India internet: The opex problem
Disclaimers:
Views expressed here are my personal reflections and not indicative of the views of my current or previous employers.
No part of this post maybe reproduced or quoted without explicit permission.
In my previous two pieces on internet companies in India, I had covered how subsidizing two sided marketplaces makes sense only in some scenarios and in how investors grossly overestimated market sizes in 2015. The main question I got from entrepreneurs on my post on market size is on how it affects them. It seems to really an investor/valuation problem right? Wrong.
Recently, I came across a post from Anand Lunia on how Indian tech businesses have bloated salaries. This is a topic that I have also been thinking about in the last few months based on data I am seeing across all the companies we have evaluated. The opex problem seems to be beyond just salaries. Here are some benchmarks on similar (or sometimes same) business models in US and India.
Revenue per unique visitor (RUV): US companies do 10–15x revenues for the same amount of traffic.
Customer acquisition cost (CAC): US businesses have ~4x the CAC of Indian businesses.
RUV/CAC: US companies have 2.5–3.75x RUV/CAC compared to Indian companies.
The first opex problem jumps out here. Unit economics (even ignoring poorer gross margins in India) are significantly worse in India. While there is some scope for improving RUV as the market matures, it is unlikely to move up significantly as a) new users being added have lesser disposable income b) GDP per capita (after purchasing power parity adjustments) in US is 9x of India. The problem lies in significantly higher marketing costs. Competition in marketing is generally higher in the US than in India (given existing incumbents). So, the higher marketing costs in India are simply due to one thing- too much funding relative to India’s market size for internet businesses.
So, how much overfunding exists in the market? Tough to get to an answer but here is one approach. $9B went into Indian startups in 2015 compared to $58.8B in the US in 2015- a multiple of 6.5x. We saw above that using RUV, GDP per capita or other forms of validating, US market sizes are 9–10x India today. The difference between 6.5x and 9–10x is one approximate indicator of how much overfunding exists. It is also important to note that a significant amount of US VC funding went into companies targeting global markets(think Uber, Airbnb, Snapchat). So, the actual extent of overfunding is likely higher.
The other place where incorrect market size estimations and overfunding affects companies is where Anand Lunia pointed out- in salaries. When companies pitch and investors underwrite a larger market than what exists, the entrepreneurs overestimate revenue growth while budgeting and recruit according to these revenue projections. Unlike revenues, the costs are front loaded (since you would hire in advance to hit the revenues) and harder to get rid of. So, on average you have companies missing revenue targets by 40–60% while improving on salary costs by only 10–20%. Even this 10–20% requires midway corrections- firing employees, reducing salaries, organizational changes which are incredibly hard to pull off. The abundance of funding has also pushed up salaries significantly leading to a double whammy in salary expenses. Salary cost is one of the big areas where the overfunding in 2015 (based on aggressive market size numbers and revenue projections) has affected the Indian tech ecosystem and it is going to take a while to clean this up.
The other reason for excessive headcount (hat tip to Sesh, founder of Zettata who has built engineering teams in both countries) is in India’s hierarchical salary structure. In the US, the delta between a junior engineer and a VP of Engineering in cash salaries is much lower than in India (curve with declining slope- plot x^(1/3) to get a sense). In India, it is the reverse where the senior people get paid close to US salaries while the junior engineers get paid significantly lower salaries than the US (curve with increasing slope similar to IT services- plot x³ to get a sense). This has a number of ramifications including multiple layers of middle management . A couple of years ago, I attempted to build the same mobile web product in US and India. In India, the ask was for 1 back end engineer, 1 front end engineer, 1 UI designer, 1 UX designer and 1 product manager to manage the whole process. In the US, it took 1 engineer to do it. This is probably an extreme example but shows the fully loaded cost for product development in India is not as low as what people conceive it to be (not just the ratio of engineer salaries).
The combination of excessive marketing and high salary costs (relative to revenues) is making companies less capital efficient than they should. To top this, when we look at the same businesses in both countries, the US business typically has 10x more revenue but the valuation differential is only 2x and this skews the risk/reward.
