# Starting a startup: from first idea to product-market fit — Summary The Founders work group answers questions about starting a company in group chats. Members cover finding a problem and testing an idea, where a new company can win, and how to tell when it has product-market fit. Updated: 2026-10-13 HTML page: [Starting a startup: from first idea to product-market fit]() Transcript: [Full Chat Transcript]() Work group: [Founders]() ## Questions ### [Reading the early signs of product-market fit]() Fit tends to show first as a small group who would hate to lose the product, then as growth you can explain, from users who look like the market you want. - **A few people would miss it.** Ben Silbermann had about 3,000 Pinterest accounts three months after launch, but a core of users loved it and called him when the site went down. He would watch who keeps coming back over who he expected to adopt it. - **Read disappointment next to NPS.** Shishir Mehrotra tracks both. Early Coda scored high on disappointment and low on NPS, and Superhuman showed the reverse. He expects a young, differentiated product to show its first real signal in disappointment scores. - **Growth settles the market question.** Emmett Shear set a 20 to 25% monthly growth bar for Justin.tv's gaming project. It grew 30% a month and kept growing for almost nine months before it became Twitch. He would apply that bar once the product works for its best customer, and calls good press plus "kinda working" the most common false positive. - **A churned customer can still be a believer.** Uber's team dropped Coda in a day and handed over about 30 fixes. Emmett reads a list like that as a believer when the items point to one underlying job, and as a different product when they scatter. - **Fit comes in stages.** Reid Hoffman separates initial fit from scale fit, and asks founders for evidence of the move from beachhead to whole market. Ivan Zhao saw no single moment at Notion: people cared, then reached out, then paid, and growth bent after Notion 2.0 in 2018. ### [Telling early traction apart from product-market fit]() Traction responds to something the founder did. Fit compounds through the product's own loop, and it rarely arrives as one moment. - **Launch spikes fade.** Emmett Shear's Kiko got 90,000 signups over a year and a half after a TechCrunch feature, and growth stopped when the press did. Twitch grew 25 to 30% a month for almost nine months against a bar set before the data came in. - **Name the loop, then test it.** Reid Hoffman asks founders for their theory of compounding loops and suggests shutting off your own pushing for a few weeks. Shishir Mehrotra would run that test on champions instead, since a product that changes how a team works can stall without help even when its users love it. Emmett, who never stopped pushing at Twitch, came round to Reid's return-on-effort version: whether each push pays back more as the network fills in. - **Check that the love repeats.** Ben Silbermann ran Pin It Forward with about 300 bloggers and started believing once strangers who matched his early fans loved Pinterest at the same rate. He would set any advance bar on depth: how many people write in when it breaks, and how often they come back. Emmett agreed for a new product. - **Watch how new users arrive.** About 99% of Notion's growth came from word of mouth when Notion had a team of 25, so Ivan Zhao could read growth as pull. He would count how many new team workspaces start with someone who was already a personal user. ### [What to do when the search for product-market fit stalls]() Make the next attempt a bounded test: pick the strongest group of users, set a target and a deadline the cash can cover, and decide in advance what a miss means. - **Bound the next attempt.** Emmett Shear would pick the strongest group of users found so far, set a usage target and a deadline that fit the cash, and change course on a miss. With no promising group or no runway for a real test, he would return the remaining capital. - **Make the test fair.** Ben Silbermann would first define one experience worth someone's time, since a weak first version can make a useful idea look like a failed one. Reid Hoffman would check that the test group had enough people for the product to work. LinkedIn's sparse early network gave each new member little value. - **Talk to the users who return.** Ben would learn what returning users rely on the product for and look for more people with that need. Ivan Zhao suggests putting the founder on support, as he did at Notion, and kept Notion's goal while changing the product built around it. - **Know who pays for the search.** Emmett describes a third option: make the business pay for its own search, as Justin.tv did by reaching profitability. Reid would put the plan, its cost and a judgment date in front of the team and board, and arrange a responsible transition if the case can't be made. ### [Pressure-testing a startup idea before you commit]() Set a behavior bar before the team gets attached, watch whether people return to a real task, and write down how the company could fail even if early users love it. - **Ask questions that can change the idea.** Emmett Shear called 40 to 50 streamers and learned that even a few dollars of income mattered to them, which led Justin.tv to share ad revenue. He would specify in advance what repeat use or growth justifies more time and money. - **Put it into someone's real work.** Shishir Mehrotra brought in Coda's first outside customer four months after the team moved its own work into it, and that team's refusal to continue exposed problems internal use had missed. Ivan Zhao would ask someone to finish a familiar task in both the new product and the old one, and watch where they switch back. - **List the ways it could fail.** Reid Hoffman would write down two or three ways the company could fail even if early users love it, with a plan and a fallback for each. He would also ask founders to name the next proof point and its date: PayPal moved to the Palm Pilot because a version could work in about six months. - **Test your own commitment.** Shishir asks whether you have an idea you can't leave alone and a teammate you can't build it without. Ben Silbermann would give the idea a bounded, serious period of work, then ask whether he still wants the hard parts he has met. ### [The warning signs that should end a startup idea]() Drop or change an idea when the people who have the problem can find it, try the full version, and still have no reason to come back. - **Nobody comes back.** Ben Silbermann would drop or change an idea if repeated conversations and improvements leave no small group that chooses to use it again. Low signups alone would have sent Pinterest away from a real audience at about 3,000 accounts. - **Judge the full claim.** Coda's first customer went from six daily users to zero after seeing mostly the tables. Shishir Mehrotra counts repeated rejection by customers who tried the full idea as a kill signal, and feedback on a slice as a list of what to fix. - **The customer stays vague.** Emmett Shear would stop a team that keeps switching products while describing its customer as "everyone who uses a calendar," as Kiko did, and choose one customer first. Investor rejection is a weaker signal: 40 VCs turned down Twitch while people were using it. - **The business doesn't hold up.** Ivan Zhao would ask who pays and why, and whether the cost of delivering the product leaves room to keep serving them. Reid Hoffman would reconsider when the next plan has worse odds than the first, or when funding forces a business model customers can't back up, as at SocialNet. ### [Where founders find a problem worth a company]() Start with a problem you will care about for years, find people whose behavior shows they have it too, and know why you can make progress where others see risk. - **Start from a need you have.** Ben Silbermann and his cofounder built Pinterest as a place they wanted to collect things, and users who wrote in when it went down gave them a reason to commit. Ivan Zhao started from making portfolio websites for artist friends. - **Study people who do something intensely.** Emmett Shear suggests asking about behavior and motives before pitching an idea. Calling 40 or 50 streamers showed that even $2 a month mattered to them, and Justin.tv built ways for streamers to earn. - **Know why you can win.** Reid Hoffman looks for a change he understands before it seems obvious, and took LinkedIn to smart critics to learn what would break. Emmett saw that game publishers had reason to welcome streams, which made gaming a better bet than general live video. - **Keep the aim, change the first product.** Notion's first two years on a developer tool did not resonate, so Ivan moved to productivity software and kept the aim of letting people make their own tools. Shishir Mehrotra had to build Coda to show what its thesis meant, because the pitch alone left people asking what was wrong with their current tools. ### [The first moves for a first-time founder]() Begin, choose a cofounder with care and get something in front of strangers early. The group splits on how much diligence and polish should come first. - **Do the diligence before the code.** Shishir Mehrotra would treat the company as an investment you can make only once: a team that can't picture losing, a space that matters and a secret good enough to start. He asks would-be founders to write "Things I'd like to see different in the world" and then talk to people already in the space, competitors included. - **Ship something rough and learn from it.** Emmett Shear's order is a cofounder, a rough product, then user conversations meant to learn rather than confirm. A gamer who wanted to make $17 a month streaming led Justin.tv to an ad button that split the money 50/50. - **Purpose before thesis.** Reid Hoffman suggests a short written theory of the game, tested against smart skeptics. For a founder with no idea yet, he would start with a statement of purpose and open-ended conversations, and write a falsifiable thesis once an idea shows up. - **Make one thing worth a stranger's time.** Ben Silbermann's first app, Tote, crammed in too many features and had nothing special about it. He would build something you want yourself and ship once one thing is worth a stranger's time. Ivan Zhao sides with Ben on that bar and with Emmett on speed. Notion spent two to three years on a premise most people didn't want, so he would first check that strangers have the problem. ### [Sizing a market you haven't entered yet]() Start from a behavior people already have, ask why nobody has won the market yet, and let growth or repeat use tell you how big it is. - **Set a ceiling with a ratio.** Ivan Zhao never built a bottoms-up TAM for Notion. He used the gap between 1B+ knowledge workers and about 25M developers, and notes as a daily habit for 1B+ people. He would find a behavior with a long history and put the idea inside a familiar form. After Ben's pushback, he counts that history as the part of the test that carries weight, with a paid predecessor as a stronger signal for a business product. - **Research the behavior first.** Ben Silbermann thinks a sizing exercise in 2009 would probably have talked Pinterest out of it. Collecting was the real opportunity. He would watch whether an unexpected use, like Pinterest's boards of inspirational quotes, holds up as the product grows past early adopters. - **Ask why nobody has won it.** Shishir Mehrotra writes down the assumption the market has made that his team sees differently, then asks whether it's true and what has kept anyone from acting on it. For Coda, file formats had been the lock-in, and web tools removed it. An idea Microsoft could ship in a release is not a company, in his view. - **Let growth stand in for size.** Emmett Shear funded Justin.tv's gaming and mobile side projects with growth goals set in advance. Gaming grew 25 to 30% a month for almost nine months before Twitch launched. In a two-sided market, he would research first the side the other side is loyal to. - **Write down what would prove there's no market.** Reid Hoffman would write the thesis before building: what you believe that others don't, what would show there is no market, and how you will get fast adoption anyway. Hardware needs harder evidence before you commit capital than software does. ### [Standing out when every product looks the same]() Pick one experience people return to, start with the people whose choice brings the rest, and check whether early users come back for a second job. - **Name the thing users come back for.** Ben Silbermann would ask whether you can name what first users return to do and where it helped their lives. Early Pinterest users planned weddings and redecorated homes with it. He would also watch which users get asked by a friend where an idea came from. - **Start with whoever brings the others.** Emmett Shear saw that viewers followed streamers, so he called 40 or 50 of them and Justin.tv built an ad button that split revenue 50/50. The team recruited streamers one by one and sometimes offered larger ones $1,000 to switch. - **Plan for density.** Reid Hoffman asks which group gets value first, what improves as each new customer joins, and how you reach enough of them before early users leave. He estimated LinkedIn needed about a million members before the first member had a fairly functional value proposition. - **Own a finished job.** Ivan Zhao would start with a frequent job, make the first useful result easy to reach and give people a reason to stay. Notion found its way in through notes and documents. With business customers, it learned that selling flexible parts alone asked them to do too much setup. - **Watch the second use.** Shishir Mehrotra would track whether a team that used the product for one task builds its next recurring process there. A team at Pinterest began with meeting notes in Coda, and later someone built a goal-tracking system in it. ### [What an AI startup can build that a model company won't easily copy]() A model company can ship the feature. The group would build around what takes sustained work: a group's accumulated taste, a team's workflow, trust between people or a creator's audience. - **Collect taste that compounds.** Ben Silbermann suggests a product where people keep adding their own taste and intent, like a garden planner that records what each person planted, season after season. For a sensitive task such as naming an emotion, he would add expert review and careful boundaries, drawing on How We Feel, whose team includes scientists and therapists. - **Own a team's workflow.** Ivan Zhao would pick one complete workflow, such as an agency's client intake and approvals, that a team can use on day one. A ready-made assistant is easy to copy, so he would let the consultants who set it up keep improving it across clients, as consultants did with Notion templates. - **Build on trust between people.** Reid Hoffman's example is a referral marketplace that needs clinician participation, verified availability and patient consent. He calls a hospital data deal alone a weak moat, and would test whether referrals get better as participation grows. - **Measure more than output quality.** Emmett Shear points to Kiko, whose opening disappeared when Google shipped a calendar attached to Gmail. For a creator product, he would measure whether creators come back because it helps them earn, get discovered and talk with their audience. - **Connect the systems a team already uses.** Shishir Mehrotra suggests customer renewal briefs that draw on the account record, support history and roadmap, placed where the account team already prepares. The test is keeping trusted access across those systems and earning regular use. ### [Putting an early startup idea in front of real customers]() Go out to learn rather than to confirm, watch who comes back on their own, then ask them why. - **Ask about their problems.** Emmett Shear interviewed streamers and followed each feature request by asking how their life would be better. Customers know their problems but are bad product designers, he says. His cheap check is to name the five people a product is for and be one of them or have their phone numbers. - **Watch who returns unasked.** Ben Silbermann's Silicon Valley friends shrugged at Pinterest, while a small group in Iowa and in creative fields used it to plan homes and meals. He would ship one thing worth a stranger's time, and set in advance the months and dollars you'll spend before a small group should be coming back on their own. - **Test the form the idea takes.** Ivan Zhao says Notion spent years on a developer tool most people didn't care about. Looking back, he would first have asked friends and small teams what they use every day and what they keep fighting with. - **Count repeat use.** Reid Hoffman would write the idea as a thesis of testable beliefs and launch a version you're embarrassed by. Socialnet got good reviews but paid too little attention to whether users engaged with each other, and he counts repeated use as validation. - **Judge feedback against the vision.** Shishir Mehrotra's first customer team at Coda stopped using it and handed over about 30 fixes. Coda chose not to build them right away. He would aim for 40% of users saying they'd be very disappointed to lose the product. ### [Choosing the category for a new consumer brand]() Look for a need people already recognize, evidence that they act on it, and economics that work across the whole purchase cycle. - **Start from a need people already recognize.** Ivan Zhao would look for a category where people buy repeatedly, have a specific frustration with what exists and can describe a better product in their own words. Product, distribution and unit economics have to work together. - **Watch what customers do.** Reid Hoffman would ask what you know about customer behavior that established brands have missed. Then he would put an early version in customers' hands and watch them choose, open, use and buy it again. - **Find people doing extra work.** Emmett Shear points to gaming, about 1 or 2% of Justin.tv, where people were already streaming StarCraft with screen-capture software. He would talk to the people making those workarounds before choosing the product or the brand story. - **Weigh planning against purchase frequency.** Ben Silbermann would give strong discovery demand a chance to make up for rare purchases if people return again and again while planning, as Pinterest users did for weddings. Each sale then has to pay for discovery and trust. Shishir Mehrotra would compare categories across the full cycle of customer value and favor the one where he can name why existing brands left the opening. ## Participants - [Emmett Shear]() — CEO of Softmax. Founder of Justin.tv and former CEO of Twitch - [Ben Silbermann]() — Founder and Chairman of Pinterest. Founder of How We Feel - [Ivan Zhao]() — Founder of Notion - [Reid Hoffman]() — Co-founder of LinkedIn, Inflection AI and Manas AI. Partner at Greylock - [Shishir Mehrotra]() — CEO of Superhuman (formerly Grammarly). Co-founder of Coda and former Chief Product Officer of YouTube --- Copyright © 2026 Super Me Network, Inc. All rights reserved. | [Terms of Service](https://www.superme.ai/terms) | [Privacy Policy](https://www.superme.ai/privacy)