Mood whiplash is a real cost
Rapid, unrelated topic switches inside a feed create emotional whiplash. Holding a topic for longer reduces that cost and makes sessions feel calmer and more intentional.
Back to deepstory
deepstory research
The thinking behind the swipe.
deepstory is built on a simple behavioural bet: people often want more perspectives on the same topic before they move on. The research below is the thinking behind the product mechanic, the recommendation direction, and the business model.
Download deepstoryResearch questions
Why do people feel drained after short-video sessions even when they enjoyed individual clips?
What happens to discovery quality when the unit of recommendation is a topic instead of a creator?
Can intent signals (a deliberate same-topic swipe) outperform engagement-bait as a ranking input?
How do you give smaller creators a fair shot when relevance, not follower count, decides reach?
What we found
Topic as the unit of discovery
Shifting the unit of recommendation from creator to topic changes what 'more like this' means, it pulls in different creators' takes on one idea instead of more of one creator.
Swipes as intent signals
A deliberate same-topic swipe is a clean intent signal. It is more honest than dwell-time alone and powers both relevance and contextual, intent-led advertising.
Recommendation rebuild
Early retention and relevance problems pushed a rebuild toward vector intelligence, metadata, sound cues, narrative markers, and visual objects rather than pure follower graphs.
Industry benchmarks
Knowing what “good” looks like before spending a rupee.
Before scaling spend I built a benchmark of early-stage short-form platforms (TikTok, Reels, Shorts, Spotlight, and India's Moj / Josh / ShareChat) to know what 'good' looked like, and where a topic-first model could win.
| Metric | Industry benchmark | What it told deepstory |
|---|---|---|
| Day-1 retention | 25 to 35% good · 40%+ great | The a16z social-app bar deepstory's cohorts were measured against. |
| Like-to-view ratio | 4 to 10% TikTok · 2 to 5% Shorts | deepstory's 19.7% left-swipe rate sits well above the like-rate band. |
| Discovery source | ~10% of TikTok views come from followed accounts | Evidence that a topic-first, non-follower feed is viable, even dominant. |
| Video completion | 40 to 70% for successful platforms | The bar a calmer, topic-held feed has to clear to be sticky. |
The money
Built on roughly ₹18 lakh, marketing and product included.
deepstory was built on a deliberately small budget of about ₹18 lakh, covering both marketing and product development. The constraint was the strategy: build only what was necessary, give the algorithm exactly what it needed, and keep the burn low enough to keep iterating.
₹18 lakh
total budget
Marketing and product development combined.
₹2.45 lakh
paid marketing spend
Google Ads over 12 months, which drove 25,642 installs.
Small team
lean development
Only the most necessary features were built, on purpose.
Algorithm-first
where the money went
Spend was pointed at giving the recommendation engine what it needed.
Every rupee was scoped against necessity. That discipline is why a topic-first platform with a working recommendation engine and real install volume came out of a budget most teams would consider impossibly small.
Topic analysis
Which topics actually worked.
Not every topic behaves the same. Tracking the same-topic left swipe by category showed which topics actually pulled people into deeper exploration, the clearest signal of genuine topic-led demand.
Automobiles
The single strongest driver of same-topic left-swipe exploration.
Football
Consistent topic depth, people stayed and explored across creators.
Visual, identity-led topics
Fast, visual subjects like sport and cars led the explore action over passive categories.
The read was clear: topics people identify with personally drive the most intentional exploration, which is exactly where a topic-first feed should win.
What changes for you
What a person actually gets out of it.
Beyond the metrics, the real point of deepstory is what changes for the person using it.
Improved focus
Staying on one topic removes the emotional whiplash of random switching, so a session feels calm and deliberate instead of scattered.
Learn a topic quickly
Because related videos sit together, you can go deep on one subject in minutes instead of hunting across a random feed.
Faster access, stand out
Getting to the right content faster means you learn quicker, which helps you stand out among peers who are still scrolling at random.
Control in your hands
You decide what to watch and which topic to pursue. The feed follows your intent instead of pulling your attention wherever it wants.
Picture someone curious about football. They open deepstory, swipe left to stay on football, and in ten minutes have seen tactics, history, and skills from many different creators. They walk away having actually learned something, fully in control of where their attention went.
Living document
This page is a living research summary. Raj can expand each section with raw notes, user-interview quotes, and the metrics behind every claim.