple and important: for this generation, talking is effortless and typing is friction. A voice note asking about a first job is not a new task; it is the thing they already do every day, pointed at something worth keeping. It also means we capture what a transcript never could: the pause before a hard memory, the laugh in the middle of a happy one, the exact word a feeling lives inside.

Across 751 replies, not one elder has typed their answer; every memory has arrived as a voice.

Ask about career, and they go quiet. Ask about family, and they don't stop.

This was the most striking pattern in the whole dataset. We can see which theme a conversation opened on, and how much the elder then said. The gap is enormous.

When the opening question was about career, elders answered with about 28 words on average. When it was about family, they gave about 205. Traditions and values-and-wisdom were close behind, both over 180. That is more than a seven-fold difference, driven entirely by the subject.

The same pattern shows up by question type. Reflective questions drew about 287 words on average; emotional ones about 198; plain factual ones only about 106. Nearly three times more talk for a question that invites reflection over one that asks for a fact.

Ask someone about their career and you get a sentence; ask about their family and you get a story.

A remembered life is mostly family and childhood

When we organize what an elder shares into life chapters, the true shape of a remembered life appears, and it looks nothing like a resume.

Of the chapters built so far, family (53) and childhood (37) dominate. Then come values and wisdom, turning points, and places. Further down sit love and marriage, traditions, and, quietly, loss and grief. Career appears in just 5 chapters, near the very bottom of the list.

We spend our working years measured by our jobs. Almost none of us, looking back, organize our lives around them.

From the inside, a life is mostly the people you loved and the place you grew up; the job comes last.

Memory lives in moments and people, not things

Our system quietly catalogues the people, places, things, and events an elder mentions. So far it has recorded 486 of them, and the breakdown is telling.

Events (179) and people (136) far outnumber places (110) and especially things (61). When elders reach into the past, they reach for what happened and who was there, not for what they owned. The objects come up only as carriers of a moment: a cycle because of who taught them to ride it, a radio because of the night the news came through it.

Memory is made of moments and people, not possessions.

Even in Hindi, they speak Hinglish

More than nine in ten of our language-tagged conversations happen in Hindi. A handful are in Marathi. Just three, out of 180, are in English.

But "Hindi" is almost never pure. About 78% of elder voice notes mix in at least one English word, the natural code-switching that any Indian family will recognize from its own dinner table. This is why we built Alfaaz to follow an elder between Hindi and English mid-sentence, rather than forcing them into one clean language. The moment you make someone "speak properly," they stop speaking freely.

Elders don't think in one language, and about 78% of their replies fold English into Hindi, exactly as they always have.

They say far more than anyone expects

The most common worry we hear from families is that their parent "won't have much to say." The data disagrees, firmly.

The average answer runs about 52 words. The longest single answer we have recorded ran 719 words, roughly five and a half minutes of unbroken memory, all from one question. And once elders begin, they come back: one has now recorded 24 separate conversations. Of the conversations we scored for emotional depth on a five-band internal scale, 51 of 53 reached "moving" or deeper.

The stories were never missing. The question was.

The fear that a parent "won't have much to say" is almost always wrong; given one specific question, they have hours.

What this means for your own parents

If you take one thing from this data, let it be this: the raw material is already there, far richer than you expect, and the only variable is whether someone asks well.

So ask about people and moments, not achievements. Ask one small, specific question at a time, not "tell me about your life." Let them answer by voice, in the language they actually think in. And don't wait for the right occasion, because the data shows the depth comes from the asking, not the setting. Our guides on the questions that open real stories and how to get your parents to open up go deeper, and you can see how Alfaaz does this on WhatsApp or give it to your own parents.

You don't need the right moment or the right words; you need one specific question, asked by voice, in their language.

A note on this data

These are early numbers, and we want to be honest about that. They come from 180 conversations with 20 elders in 16 families, recorded between March and June 2026. That is a small sample, and our families skew toward Hindi-speaking households, so the patterns may shift as we grow. Everything above is aggregated: counts, averages, and proportions, with no names, transcripts, or identifying details. The emotional-depth figures come from our own scoring rubric, not an external standard. We will revisit these numbers as the archive grows, and we will tell you if the story changes.