Hightouch’s Kashish Gupta on Founder Mode, Customer Context, and Founder Risk

概览

This episode features Kashish Gupta of Hightouch, a YC Summer 2019 company that evolved from a travel startup into a data and AI platform for marketers. Hightouch helps companies use customer data for advertising and lifecycle marketing, and now builds AI agents for more personalized campaign creation.

A major thread is customer-driven strategy. Gupta says Hightouch ignored general advice to start with small businesses because customer feedback and product architecture pointed clearly toward enterprise customers.

The conversation then turns to “Founder Mode.” Gupta defines it less as micromanagement and more as taking the risks and making the hard decisions that only founders can make, especially when data is incomplete but customer and market context is strong.

分段落总结

[00:00] Hightouch’s Product and Customers

[事实] Hightouch is described as a data and AI platform for marketers.

[事实] The company helps marketers access customer data, run advertising and lifecycle campaigns, and deploy AI agents to build more personalized campaigns.

[事实] Gupta names PetSmart, Spotify, and Men’s Wearhouse as examples of large B2C brands Hightouch works with.

[事实] Hightouch focused on enterprise customers early, even though YC and others advised selling to smaller companies first.

[02:18] Customer Feedback Over Outside Advice

[事实] Gupta says Hightouch’s architecture was better suited to enterprises than to small businesses.

[事实] He says the company “basically only” listens to customers, regardless of who gives other advice.

[事实] Inside Hightouch, co-founders challenge ideas by asking whether a customer actually said them.

[推测] The company treats customer evidence as a guardrail against founder opinions becoming strategy too quickly.

[02:48] From Travel Startup to Data and Marketing Platform

[事实] Hightouch originally entered YC as a travel business that used natural language processing to help users book travel.

[事实] COVID made the travel business unattractive, so the company began testing new business ideas.

[事实] Gupta says potential customers would often say yes to an idea, but would reveal their true priorities when asked broader questions.

[事实] Those conversations eventually led to the idea of helping companies use their data in sales and marketing.

[03:42] The Enterprise Data Gap

[事实] Gupta says enterprises had a large gap between having data and being able to use it in production sales and marketing workflows.

[事实] He names Snowflake and Databricks as examples of data warehouses enterprises were using as their databases.

[事实] Two Hightouch co-founders had been early engineers at Segment, and Segment’s approach influenced the team’s understanding of the market.

[事实] Gupta says the Segment-style model broke in enterprise because enterprise customers wanted to bring their own database.

[05:04] Why Hightouch Does Not Store Customer Data

[事实] Gupta says the difficulty in connecting enterprise data to marketing comes from data volume, governance, and the lack of tooling between databases and marketing platforms.

[事实] He says marketing platforms did not want that gap filled because they wanted to own the data.

[事实] Hightouch chose to make SaaS tools reflect the customer’s own database rather than storing the customer’s data itself.

[事实] Gupta says Hightouch still does not store any data and has no database.

[05:49] Company Scale and Product Reinvention

[事实] Gupta says Hightouch has about 250 employees, roughly 2,000 to 3,000 customers, and works with about 25 Fortune 500 companies.

[事实] He says the company has redefined itself about every two years.

[事实] Hightouch moved from a data connections product to a marketer-friendly UI, then to an AI decisioning product that uses reinforcement learning.

[事实] The AI decisioning product chooses the best marketing message for each customer, with the goal of reducing message volume while improving relevance and engagement.

[07:19] Preparing the Team for Founder Mode

[事实] Gupta says he has explained Founder Mode to Hightouch’s executive team and sales leaders so they know what to expect when he or his co-founders practice it.

[事实] He believes telling people in advance prevents them from being surprised or thinking the founder is angry or acting out of character.

[事实] He says asking for permission in advance gives founders more credit when they later intervene directly.

[推测] For Gupta, Founder Mode requires social preparation as well as decisive action.

[08:12] The CEO’s Job Is Company Success

[事实] Gupta says he was taught that the CEO’s job is to make the company successful, not to make everyone around them happy.

[事实] He says he struggled early with trying to listen to everyone, build consensus, and lead democratically.

[事实] He says he became a better co-founder and co-CEO around 2024 when he “stopped negotiating.”

[事实] His definition of Founder Mode is that the founder’s job is to deliver results and make the company successful.

[10:13] Founder Context Versus Executive Experience

[事实] Gupta says founders often hire executives who are more experienced in a domain than they are.

[事实] He says those executives may be better in their domain, but the CEO has broader context from the market, customers, and the rest of the company.

[事实] He says the CEO will usually have more context than everyone else in the business.

[推测] Gupta is arguing that founder conviction can be valid even when it conflicts with a more experienced executive’s view, if it comes from broader customer and market context.

[11:01] Writing as a Tool for Conviction and Disclosure

[事实] Gupta says he writes almost every hour of every day and writes down almost every thought he has.

[事实] He tries to share everything that remains true by the end of the week.

[事实] He says if a founder knows something but has not told executives or leadership, they are failing to do their job or hiding information.

[事实] He says good people often sense the same issue and are relieved when it is shared openly.

[12:32] The Founder as the Person Who Can Take Risk

[事实] Gupta credits leadership coach David Clements with teaching him that the founder is often the only person who can afford to be wrong.

[事实] He says most employees are risk-averse because they can be fired.

[事实] He argues that if a founder is right 100% of the time, they have not taken enough risks.

[事实] He describes risk-taking as something founders can uniquely do because others will often ask permission, hedge, or take only partial risk.

[13:31] Doubling the Sales Team Before the Data Proved It

[事实] Gupta says that in the middle of the prior year, many people did not believe Hightouch would still hit its revenue plan.

[事实] He decided to double the sales team headcount because he believed the market was showing more demand than historical metrics captured.

[事实] He learned this by managing the sales team directly while learning how to hire a chief revenue officer.

[事实] He says ground-level reps and managers had strong qualitative optimism that was not yet visible in the data.

[推测] This example illustrates Gupta’s view that founder risk can come from direct market exposure rather than formal metrics.

[15:37] Founder Mode Among Co-Founders

[事实] Gupta says coordinating among equals can be the hardest version of Founder Mode.

[事实] He says aligned co-founders can “move mountains.”

[事实] Earlier in the year, Hightouch was working on an AI product based mainly on reinforcement learning rather than LLMs.

[事实] Gupta says CMOs had been scared of LLMs and averse to content generation during H2 of the prior year and most of H1 of the current year.

[16:35] Choosing a Larger AI Opportunity

[事实] Gupta says Hightouch had about four main products and around 50 engineers, making the company understaffed for its product surface area.

[事实] He told his co-founders that the current AI opportunity might not be the largest one available.

[事实] He believed there was a bigger opportunity that no one had cracked, and that pursuing it could create a longer-lasting business.

[事实] He framed the decision by saying the company was now strong enough that it would still double that year even if the three founders stopped doing their current jobs.

[17:47] Reimagining Marketing With Customers

[事实] Hightouch began asking customers how they would build a marketing team from scratch using today’s technologies.

[事实] Gupta says customers responded that they would not do much of what they currently do, including writing briefs and relying on trial and error.

[事实] Customers said they would make more data-informed decisions and use LLMs to analyze campaigns and data with them.

[事实] Hightouch began developing the new product with customers, offering it for free in exchange for collaboration.

[18:29] Co-Founder Alignment Took Time

[事实] Gupta says it took two to three months for the co-founders to convince one another.

[事实] The challenge was not choosing yes or no on a clear proposal, because none of them initially knew exactly what to do.

[事实] The shared belief was that there was a technical shift in the market and that Hightouch had unusual access to marketers.

[推测] The episode presents founder alignment as an iterative process of shared discovery, not just one founder persuading the others.

播客点评/总结

[推测] The episode’s strongest value is its concrete treatment of Founder Mode. Rather than staying abstract, Gupta gives specific examples: doubling the sales team without complete data, directly managing sales to understand demand, and spending months aligning co-founders around a larger AI opportunity.

[推测] It is especially useful for enterprise SaaS founders, because much of the discussion centers on customer context, enterprise product-market fit, sales capacity, and when to trust qualitative market signals before metrics fully catch up.

[推测] The main limitation is that the conversation is brief and mostly reflects Gupta’s perspective. It does not deeply examine counterexamples where founder conviction might be wrong, or how Hightouch handles the organizational cost of risky decisions when they fail.

[推测] The episode is best suited for founders and startup leaders who are scaling teams, hiring senior executives, or deciding when to intervene directly instead of relying on consensus.