Most founders mistake early traction for real GTM signal. They are not the same thing - and acting on that confusion is what quietly breaks GTM machines before they are built.
What this covers: What genuine GTM signal is and what founders routinely mistake for it. The six behavioral markers of real signal. How to actively hunt for signal in Phase 1. Three India-specific examples of what signal looked like before scale. And the eight questions that close Phase 1 and unlock Phase 2.
Who it is for: Early-stage founders, product leads, and growth operators in Indian startups trying to understand whether what they have is signal - or something that looks like signal under the right lighting.
The core argument: You cannot scale what you do not understand. Signal is the understanding. Finding it is not an event. It is a discipline.
When we built the waitlist for OneCard - India is premium metal credit card - we hit 75,000 signups in 15 days. The number looked like signal. It felt like signal. The team was energized in the way teams only get when the data looks good.
But the number was not the signal.
The signal was something quieter, and more specific. It was the subset of waitlist members - roughly 12% of them - who referred a second person before they had even received their own card. They had not yet experienced the product. They had not used a single feature. But they were already distributing it. That behavior told us something the 75,000 number could not: that the positioning had landed with enough force to create advocates out of people who were still waiting.
That 12% became the blueprint. We studied who they were, what had brought them to the waitlist, what they had said in their referral messages, what credit card they currently held. Every activation decision, every onboarding message, every product priority in the next six months was optimized around recreating that behavior - not in 12% of users, but in as many as we could reach.
That is what signal actually is.
Not the spike in downloads. Not the press coverage that moves a number for a week. Not the cohort that was acquired through a specific discount that will never run again. Signal is the repeatable behavioral pattern that tells you who your product is really for, why they actually want it, and how they naturally spread it.
Everything else is noise wearing signal is clothes.
GTM signal is a repeatable behavioral pattern - not a metric - that tells you why your best users came, what triggered their activation, and what causes them to stay or refer. It is specific, observable, and reproducible. If you cannot answer those three questions with real behavioral data, you have not found signal yet.
Most early-stage founders mistake one of three things for signal:
Spike traction. A product launch gets press. Downloads jump. The cohort looks strong for three days, then falls off the same cliff as every cohort before it. The press cycle is not signal. It is borrowed attention. Signal is what survives after the press cycle ends.
Discount-driven usage. A cashback campaign drives strong first-week engagement. Retention at Day 7 looks better than it has ever looked. But retention at Day 30 returns to baseline the week the cashback ends. The cashback was not signal. It was rented behavior. Signal is behavior that persists without financial incentive.
Surveyed enthusiasm. Users say they love the product. NPS is high. But actual usage frequency is low, and no one is referring anyone. Stated intent and revealed behavior are different data sets. In consumer products in India - where users are often more polite than they are honest in formal surveys - revealed behavior is always the more truthful signal.
The test for real signal:
Remove the thing you think is driving the behavior. Does the behavior persist? If yes, you have signal. If it collapses, you were measuring the driver, not the behavior.
India is consumer behavior creates four specific conditions that make GTM signal harder to read - and far easier to misread - than any Western startup playbook accounts for.
I have seen the same mistake in post-mortems from multiple fintech teams. Aggregate metrics looked flat. The team concluded the channel was not working and shut it down. In two of those cases, a cohort analysis run after the fact revealed that one specific segment - metro women in their late twenties, users who arrived through a vernacular creator - was retaining at twice the national average. They had abandoned a working signal because they were looking at the wrong layer of data. Aggregate India hides the signal. You have to go one level deeper before any pattern becomes legible.
Here are the four conditions that create the misread:
Trust-delayed adoption curves. In categories like fintech, health, and lending, Indian users often take significantly longer to convert from awareness to activation - not because the product is wrong, but because trust has not been established yet. A product can be in consideration for weeks or months before a user takes the activating action. This means early adoption metrics undercount real interest, and teams often give up on channels or segments just before the trust threshold is crossed.
WhatsApp-driven discovery that does not show up in attribution. A significant portion of consumer discovery in India - especially in Tier-2 cities and vernacular segments - happens through WhatsApp forwards, group shares, and voice messages that are invisible to standard attribution models. A product can be growing through organic trust propagation while the analytics dashboard shows "direct" traffic and the team concludes nothing is working.
Price sensitivity that masks preference. Indian users will use a free product with low frequency. They will use a product they genuinely value with high frequency, even at a price. Frequency of use on a free product is not signal. Willingness to pay - or dramatically increased frequency after a specific feature ships - is signal.
Heterogeneous markets that look like one market. An activation rate that looks flat nationally may be hiding a strong signal in one specific segment - metro professionals, or women in Tier-2 cities, or users who arrived via a specific creator. Aggregate metrics obscure the signal that lives inside the segments. The teams that find signal earliest are usually the ones who refuse to look at aggregate numbers until they understand what is driving them.
Finding signal is not about waiting for a number to reach a threshold. It is about observing specific behaviors across enough users that a pattern becomes undeniable. These are the six behavioral markers I look for - and none of them are volume metrics.
1. Unprompted return.
Users come back to the product without a push notification, an email, a re-engagement campaign, or a discount. They return because the product created a reason to return. This is the most basic signal - and it is surprisingly rare in early cohorts. The question is not whether users return when pushed. It is whether they return when you go quiet.
2. Organic referral before full value delivery.
Users refer other users before they have experienced the product is full value proposition. This happened with the OneCard waitlist. It happens with fintech products where users forward a link to a family member before they have even received their first statement. If users are willing to put their social credibility on the line for a product they have barely used, the positioning has landed at a depth that most products never reach.
3. Specific language in organic feedback.
Users describe the product in their own words - in WhatsApp forwards, in the App Store, in social comments - and the language is specific, not generic. "It is like a credit card for people who actually care about design" is signal. "It is a good app" is not. Specific language means the product has created a distinct mental category. Generic praise means the product exists but has not yet differentiated.
4. Behavioral intensity in a narrow use case.
A small number of users are using one specific feature or completing one specific flow with dramatically higher frequency than the average. This is a signal about where the real value lives - which is almost never where the founder thinks it is. Zerodha found early signal not in trading frequency but in Varsity completion rates: users who finished an investing module were converting to funded accounts at a rate that made everything else look like noise. The reading behavior was the signal. The trading was the outcome.
5. Retention inflection in a specific cohort.
Not all cohorts are equal. A retention curve that is mostly flat may contain one cohort - from a specific acquisition channel, a specific city, a specific referral source - whose Day-30 retention is two or three times the average. That cohort is the signal. Everything before finding that cohort is research. Everything after is about understanding what made it different and reproducing it.
6. Proactive problem-sharing.
Users surface problems before you ask. They send screenshots. They describe edge cases in support tickets with the specificity of someone who cares. They post in communities not to complain, but to ask if others have figured out a workaround. This behavior indicates investment. Users who do not care do not spend this kind of energy. High-quality complaint behavior - detailed, specific, and constructive - is one of the clearest signals of a user who values the product enough to want it to be better.
What these six markers share:
They are all behaviors that cost the user something - social credibility, time, or emotional energy. Behavior that costs the user something real is signal. Behavior that costs them nothing (a one-click rating, a passive scroll) is noise.
Signal is not found passively. It is hunted - through structured observation, deliberate experimentation, and a discipline of looking at behavior before you look at metrics.
Most early-stage teams make one critical error: they instrument for growth before they instrument for understanding. They set up dashboards that track installs, signups, and revenue. They do not set up the qualitative infrastructure needed to understand why the numbers move.
Here is what has worked when I have tried to hunt signal from scratch - and what I watch for when I am advising a team in Phase 1:
Run behavioral interviews before you run campaigns.
In the first 500 users, you should be talking to 30-50 of them directly. Not surveys - conversations. Ask them where they were when they decided to try the product. Ask them what they were hoping it would do. Ask them to walk you through the moment they first felt it working. These conversations will tell you more about where signal lives than any dashboard you build.
I did this during COVID - right at the point when we were rolling out OneCard to our waitlist. Every day, I called four or five users personally. Not to pitch. Not to retain. To listen. I would ask what they were hoping the card would do for them. I would ask what frustrated them in the first week. I would ask - directly - what it would take for them to tell a friend about it without being asked. Most teams never ask that last question. It is also the most important one.
What surprised me was not the answers themselves. It was how specific users could be when someone actually called. They had opinions they had never shared anywhere - not in reviews, not in support tickets. They had been waiting for someone to ask. The calls were not scalable. That was the point. Unscalable closeness to the user is the fastest way to find signal. You systemize it later.
When full rollout began, I built the same discipline into the onboarding team. Every onboarding call had the same three closing questions: What was confusing? What did you like? What would you need to feel comfortable recommending this to someone you know? The third question is what changed the onboarding. The answers came back in two or three patterns, consistently, across hundreds of calls. The onboarding experience was rebuilt around those patterns - not around what I assumed users needed, but around what they said, in their own words, on the phone. The retention improvement in the cohorts that followed was not because of a better campaign. It was because I finally understood what users needed in order to stay.
Build cohort clarity before you build cohort size.
A 200-user cohort you understand deeply is more valuable than a 2,000-user cohort you understand superficially. Smaller, more studied cohorts in Phase 1 accelerate Phase 2 because the system you build is based on real behavioral understanding - not on statistically comfortable but shallow data.
Create controlled variation to isolate signal.
Run two or three acquisition experiments simultaneously - different sources, different messages, different audiences - with cohort tracking turned on from day one. You are not optimizing for conversion. You are learning which source produces users who behave differently. The cohort that retains better, refers more, or reaches activation faster is the segment where your signal is strongest.
Track the moment, not just the milestone.
Most product analytics track milestones: account created, first transaction, first referral. Signal lives in the moment before the milestone - the specific action that predicted the milestone. What did the users who completed KYC have in common that the drop-offs did not? What happened in the 90 seconds before a user sent their first referral? Instrumenting the moment gives you a lever. Instrumenting only the milestone gives you a count.
Give users a reason to reveal preference.
A/B test your messaging. Offer two different value propositions and see which one drives the activation behavior you care about - not the click. In India specifically, this is how you discover whether your product is being adopted for the reason you think it is. Often it is not.
The signal-hunting rule that overrides everything else:
Before you instrument for growth, instrument for understanding. Every dashboard built before you understand why users behave the way they do will tell you what is happening - never why. And in Phase 1, why is the only number that matters.
Three examples I keep returning to - not because they are famous, but because the signal each company found was non-obvious, easy to have missed, and in each case required someone close enough to the user to see the behavior before a metric captured it.
CRED is early invite-only structure meant that every new user arrived because someone they knew had chosen to extend an invitation. The signal CRED found early was not retention or engagement metrics on the app itself - it was the social behavior around the invitation. Users were treating the invite as a status object. They were being selective about who they invited, which meant acceptance carried implicit social endorsement. CRED understood that the trust architecture was the product - before the rewards, before the features, before the full app experience existed. The invite behavior was the signal. Everything else was built to protect and scale it.
Zerodha launched Varsity - their investor education platform - before most people understood why a brokerage would build free content. The signal they found was specific: users who completed a Varsity learning module converted to funded trading accounts at a rate that was dramatically higher than users who came directly to the trading platform. The educational behavior was predicting the financial behavior. Zerodha built its entire GTM architecture around this - content as trust-building, trust-building as acquisition, acquisition as the start of a compounding loop that did not require performance marketing to sustain.
Meesho is early reseller network in Tier-2 and Tier-3 India revealed a specific behavioral signal: some resellers were actively sharing products in their communities without any direct financial incentive in that moment - they were building social credibility as curators. The signal was not the transaction. It was the social act of recommendation before the transaction. Meesho built its distribution model around enabling and amplifying that behavior - the reseller as a trusted node in a community trust network. The financial incentive came later, and it was designed to reward behavior that was already happening naturally.
What these three examples share: the signal was behavioral and non-obvious, it appeared in a small subset of early users, and it required founders who were close enough to the user to observe it rather than just measure it.
You are ready to move from signal to system - from Phase 1 to Phase 2 - when you can answer these eight questions with behavioral data, not instinct.
These are the questions whose answers become the architecture of the GTM machine you build in Phase 2. If you build that machine before you have the answers, you are building on assumption. And in India, where the cost of a wrong GTM assumption is paid in CAC that never recovers and cohorts that never compound, that is an expensive way to learn.
→ Who are your best users?
Not your most frequent users. Your best users - the ones who stay, refer, and expand their usage over time. What do they have in common demographically, behaviorally, and in terms of acquisition source?
→ What triggered their activation?
Not what feature they used first. What was the specific moment - the action, the result, the realization - that made them feel the product was worth staying for?
→ What is driving organic referral?
If referrals are happening, why? What are users saying when they refer? What is the emotional trigger - social credibility, financial benefit, genuine delight? If referrals are not happening yet, why not?
→ Which acquisition source produces your best cohort?
Not your biggest cohort. Your best cohort - by Day-30 retention, by activation rate, by referral rate. The channel is less important than what it produces.
→ What does your retention curve tell you about your product?
A curve that falls sharply at Day 7 indicates an onboarding or first-value problem. A curve that falls at Day 30 indicates a habit formation problem. A curve that is flat but low indicates a positioning problem. Each diagnosis requires a different solution.
→ What is your time-to-first-value?
How long does it take a new user - from first open to the first moment they felt the product working? If you cannot answer this precisely, you have not instrumented the right moment.
→ Would your best users still come if your performance marketing disappeared?
This is the most important question in Phase 1. If the answer is no - if your best users are best because they arrived through a paid incentive that would disappear with the budget - you have not found signal. You have found rented behavior.
→ Can you describe your GTM in one sentence that a stranger would immediately understand?
Not your positioning statement. Your actual GTM - the mechanism by which a new user hears about the product from someone who already uses it, decides to try it, and becomes the next person to tell someone else. If you cannot describe that chain in one sentence, it does not yet exist as a system.
When you have real answers to all eight - answers that come from data, from user conversations, from cohort analysis, not from hypothesis - you are ready to build the machine.
That is what the next essay is about.
Here is what I keep returning to after every early-stage conversation, every Phase 1 review, every team that scaled before they should have.
The urgency to scale is understandable. Investors want to see growth. Competitors are moving. The team needs to feel momentum. But scaling before signal is understood does not create momentum. It creates speed in a direction that has not been verified.
The founders who build the most durable GTM machines in India are not the ones who scaled fastest. They are the ones who were most disciplined about understanding signal before they acted on it. They talked to more users than their peers. They ran smaller, more instrumented experiments. They refused to move to Phase 2 until Phase 1 had taught them what they needed to know.
That patience is not a strategic liability. In a market as heterogeneous, trust-driven, and behaviorally complex as India - it is the most important competitive advantage available to an early-stage team.
Signal is not found. It is earned. Through observation, through patience, through the discipline to look at behavior before you look at metrics.
Find the signal. Then build the machine.
← This essay
Finding the Signal
Next →
From Signal to System: Scaling GTM in the AI Era
GTM signal is a repeatable behavioral pattern - not a volume metric - that tells you why your best users came, what triggered their activation, and what causes them to stay or refer others. In India, signal must be read from behavioral data rather than stated preference, because trust-driven adoption creates delayed and non-linear curves that aggregate metrics obscure.
Deep dives on growth systems, fintech strategy, and building compounding distribution in India. New essays every month.
IN THIS ESSAY