The numbers Brief Episode map Timeline Takeaways Charts Quotes Clips Glossary Go next Transcript

Episode 100 · Enterprise · 51 min

The moat is the engine, not the output

Public markets got Bloomberg and Capital IQ because 50,000 listed companies arrive pre-structured; private markets got Google search and a week of an analyst's life. Tracxn's claim is that a data business carries a deeper moat than software: the answer on screen sits on an engine tracking more than 700 million entities, refreshed globally by the hour. Nine years after a Java prototype written to solve her own problem as a VC, Neha Singh rang the bell, and calls that day zero rather than an exit.

Data & digitisation SaaS & enterprise Venture capital

Neha Singh

Co-founder & CEO, Tracxn · with Vishal Krishna

51:23

Said in this episode

▶ 16:51

700M+

Entities tracked on the back end

Every domain registered globally is picked up, so a domain registered yesterday is on the back end today. That is the raw funnel from which companies are curated.

▶ 7:02

$5,000 vs $100

The price test curation won

Two early versions were trialled at Stanford: a roughly $100 pure-technology product and a $5,000-a-month curated one, and the curated version found more takers.

▶ 19:34

~$10,000

Average realised pricing per customer

Low enough that buyers are comfortable purchasing remotely; realised pricing runs higher in the US, where more individuals within a company subscribe to a plan.

▶ 19:50

1.5 months

Sales cycle, first demo to close

Against the roughly nine-month B2B norm the host cites for that era, helped by a product that needs no integration and leads that arrive already having seen the data.

▶ 25:20

750

People behind the platform

Including a 100-member central technology team and a 90-member sector-focused analyst team doing final curation; every department also runs its own automation engineering team.

▶ 32:27

30 → 70,000

Shareholders, before and after listing

Roughly 30 investors pre-IPO became 30,000 on day one and more than 70,000 a year later, the responsibility Singh says changed how the company is run.

The brief

The argument in sixty seconds

Singh's claim is that a data business has a deeper moat than a software business, and Tracxn is the case she wants to argue. Public markets have Bloomberg, FactSet and Capital IQ because roughly 50,000 listed companies arrive pre-structured; private markets have millions of companies and almost no structure, so the answer a user sees, every generative-AI company mapped across the value chain, sits on an iceberg: an engine tracking more than 700 million entities that picks up a domain registered yesterday, and a 90-person sector-analyst team doing final curation inside a 750-strong org. The commercial lesson came early and counterintuitively. At Stanford in 2013 the founders tested a $100 self-serve product against a $5,000-a-month curated one, and the expensive, human-touched version won, because the VC and PE professionals buying it are paid too well to scrape the web themselves. Everything downstream follows: inbound content instead of cold calls, a one-and-a-half-month sales cycle at roughly $10,000 realised pricing, data produced in India and sold to the world, corporates who scan markets for vendors rather than acquisitions, and AI deployed behind the interface rather than as the chat box customers keep saying they don't want. The listing, she argues, is day zero rather than the exit: a twenty-year commitment with an exam every quarter and more than 70,000 shareholders now watching.

Worth your time if you are

Founders building data products rather than dashboards

VC and PE analysts still assembling sector maps by hand

Corporate innovation and M&A teams scouting emerging vendors

Founders weighing an Indian listing against an acquisition

Operators wiring AI into back-end workflows, not front ends

Episode map

Where the conversation travels

Every block is a chapter, coloured by what it's about. Click any of it to jump straight to that minute on YouTube.

Data & digitisationSaaS & enterpriseVenture capitalSales, GTM & growthAI & machine learningFounder journey

01Cold open: the map nobody had 0:00 Vishal opens on twenty years of using data with intuition rather than intuitive platforms, and Singh defines Tracxn as a private-market platform tracking startups globally for VC funds, PE funds and large corporates. 02A VC's own problem, coded in Java 1:43 IIT Bombay computer science, then BCG, then a VC seat where mapping every D2C brand meant an analyst losing a week to Google, so she brushed off dormant coding skills and wrote the first version of Tracxn herself, in Java. 03Bloomberg exists; private markets had Google 3:43 In 2013-14 the top angel investors in India were whoever your friends could name, and the founding bet was that private markets, now 10-15% of a typical LP allocation and up to 30% at a big university endowment, would need the platforms public markets already had. 04Sand Hill Road and the $5,000 answer 5:30 Tracxn was born in the US during her Stanford MBA, incubated in a no-equity summer programme on Sand Hill Road, where the founders tested a $100 pure-tech product against a $5,000-a-month curated one and found the expensive, curated version had more takers. 05Product-market fit, then an Indian incorporation 7:50 Paying customers in enterprise sectors signalled fit, and the founders deliberately incorporated in India rather than the US because a financial-data business would eventually suit Indian public markets, a choice that quietly set the IPO in motion on day one. 06Fifty thousand listed, millions private 10:17 Public markets offer 50,000 listed companies with structured data; private markets demand you first find the relevant million out of millions and then structure information that arrives unstructured. That is why the build needed heavy technology and a curation team from the start. 07The iceberg under the search result 13:18 Against VCs who pitched armies of small-town annotators, Singh argues data carries a higher moat than SaaS: you see the output, never the engine refreshing it hourly across languages and geographies, or the 700 million entities tracked so that yesterday's domain registration surfaces today. 08Inbound leads, remote demos, six-week close 17:35 Half the customer base is private-market investors and half corporates, none of them cold-callable, so content-led inbound feeds a short funnel of roughly $10,000 realised pricing, a one-and-a-half-month cycle and no integration required. Once inside sales proved out, the entire global sales team moved to India alongside data production. 09Corporates scan; VCs get deal flow 22:05 Corporates have narrow mandates and thin deal flow, so they use Tracxn to scan a sector before an M&A move, or increasingly to find new-age vendors and partners, from eKYC providers for a bank to drone-based crop-health companies for an agri firm. 10Ninety analysts behind the button 25:20 A 90-member sector-focused analyst team decides what belongs on the platform and where it maps, turning a week of manual collation into two hours, and the same coverage logic serves more than 70 Fortune-500-class corporations abroad, where realised pricing runs higher on multi-seat plans. 11Why a data company runs 750 people 27:00 Asked why a product company needs 750 staff, Singh points to financial data in 20-plus countries needing enterprise-grade curation that even public-market data vendors still perform manually, with a 100-member central tech team and a co-founder (an IIT Kanpur computer science graduate and former Accel investor) owning technology while she runs go-to-market. 12IPO is not an exit 29:07 A private company chooses between an M&A that frees you in two years and a listing that commits you for twenty; Tracxn turned cash-flow positive at the end of 2020, filed its DRHP in 2021 and listed in year nine into poor sentiment, taking a cap table of about 30 investors to 30,000 on day one and past 70,000 since. 13The first AGM question was GenAI 33:42 She expected quarterly numbers and got long-term technology questions instead. Her answer is that AI belongs behind the interface, classifying companies out of 700 million entities, sharpening comps and validating accuracy, because customers asked whether they would trade their working views for a chat box said no. 14Selling harder, filing deeper 40:00 After a decade spent building the platform, the next year's effort shifts to sales and marketing reach, deepening customers in geographies that have already grown, and adding regulatory filings (financials, cap tables and transactions in over 20 countries) for later-stage investors. 15Nothing in her head, everything on the calendar 41:43 The IPO reframed the job as a marathon rather than a sprint, producing a daily 6am fitness hour and a weekend calendar as full as a weekday, because the vacation you keep meaning to book for your parents takes an hour to arrange and five years to remember. 16BARC Colony to the bell 45:10 Raised in a Bhabha Atomic Research Centre colony where half the parents were scientists and her father a nuclear scientist, she was the family's first entrepreneur, with parents unconvinced until the angel cheque arrived, and she still advises founders to spend a couple of years inside a good organisation first.

Timeline

The arc, briefly

Ten years from a Sand Hill Road summer programme to a listed company, dated from the conversation.

2013Tracxn starts, in the US. The company is born during Singh's Stanford MBA and incubated that summer in a Lightspeed programme on Sand Hill Road that takes no equity. She had left a VC seat where mapping one sector cost an analyst a week of Google.

2013The price test. The founders run a roughly $100 pure-technology product against a $5,000-a-month curated one, and the curated version finds more takers. The VC and PE professionals buying it are paid too well to gather data themselves.

2013-14The gap they were betting on. Public markets had Bloomberg and Capital IQ across 50,000 listed companies with structured data; to name India's top angel investors, a journalist asked friends and searched Google.

2013-14Incorporated in India, not the US. Paying enterprise customers signalled fit, and the founders chose Indian registration because a cash-generating financial-data business would eventually suit Indian public markets. Asked when they first thought about an IPO, Singh says incorporation day.

End 2020Cash-flow positive. Indian public markets like profitable stories, so breakeven is the point at which the listing becomes a plan rather than an idea.

2021-22Bankers, DRHP, bell. Preparation starts the year after breakeven, the filing buys an approval valid for twelve months, and Tracxn lists in its ninth year into poor sentiment.

TodayQuarterly exams instead of an exit. A year and four earnings calls after listing, a cap table of about 30 investors has become more than 70,000 shareholders, and 750 people run the platform, 100 of them in central technology and 90 in sector analysis.

Takeaways

Ideas to carry out of this hour

A data business has a deeper moat than a SaaS business

The pitch rooms of 2013-14 were full of models that promised to beat Tracxn with cheap labour scraping and annotating the web, and Singh's counter is that the visible answer is the smallest part of the product. What you copy off the screen sits on an iceberg: an engine that keeps millions of records current every hour, across countries where the source documents are not in English. That refresh machinery, not the interface, is what a competitor would have to rebuild.

▶ 13:32 ▶ 14:35 ▶ 14:50

The expensive, curated product outsold the cheap, automated one

Tracxn's first real experiment was a price test: a roughly $100 pure-technology solution against a $5,000-a-month curated one. The curated version won, because the buyers were highly paid VC and PE professionals who needed the data judged before it arrived, not simply gathered. That single result set the company's shape: analysts alongside engineers, coverage decided by humans, and pricing that assumes the customer's time is worth more than the software.

▶ 6:46 ▶ 7:02 ▶ 7:17

Choosing where to incorporate is choosing your exit

The founders debated incorporating in the US versus India and picked India, reasoning that a cash-generating financial-data business would eventually appeal to Indian public markets. Having been investors themselves, they knew the option runs one way: register outside and the Indian listing route effectively closes. Asked when they first thought about an IPO, Singh's answer is incorporation day.

▶ 8:17 ▶ 8:31 ▶ 8:47

An IPO is day zero, not the finish line

A private company facing liquidity has two roads: an M&A that lets the founder leave in two years, or a listing that is a public commitment to the next twenty. Singh frames the bell as day one of a new company rather than an exit, with quarterly results functioning as trimester exams that keep the whole organisation focused. The scale change is concrete. A cap table of roughly 30 investors became 30,000 on listing day and more than 70,000 a year later, which is itself a new kind of responsibility.

▶ 29:35 ▶ 29:50 ▶ 32:27

Customers do not want a chat box in front of the data

Asked directly whether they would swap their existing interface for a conversational one, Tracxn's customers said no, because an investor wants working capital laid out in the format they already read. So the AI went behind the screen instead: classifying which of 700 million tracked entities belong in fintech or robotics, producing more comprehensive comparable-company sets, surfacing new companies in a sector, and validating accuracy across data sets.

▶ 38:27 ▶ 38:42 ▶ 39:12

Automation shows up as flat headcount, not fewer people

The clearest evidence of AI paying off is a ratio rather than a product feature: revenue rose over three years while team size stayed range-bound, because the people needed to capture financials in a given country or global transactions keeps falling. Singh's structural answer is to refuse a standalone AI department hunting for applications. Instead the 100-person central tech team owns the platform, and every department runs its own automation engineering team. Her claim is that only tech-first companies survive this transition.

▶ 37:06 ▶ 37:28 ▶ 37:44

At $10,000 a seat, the sale can be entirely remote

Neither VC partners nor corporate innovation heads can be cold-called, so Tracxn built a content funnel that catches someone searching for, say, semiconductor companies, and converts them over Zoom demos. Average realised pricing around $10,000 and a product that needs no integration mean buyers are comfortable purchasing remotely, compressing the sales cycle to about one and a half months against the nine-month norm for B2B at the time. Once inside sales proved out, the India team took over selling to Europe and APAC too.

▶ 18:20 ▶ 19:34 ▶ 21:21

After ten years of building product, the constraint is reach

Customers now tell Tracxn the data and the interface are good, which reframes the problem as distribution rather than capability. Hence the recent quarters loaded with sales and marketing initiatives, a push to deepen penetration in geographies that have already grown large, and an expansion into regulatory filings, financials, cap tables and transactions across 20-plus countries for later-stage investors who want compliance-grade depth.

▶ 40:04 ▶ 40:29 ▶ 40:50

The numbers, drawn

What the episode measures

Everything drawn below was said on air, with the timestamps attached and the caveats kept.

The price test curation won

$ per month

The two price points Tracxn tested during its Stanford-era incubation, as stated on air; the $5,000-a-month curated version had more takers among highly paid VC and PE professionals.▶ 7:02

How long the enterprise sale takes

months

The nine-month figure is the host's characterisation of B2B sales cycles at the time; Singh gives Tracxn's cycle as roughly one and a half months, plus or minus.▶ 19:50

The iceberg under one search result

what a competitor would actually have to rebuild

Singh's answer to the 2013 pitch rooms full of models promising to beat Tracxn with cheap annotators. The output is the small part; the refresh machinery under it runs across countries where the source documents are not in English.▶ 14:35

Where the automation actually sits

750 people, and the department that was never created

Her structural objection is to a central AI team hunting for applications. The tech team owns the platform, everyone else automates their own work, and the people needed to capture a country's financials keeps falling.▶ 37:28

Worth keeping

Lines that stay

In data the moat is probably higher than in a software business. You see the output — but what goes behind it is a whole iceberg: the engine that keeps it updated every hour, globally, across so many sources.

▶ 14:35

An IPO is not an exit. An IPO is like starting off day one — because here you are committing that you are with the company for the next twenty years.

▶ 29:35

After getting listed, you are having exams every quarter. It just keeps you on your toes, it keeps the entire organisation focused.

▶ 29:50

When we ask customers whether they would give away their current interface for a chat-like interface, the answer is typically no.

▶ 38:42

Nothing is in my head — it is on my calendar.

▶ 44:03

Clips that travel

Short on time? Start here

Founders pricing a data or research product

The $100 product that lost to the $5,000 one

The price test that decided Tracxn's shape, and why buyers who are paid well want data judged rather than merely gathered.

6:46 → 9:16 · 2 min ▶ Watch clip

Investors sizing defensibility in data businesses

The iceberg under the search result

The moat argument in full: annotator-army competitors, the hourly refresh engine, and 700 million tracked entities.

13:18 → 17:35 · 4 min ▶ Watch clip

B2B founders building an inside-sales motion

A six-week enterprise sale, entirely remote

Content-led inbound, $10,000 realised pricing, no integration, and the moment the whole global sales team moved to India.

17:35 → 22:05 · 4 min ▶ Watch clip

Founders weighing an Indian listing against an acquisition

IPO is not an exit, it is day zero

The M&A-versus-IPO fork, cash-flow positivity as the trigger, and going from 30 investors to 70,000.

29:07 → 33:42 · 5 min ▶ Watch clip

Operators wiring AI into back-end workflows

AI behind the interface, not in front of it

Why customers reject a chat box, where automation actually pays, and the case against a standalone AI department.

35:25 → 40:00 · 5 min ▶ Watch clip

Glossary

The jargon, unpacked

Private market data platform

The equivalent of Bloomberg or Capital IQ for unlisted companies: discovery, sector maps, funding, financials and comparables for businesses that publish almost nothing by law.

Realised pricing

The average revenue actually collected per customer across plans and seats. It is around $10,000 for Tracxn, and higher in the US where more individuals in a company subscribe.

Comps

Comparable companies used to benchmark a business you are evaluating; producing better and more comprehensive comps is one of the places Tracxn puts AI to work.

Deal flow

The stream of investable opportunities that reaches an investor unprompted. VCs have plenty, corporates have far less, which is why corporates must actively scan a market before an acquisition.

Limited partner (LP)

The institution whose capital a fund invests. Rising LP allocations to private markets, 10-15% typically and up to about 30% at a big university endowment, are what make Tracxn's category grow.

DRHP

The draft red herring prospectus filed with the regulator ahead of an Indian IPO; once approved it stays valid for a year, which is the window Tracxn listed inside.

Inside sales

Selling remotely by call and video demo rather than putting feet on the street. This is the model Tracxn discovered worked, which let one India-based team cover Europe and APAC.

Connections

If this resonated, go here next

EP 124 · Enterprise

Full transcript

The whole conversation, searchable

Auto-generated captions, lightly cleaned. Click a timestamp to open that moment on YouTube.

The argument in sixty seconds

Worth your time if you are

Where the conversation travels

The arc, briefly

Ideas to carry out of this hour

A data business has a deeper moat than a SaaS business

The expensive, curated product outsold the cheap, automated one

Choosing where to incorporate is choosing your exit

An IPO is day zero, not the finish line

Customers do not want a chat box in front of the data

Automation shows up as flat headcount, not fewer people

At $10,000 a seat, the sale can be entirely remote

After ten years of building product, the constraint is reach

What the episode measures

The price test curation won

How long the enterprise sale takes

The iceberg under one search result

Where the automation actually sits

Lines that stay

Short on time? Start here

The $100 product that lost to the $5,000 one

The iceberg under the search result

A six-week enterprise sale, entirely remote

IPO is not an exit, it is day zero

AI behind the interface, not in front of it

The jargon, unpacked

If this resonated, go here next

Cost is the part you can control

No one downloads a second app

The weakest link has a login

The whole conversation, searchable