How We Grew Dr. Shilpa Arora From 3,000 to 1 Million Followers
The first thing I did when we took on Dr. Shilpa Arora's Instagram was open her followers list and start scrolling.
I don't know why I did it. Habit, maybe. But about forty accounts in, I stopped, because I was looking at a wall of profiles with no photos, no posts, and usernames that looked like someone had fallen asleep on a keyboard. Numbers where names should be. Accounts following eight thousand people and followed by three.
The counter on her profile said roughly 3,000 followers. The counter was lying.
A big share of those three thousand were bots — leftovers from some earlier phase where somebody, at some point, had talked her into one of those "grow now, figure it out later" schemes that get pitched to every expert who's too busy doing actual work to smell the con. Strip the fake accounts out and the real, breathing, watching audience was close to zero.
Here's the part that most people don't get, and honestly the part I didn't fully get either until this project: that's not a bad starting point. That's a worse starting point than zero.
Why a dead audience is worse than no audience
If you have zero followers, the platform has no opinion about you. You're a blank file. Every video you post gets judged fresh — shown to a small test batch of strangers, and if they watch, it travels.
- Her 3,000 followers were mostly bots — a dead audience that actively fought the algorithm.
- The knowledge was expert; the edit wasn't. We closed the correct-to-compelling gap with craft.
- "Double-down" on breakout series took her to 1M followers and a 16M-view video.
But if you have three thousand followers and none of them ever watch anything? The platform has a very firm opinion about you. Every post goes out to that dead room first, gets nothing back — no watch time, no saves, nothing — and the algorithm reads that silence as a verdict. We showed this to her own audience and even they didn't care. Why would it push you to anyone else?
So a bot-inflated account isn't neutral. It's actively working against you. Every fake follower is a tiny anchor.
The first job wasn't growth. It was convincing a cold algorithm that real people wanted this — while dragging three thousand anchors behind us.
I want to be honest about something here. In the early days of With Media, I would have looked at "3,000 followers, decent-looking profile" and quoted a plan without ever opening that followers list. I undercharged, I said yes to everything, and I definitely didn't audit anyone's audience before starting. It took getting burned a couple of times — polishing videos for accounts that were structurally incapable of letting anything travel — before checking became step one. So when I say the counter was lying, I'm not being clever. I've been the guy who believed the counter.
What we actually had to work with
Now the other side of the ledger, because this matters for everything that follows.
Dr. Shilpa is a genuinely expert nutritionist and dietician. Not an influencer who read three books and bought a ring light — an actual expert, with science-backed, practical advice that regular people need and can use tonight. When she talks about food, she knows what she's talking about at a depth most health-content creators are only performing.
Which means the knowledge was never the bottleneck. It almost never is with experts. I've sat across from enough coaches and doctors and founders by now to say this with a straight face: the smartest person in the niche is very rarely the most-watched person in the niche, and the gap between those two people is not intelligence. It's translation.
Her content, before us, was what I'd call correct. The information was right. The advice was sound. And the videos were flat — no hook, no rhythm, no craft in the first two seconds, no reason for a stranger mid-scroll to stop being a stranger. Correct content informs the people who already follow you. Compelling content recruits the people who don't. She had plenty of the first kind and none of the second, which is exactly why she had an audience of bots and nobody else.
That gap — correct to compelling — was the entire job. Close it without diluting the expertise, because the moment a doctor's content starts feeling like clickbait, you've traded her most valuable asset for a few views. The trust is the product. The edit just has to make the trust travel.
I run a studio that does this for a living, so yes, obviously I'd frame the problem as an editing-and-craft problem. Biased. But watch what happened next and decide for yourself.
The unglamorous part nobody puts in case studies
Every case study you've ever read skips this part, so let me put it in.
The first stretch of this project was not cinematic. It was us posting carefully made videos into a room we knew was mostly empty, watching them do very little, and holding our nerve.
That's what a cold start actually feels like from the inside. You can't argue with the algorithm. You can't email Instagram and explain that the three thousand silent followers are bots and please judge the content on its merits. The only argument you're allowed to make is retention — real humans, from the small test batches the platform still deals out, watching longer than expected. Saving. Sharing. Watching again.
So that's what those early weeks were. Making every single piece as if it were going out to a million people, knowing it was going out to a few hundred, most of whom didn't exist. Reading the numbers on videos most people would call failures and looking for the one signal that mattered: are the real humans who did see it staying till the end?
There were days I questioned the whole plan. There's a specific kind of doubt that shows up when you know the work is good and the numbers don't agree yet — you start wondering if you're the delusional one. I've hired somewhere near a hundred editors building With Media, and one thing I've learned is that this exact stretch is where most editors, and most creators, quietly downgrade their effort. The video "isn't going to do numbers anyway," so why isolate the audio on the pour shot, why recut the hook a fifth time.
We didn't let that happen, and I'd love to tell you it was because of some steely founder conviction. Mostly it was because we'd seen the pattern before: the algorithm doesn't reward the breakout video. It rewards the body of work that made the breakout video statistically inevitable. Your job during the cold stretch is to keep the quality flat-lined at maximum so that when the platform finally deals you a good hand, the video it lands on can actually hold the attention it's given.
Then one held.
The diagnosis, properly
Before I get to the system, let me pin down the diagnosis, because "make it compelling" is the kind of advice that sounds useful and does nothing.
When I watched Dr. Shilpa's older content, the problem wasn't any single thing. It was that every element was doing the polite version of its job. The framing showed you the food; it didn't make you feel the kitchen. The information arrived in lecture order — context first, payoff last — which is exactly backwards for a feed where the viewer decides in two seconds. The sound was whatever the phone mic picked up. The text on screen, when there was any, could have belonged to ten thousand other accounts.
None of this is a criticism of her. Why would a nutritionist know that a ghee pour needs its own audio pass? That's my team's job to know.
But add it up and you get content that a follower will tolerate and a stranger will skip. And growth — all growth — comes from strangers. The people who already follow you can't make you bigger. Only the people who don't follow you yet can do that, and they owe you nothing. They haven't heard of Dr. Shilpa. They don't know she's a real doctor. They have a thumb hovering and about 1.5 seconds of grace.
Every decision we made after that was for the stranger with the hovering thumb.
Double-down: the system that did the heavy lifting
Most creators — and honestly most agencies — run channels like a slot machine. Make a video, post it, hope, make a completely different video, post it, hope. Every post is an independent gamble, and every week starts from zero.
We run channels on a system we call double-down, and Dr. Shilpa's account is the cleanest demonstration of it I can show you.
It works like this. The moment a piece breaks out — and by breaks out I don't just mean views; I mean more saves than usual, more shares, a retention curve that holds instead of sliding — we stop. We don't celebrate and move on to next week's idea. We sit with the video and interrogate it. What was the underlying topic, separate from this specific execution? What did the hook actually promise? Which shot were people rewatching? Was it the recipe, the routine, the ingredient, the claim?
Then — and this is the part that feels wrong to almost everyone — we make it again. Not the same video. The same winning concept, rebuilt into new scenarios: a different hook on the same topic, a different recipe through the same format, a different visual treatment of the same promise. Three, four, five variations, shipped while the algorithm is still hungry for that exact thing from that exact account.
When food-and-routine content started hitting for Dr. Shilpa, we didn't make one "What I Eat In A Day" and file it under done. We built it into a recurring, recognisable series. Then we spun adjacent winners off it — the Ghee Series, the Banana Smoothie, the Morning Routine — each one inheriting the audience and the algorithmic warmth of the last. A viewer who arrived through a ghee video landed on a profile full of things that person would obviously also want. So they followed. The platform saw arrivals turning into followers and pushed harder. Each series fed the next.
That's the difference between posting and campaigning. A post is a bet. A campaign is a position you keep pressing while it's winning.
Was it boring sometimes? Yes. Genuinely. There were weeks where the creatively exciting move was some brand-new format we were itching to try, and the correct move was ghee video number six. I remember the pull of that clearly, because I feel it on every account we run — the itch to be novel is the single most expensive itch in this business. New ideas feel like progress. Most of the time they're just variance. The discipline of double-down is choosing the sixth iteration of a proven thing over the first iteration of an exciting thing, and doing it again next week.
One breakout, doubled down on correctly, is worth more than ten unrelated experiments.
And no — before you ask — the audience did not get bored. That fear is almost always the creator's ego talking, not the data. Your followers see a fraction of what you post. The strangers the algorithm is recruiting have seen none of it. What reads as repetitive from inside the edit bay reads as consistent from the feed. Consistency is what makes an account feel like a channel instead of a diary.
The craft layer: what "cinematic" actually meant here
Double-down tells you what to make. It does nothing about whether the thing is worth watching. Strategy without craft is just a well-organised way to bore people. So underneath the system, every single piece carried a layer of craft decisions, and I want to walk through them properly, because "cinematic editing" has become a phrase that means nothing.
The grade. Kitchen content shot on a phone looks like a phone clip — flat, slightly blue, the food looking like evidence rather than dinner. We graded everything warm. Skin tones alive, the ghee golden instead of beige, the whole frame sitting closer to a streaming cooking show than a WhatsApp forward. This isn't vanity. Perceived production value buys trust before a single word lands, and for a doctor whose entire offer is credibility, that pre-verbal trust is worth more than any caption.
The sound. Food content lives and dies on sensory triggers, and sound is the trigger almost everyone ignores. We treated the food beats — the pour, the sizzle, the chop, the stir — like they were dialogue. Isolated, cleaned, mixed forward, ASMR-grade. You should feel the ghee hit the roti. I've sat through edits where the room went quiet at a pour shot, and that's when you know it's working — sound is the difference between watching food and craving it, and craving is retention.
The typography. We built her a signature on-screen text system — consistent type, consistent placement, consistent colour — so that MORNING ROUTINE or WHAT I EAT IN A DAY was recognisable as hers within one frame, even on mute, even mid-scroll, even cropped in a share. Most creators change their text style every month because the editor changed or the trend changed. Then they wonder why nobody recognises their content. A frame from her videos became identifiable the way a frame from a show is identifiable. That's not decoration; that's brand memory, built shot by shot.
The hooks. Every video's first two seconds were scripted and cut as their own miniature production, because they're the only two seconds that exist until they've done their job. We engineered them the way you'd engineer a headline — the promise up front, the payoff visible, the reason to stay landing before the viewer has consciously decided anything. We've recut hooks more ways than I'd like to admit. It's tedious. It's also where the entire result gets decided, so tedium seemed like a fair price.
None of these four is exotic on its own. Together, applied to every single piece with zero exceptions for eighteen straight months of output, they compound into something most channels never have: a floor. Her worst video still looked, sounded, and read like hers, at a level most accounts don't hit on their best day.
The results
Instagram went from that lying 3,000 to 1,000,000 followers — and it crossed half a million real, engaged followers in under six months. Not bought, not botted, not bartered through giveaways. Recruited, one hovering thumb at a time.
YouTube grew alongside it to 700,000+ subscribers, because the same system and the same craft layer travel across platforms with adjustments, not reinventions.
Her best-performing video crossed 16 million views — the single most-watched piece in our entire portfolio, and I'd point out that it came from inside a series, not from some lightning-strike one-off. That's double-down working exactly as designed: the breakout wasn't an accident we got lucky with; it was the sixth or seventh press on a position we already knew was winning.
And under the headline numbers, the hero posts that compounded the run: Morning Routine at 58.5K, Ghee Ka Roti at 77.9K, the Ghee Coffee and Banana Smoothie pieces, What I Eat In A Day. Individually, nice numbers. As a sequence, they were the ladder — each one warming the account for the next.
From an audience that was mostly bots to a million real followers, with a video watched more than sixteen million times. I've produced well over a thousand videos through With Media at this point, and this run is still the one I reach for when someone asks me what the studio actually does.
"His creative side, his work speaks for him. An inspiration to younger people. Going a long way."
Dr. Shilpa Arora
The objections a smart reader should raise
Let me argue against myself for a minute, because I would, if I were reading this.
"She's a real expert with real knowledge — of course it worked." Partly fair. The expertise was necessary. It was also sitting right there for years, attached to three thousand bots. Expertise is fuel; it is not a vehicle. If knowledge alone grew channels, every professor in the country would have a million followers.
"Health and food content grows fast anyway." The niche helps, no argument. Food is visual, universal, and endlessly shareable. But the niche was equally available to her before, and equally available to every other nutritionist posting flat kitchen clips into the void right now. A rising tide doesn't lift boats with holes in them.
"Would this work for me?" Honestly — it depends, and anyone who says otherwise is selling something harder than I am. The system transfers: audit the audience, find the correct-to-compelling gap, double down on breakouts, hold the craft floor. What doesn't transfer automatically is the timeline. Some niches warm slower. Some faces take longer to trust. If you're a B2B consultant, your version of the ghee pour is different, but you still have one — some moment of proof your content either delivers viscerally or doesn't. I won't pretend every account we touch does Shilpa numbers. What I will say is that the accounts that fail usually fail at the discipline, not the strategy. They get the plan and then chase a new idea in week three because iteration got boring.
"Couldn't she have just deleted the bots and posted more?" Volume without craft is just a faster way to teach the algorithm you're skippable. And bot-removal helps hygiene, but it doesn't create the thing that was actually missing — a reason for strangers to stay.
What this run actually proves
I keep coming back to that afternoon scrolling her followers list, because everything the project became was already visible in it. A world-class expert. A number that lied. And a gap in the middle that nobody had bothered to name.
Three things I'd want you to take from this, especially if you're a creator or a coach staring at your own plateau.
Your follower count is vanity until it's real. A clean, engaged ten thousand beats a hollow hundred thousand every single time, because one of those audiences pushes you uphill and the other one is an anchor. Sometimes the first job is subtraction, not addition — and it's worth asking, honestly, how much of your number would survive the scroll test I did that first day.
Expertise doesn't go viral on its own. It never has. The most-watched person in your niche is not the smartest one; they're the one whose knowledge got packaged with the intention of a film. That's not an injustice to resent. It's a lever to pull.
And compounding beats chasing. The unglamorous sixth iteration of a working idea will outperform the thrilling first iteration of a new one far more often than feels fair. Growth on this account was never a spike. It was a curve — and curves are built by people willing to be a little bit bored.
That's the whole thesis behind With Media, really. Not transitions, not zoom-ins, not motion graphics that twirl. A system, a craft floor, and someone who actually gives a damn whether the video lands.
If you want to see what one piece from this run looks like up close, I've embedded "3 Dinners in a Day for Weight Loss" below — watch the first two seconds twice, and then watch what the sound is doing under the food. That's the stuff this entire essay was about, compressed into one video.
