Vikash Kumar

I solve growth problems by building systems that compound.

Brand and marketing matter because they create demand. Demand alone is not growth. Product, retention, operations, and unit economics decide whether that demand turns into a durable business.

I do not do growth hacking. I run rapid experiments to find where the system is weak, then build the useful learning into the next cycle. The goal is a loop that keeps getting better, not a spike that has to be recreated every month.

I now split my time between operating consumer businesses and building Dataflux Labs, an AI-native growth agency. I live in Bengaluru.

Work

At CRED, I was responsible for brand and performance together, including product launches, sponsorships, measurement, and the marketing P&L. The interesting part was not any channel on its own. It was understanding how brand created demand, how performance captured it, and how measurement changed the next allocation.

At Dream11, the same system operated at much larger scale. We built the marketing analytics and technology needed to test incrementality, improve media efficiency, and carry the learning into new launches such as FanCode.

At NEWME, my remit widened to the online P&L. Retention, quick delivery, returns, and contribution margin became growth problems too. Earlier, nearbuy.com taught me how acquisition and product feed each other. Syntel taught me to look for the process underneath the result.

The thread across all five roles is simple. Run experiments quickly, keep what they teach you, and turn the learning into a loop the business can repeat.

CRED, Dream11, NEWME, nearbuy.com, Syntel

Dataflux Labs

Dataflux Labs is where I apply the same approach with other businesses. We find the constraint in a growth system, test ways around it, and automate the work that repeats.

We are not selling AI for its own sake. It is useful when it shortens the learning cycle or makes a good process repeatable. When a workflow keeps proving useful, we turn it into software with opinions built in.

We have worked with Oolka, PICE, Frex, Tribe Money AI, August AI, BukuWarung, Knot Now, and Punch Trading.

Creobra A Meta creative audit that finds fatigue, weak hooks, and missing formats.
Creative IQ Ranks creatives to scale, pause, watch, or replace.
SignalOps and Meta Ads Uploader Internal tools that we still use before we talk about them.

Writing

I write to see whether an idea survives a full page.

The boring work behind the visible 20% The 80/20 rule is good at describing yesterday. It is less useful for deciding what to skip tomorrow.
Use the analogy. Then check the mechanics. A comparison gets an idea into the room quickly. Reality still gets the final word.
Builders and optimisers need different clocks New work needs room for waste. Mature work needs someone to remove it.

All writing · RSS

A few things I believe

Brand and performance are parts of the same demand system. Treating them as rival camps makes both worse.
I do not care about an experiment that creates a temporary spike and teaches us nothing.
A good growth loop makes the next cycle smarter than the last one.
Compounding systems look unimpressive at the beginning. That is why most teams abandon them early.

Books

I am drawn to books about long arcs and stubborn problems. I like people who keep working after the early excitement disappears.

The Goal by Eliyahu M. Goldratt The clearest book I know on constraints and improving the whole system.
Shoe Dog by Phil Knight A reminder that compounding businesses look chaotic while they are being built.
Open by Andre Agassi Repetition is not glamorous, but it is where judgment and performance come from.
Fermat's Last Theorem by Simon Singh Some problems give way only after years of accumulated work.

Contact

Got a growth problem that will not behave? Email vikash@datafluxlabs.com. Book recommendations are also welcome.