Why Indian Startups Are Unprofitable | KATA 6

Core idea: why many Indian startups stay unprofitable

The speaker argues that unprofitability often comes from misattribution in marketing/content—optimizing for what’s measurable today (e.g., last-click ROAS) instead of what truly drives conversion over time.

When teams chase only “easy-to-measure” performance signals, they fall into a rising acquisition-cost trap, limiting profitable growth. Eventually, they try to compensate by raising prices—but there are constraints.

---

Attribution as the “secret success key”

Framework / process proposed

• Attribution = identifying what was responsible for success/failure

Why it matters

• Converts “random luck” into repeatable strategy
• Prevents teams from crediting the wrong work (or company functions)
• Improves risk-reward decision-making

Practical implication

• In multi-touch journeys (especially content platforms like YouTube/Instagram), single-point conversion signals (e.g., last-click, one campaign link, coupon codes) are unreliable.

---

The measurement trap: ROAS-only (or overly performance-driven) marketing

What’s wrong with ROAS-heavy/performance-only attribution

• Brands often believe they know which asset converted, but buyer journeys are slow and multi-touch.
• This leads to:
  • Overspending to acquire incremental users based on flawed attribution
  • Cat-and-mouse ad costs rising because everyone uses the same tactics

Concrete mechanism described (ad cost spiral)

• ROAS degrades over time.
  • Example: spending ₹10 → returns ₹20; later ₹15/₹16 → returns still ₹20
• To “fix” it, companies raise prices:
  • ₹20 → ₹25, but ad costs keep rising → the business becomes structurally worse
• In India, raising prices is constrained by willingness-to-pay → eventual unprofitability

---

Case examples used to support the argument

Nike (strategic shift toward measurability)

• Nike allegedly lost $25B in market valuation (as presented in subtitles) after shifting heavily toward data-driven, measurable digital/sales activation.
• Claim: this reduced “aspirational/story” brand building and overemphasized “sell one unit” targeting (e.g., pixel-based retargeting), which is easier to measure but weaker for brand-building.

Zerodha (organic + low/no paid acquisition)

• Speaker claims Zerodha had no cost of acquisition for a long time and doesn’t run ads.
• Their margin is framed as roughly equivalent to what paid acquisition would have cost (i.e., organic reach substituting for CAC).

Physics Wallah (YouTube-driven CAC)

• Presented as “entirely YouTube-driven cost of acquisition,” where content supports customer acquisition rather than relying on paid acquisition costs.

Instagram link-tracking failure & “proxy metrics”

• Speaker argues that on Instagram/YouTube, link clicks aren’t normal user behavior, making linking/coupon tracking unreliable.
• Recommendation: measure proxies of intent (e.g., shares) rather than relying on views/likes alone.

Affiliate/referral attribution unfairness (example: Zerodha affiliates)

• Problem described:
  • A user clicks multiple affiliate videos across days, but only the last one converts/gets credited.
  • This misattributes value to the last-touch creator instead of the full persuasion journey.

---

Playbook: “proxy attribution” and qualitative attribution

1) Proxy attribution (use better outcome signals than views/clicks)

Examples of proxy metrics

• On Instagram:
  • Shares are suggested as a better proxy for intent than views/likes.
• Views can be cheap and misleading:
  • The speaker notes you can buy/engineer large views without sales impact.

Example metric set (as stated)

• Nvidia GTC video:
  • ~14,000 likes, ~7.6k shares (shares treated as the meaningful intent signal)
• Mumbai Tech Week video:
  • ~3.1k shares (still treated as meaningful, though lower)

2) Qualitative attribution (hard but “honest”)

Process

• Run qualitative research:
  • Call buyers/users who engaged/converted and ask why they purchased
  • Use in-house surveys and interviews rather than trusting only attribution dashboards

Goal

• Identify causal clusters rather than single-point “conversion causes”

---

Buyer journey truth (why “coupon code” and last-click logic fail)

Claim

For high-consideration purchases (phones, shoes), users don’t typically snap-buy after one influencer post/ad.

The speaker describes a multi-step journey:

• Watch reviews of multiple alternatives
• Compare options over weeks
• Final purchase is often a “gut decision” shaped by accumulated persuasion

Therefore:

• Coupon codes / single link attribution can severely undercount the assets that influenced consideration.

---

Organizational/leadership tactics tied to attribution

Correctly credit teams and individuals

• Attribution is also a talent management issue.
• In pods/teams (e.g., script/editing/motion graphics), multiple people contribute, and “who gets credit” may not match actual impact.
• The speaker argues:
  • The loudest contributors often get credit
  • Companies need systems to credit the right work and promote better talent

---

Product/engineering lesson (misattribution leads to wrong building)

Example

• In a past SaaS company, the team allegedly believed adding features would drive acquisition/acquirer value.

Critique

• The speaker notes India acquisitions typically value revenue, not custom tech features.

Lesson

• Don’t build “because it seems measurable/useful” if it doesn’t match the buyer’s true decision driver.

---

“Cluster” thinking instead of single-cause thinking

Framework referenced

• Lollapalooza effects (Charlie Munger):
  • Success often comes from a cluster of 4–5 things moving in the same direction

Emphasis

• Correlation ≠ causation
• Attribution should identify the cluster, not one isolated factor

---

Key metrics / KPIs mentioned (and how they’re used)

• ROAS (Return on Ad Spend) as a central (misused) metric
• Ad-spend vs returns unit economics (examples in rupee terms)
• Views / likes / shares for influencer/content proxy attribution
• Example of large-scale waste:
  • “300 people × ~₹10 lakhs” ≈ ₹30 crores wasted (life/education analogy for attribution cost)

Targets/timelines

• No explicit numeric growth targets or deadlines.
• The speaker frames effects as:
  • content impacts unfolding over months / long-term
  • buyer journeys lasting 30+ days for some categories

---

Actionable recommendations (implied playbook)

• Don’t optimize only for measurable last-touch performance
• Use proxy metrics (e.g., shares over views/likes; other intent signals)
• Run qualitative attribution:
  • interviews/calls + in-house surveys
• Build around repeatable formulas, not “random luck”
• Ensure teams:
  • credit the right drivers (not what’s easiest to measure)
  • avoid shortcuts that temporarily boost ROAS but harm long-run profitability
• Maintain brand-building/top-of-funnel investment even when performance metrics tempt teams to cut it

---

High-level “success formula” offered

Business success depends primarily on:

• Finding an untapped market
• Hiring good talent
• Executing well
• Sticking around long enough to reap benefits (don’t die)

---

Presenters / sources

• Presenter: Varun (name appears as “Varun” during the talk; no surname given in subtitles)
• Referenced third parties: Elon Musk; Nitin Kamath (podcast mentioned); Jason Lemkin or AppSumo CEO (unclear which, mentioned as a quote source); Charlie Munger; A16Z; MKBHD (buyer journey example); Netflix (commissioning logic); Sony/PlayStation (console exclusives example); Google/Facebook pixel systems; Zerodha; Physics Wallah; Nike; Dyson; Apple.