How should founders balance their own product vision against customer feedback in setting the roadmap?

Linda Lian · Apr 1, 2021

In customer discovery you should never ask customers for features — only for their problems — because inventing the solution is the company's job.

Customers should be asked about pain points and what gets in the way of what they want; the company must then innovate on their behalf, and the solution may be a product feature or something else entirely like sharing knowledge across organizations.

29:16 20VC: How To Deliver Feedback with Compassion and Clarity, WTF Are Values Really and How To Instill Them & Raising $50M+ From Index and Greylock but Remaining in Stealth for 12 Months with Linda Lian, Co-Founder & CEO @ CommonRoom

Jag Duggal · Mar 29, 2023

Product teams should not take dictation from customers; the job is to observe and interpret unmet needs

If we simply implemented what customers asked for, there'd be no need for the product middleman — you'd just send the engineer

7:19 20Product: Nubank's CPO on Why Product is 90% Science and 10% Art, Why Execution is Overrated and Strategic Clarity is Under-Appreciated, Why You Should Never Fall in Love With Your Own Ideas & Nubank's Biggest Product Challenges Scaling to 80M Users

Kevin Niparko · Sep 29, 2023

Product teams should always listen to users' problems but rarely to their proposed solutions

Scope: exception: when the user has built a working solution and will show it to you

42:18 20Product: Why Product Memes Are More Important Than a Product Roadmap, Why Writing is the Essential Skill for Product People, How AI Changes The Role of Product, Big Mistakes Founders Make When Hiring Product Teams with Kevin Niparko, VP Product @ Twilio

David Meyer · Nov 1, 2023

You should never listen to customers on how a feature should work, only on what they need to accomplish — and then check whether that need is common across the collective customer

The job to be done generalizes; the requested implementation does not

7:25 20Product: Why You Should Not Go Into Product Management, Why the CEO is Always the CPO, How to Build the Best Product Teams & Why You Should Hire People Who Aren't In Product Already with Databricks SVP Product, David Meyer

KR Sridhar · Jun 29, 2026

A leader must practise empathy across every part of the business — talking to shop floor workers and asking customers what pain they want solved rather than pitching product greatness — and that is what separates a good company from a great one

The things front-line employees and customers tell you when you actually listen are what make the difference; you then design the product accordingly

12:42 20VC: Leo Aschenbrenner's Largest Holding: Inside the $90BN Bloom Energy | Why Electricity, Not AI Models, Will Decide the Winners of the AI Race | Why We Are Not in an AI Capex Bubble | Energy Sovereignty and The Future of Power with KR Sridhar

David Lieb · Sep 21, 2022

Building the solution a customer literally asks for is the wrong way to act on feedback; you should absorb everything they say, their expressions and all the data, and use it to understand why they said it

The stated request may hide the real issue — they may not actually be confused, or the problem you're solving may not be big enough for them to bother understanding the screen

18:41 20 Product: The No 1 Metric You Need To Look at When Building Product | Why the Best in Product Have No Domain Experience | Why You Should Not Hire From Incumbents & The Difference Between Good vs Great PMs with David Lieb, Visiting Group Partner @ Y Comb

Hugo Barra · Nov 9, 2022

Listen to users' behaviors and ignore most of their opinions

44:59 20 Product: Hugo Barra on Lessons Building Hardware Products at Android, Xiaomi, Oculus, and Detect; Feature Kings vs. Budget Kings; 996 Work Culture in China; There Are No MVPs in Hardware; The 3.5-Hour Recruiting Interview

Kevin Niparko · Sep 29, 2023

Product teams should sit with their most advanced customers and study what they've built with the product, because those hacky workarounds give early signal of what will be big

At Segment, advanced customers had already built hacky workarounds to load data into a warehouse from an S3 bucket, which revealed the potential of what became Segment Warehouses

Scope: applies to the smartest/most advanced customers

40:14 20Product: Why Product Memes Are More Important Than a Product Roadmap, Why Writing is the Essential Skill for Product People, How AI Changes The Role of Product, Big Mistakes Founders Make When Hiring Product Teams with Kevin Niparko, VP Product @ Twilio

Cem Kansu · Jun 20, 2025

On-demand human tutoring fails as a language-learning product because users find speaking live with a human too intimidating

Observed users press the button, get connected, and hang up immediately; stumbling in front of a human on video is embarrassing

Scope: based on observed user behavior; even though users had been requesting the feature for years

65:39 20Product: How Duolingo Build Product 10x Faster with AI | Duolingo's Biggest Lessons on Paywalls, Push Notifications and In-App Purchases | Why Small Teams are the Future of Product | Why PMs Will Become Extinct with Cem Kansu, CPO @ Duolingo

Max Junestrand · Aug 15, 2025

You should not build what legal customers ask for — they ask for a faster horse — instead read their behaviour with your tools, because ~99% don't understand the technology at all

There is a huge information gap between builder and customer, so stated requests are a bad signal while observed behaviour is a good one

Scope: specific to selling technology into law

19:23 20VC: 15 Term Sheets in 7 Days and Choosing Benchmark | Harvey vs Legora: Who Wins Legal and How to Play When You Have $600M Less Funding | Are AI Models Plateauing Today | Building a 9-9-6 Culture From Stockholm with Max Junestrand

Alex Bouaziz · Feb 1, 2023

When scaling this fast everything breaks and it's purely a question of focus — his own failures were not planning support hiring early enough and not building sales ops sooner.

He and his co-founder were personally doing support late into the night across time zones because he hadn't planned hires in time, and the company paid the price of an imperfect, hard-to-use Salesforce.

19:20 20VC: From $57M in ARR to $297M in Just 12 Months; Why Speed of Execution is the Most Important Factor to Success, Hiring 2,000 People in 3 Years Remotely & Secondaries; Why, When and How Much To Take Out with Alex Bouaziz, Co-Founder & CEO @ Deel

Chris Degnan · Nov 10, 2023

The founder must own the product vision but must be genuinely open to customer feedback; founder arrogance that they know better than the customer means the company loses

Snowflake's culture of putting the customer first came from its founders, and sales' role was to help them see which features customers needed while the vision stayed theirs

14:02 20Sales: Five Lessons Scaling Snowflake to $1BN ARR, Why Customer Success is BS and Should Be Removed, Why All Sales Reps Should Do Eight Calls Per Week & Why You Should Hire a Head of Sales Sooner Than You Think with Chris Degnan, CRO @ Snowflake

Paul Erlanger · Jun 27, 2026

Founders should be epistemically modest about user feedback because users often don't know what's best for them; feedback should be filtered through your own instinct about the fundamental product experience and whether it fits the larger vision.

Some product decisions have large enough downside that they could kill the product, so they demand extra thoughtfulness before implementation.

Scope: applies most to changes with large potential outcomes

8:53 20VC: How We Got Fred Wilson, Benchmark and Index to Invest $94M | Why Robinhood's Strategy is Wrong | Why 1-1s are BS and What Every Founder Gets Wrong About Equity | Why Taste Beats AI But How AI Kills Org Charts with Paul Erlanger, CEO @ fomo

Sanjit Biswas · Dec 8, 2023

Use an 80/20 rule for customer requests: build it onto the platform if roughly 80% of customers would benefit and you hear it across multiple geographies and segments; partner for exotic or esoteric requests

Hearing the same request all over the place across markets is the signal it belongs on the platform

19:45 Build only what most customers ask across segments

Chris Degnan · Oct 10, 2025

Snowflake's engineering team was wrong to refuse to build a data science notebook and world-class data science experience, and that refusal gave Databricks room to grow

He was diagnosing where customers were 'getting sick' with Databricks — every company needed a data science notebook and the best customers said so — but engineering said it was stupid, and Databricks' current position proves it wasn't

28:27 Refusing a repeatedly requested core capability cedes the market to a competitor

Alex Bouaziz · Feb 1, 2023

Customer feedback should be triaged by type: direct complaints that the product is bad should be executed on no matter what, while directional 'where should we go next' input should only be used to validate an existing hypothesis across many conversations.

Direct product criticism is unambiguous, whereas directional input needs a hypothesis and a count of how many of a hundred people push you in that direction.

18:08 Informal direct messaging with customers beats any other feedback channel

Alex Bouaziz · Feb 1, 2023

Vision should be mostly predefined internally, while product iterations, execution detail and sequencing should be customer-led.

In HR many products are adjacent and could be sold to the user base, so the hard question customers answer is what to focus on and when.

18:41 Distinguish direct complaints acted on immediately from directional input requiring validation

Daniel Erickson · Jun 5, 2020

Open-ended survey feedback should be read differently by segment: themes from 'very disappointed' users reveal the strengths to double down on, while themes from 'somewhat disappointed' users reveal what to build to convert them into strong fits.

Doubling down on strengths shows customers you understand your product's value and maintains your lead against competitors, while addressing the fence-sitters' blockers moves them into the strong-fit segment.

14:29 Segment feedback by fit score to decide what to build

Raaz Herzberg · Dec 12, 2025

When a customer requests something outside your domain, that's a communication failure upstream; when the request is inside your domain, enough dialogue will surface a generic way to solve it

If you've communicated what you own and they've communicated what they need, an out-of-domain ask means the conversation shouldn't have gotten there; in-domain asks are real problems the customer cares about, so the right generic solution can be found together

32:17 In domain requests get generalized out of domain asks signal miscommunication