30 Dec 2017

To start up or not to start up

An well respected entrepreneur friend who's doing very well and has huge potential ahead of him asked me yesterday if I am considering starting my own sometime.

We got into a thought provoking discussion around why one should do so and how each of us was thinking about it. We also exchanged philosophical ideas that motivate each of us. I decided to write this to gather a balanced set of views from others that may help people who find themselves at such cross-roads.

For me an important question to ask is what really should be the motivation for someone to start their own. My experience has taught me that you should think about where you can have the most impact on the world. It can be either economic or social depending upon your interest. That doesn't matter because you can be an employee or an owner in either of these spheres.

Your impact in a corporate job depends upon the kind of role you're in and the kind of company you're in. If you're in a limited role in a low-moderate growth company you may have a higher chance of increasing your impact by joining or setting up a start up. The importance of the transformational nature of your start up idea is lesser in this situation. And if you're in a broader role or in a good trajectory in a fast growth company then you may have higher impact in the span of your career in a corporate role than through your own business. Unless you work on something truly transformational, by which I mean what the likes of Facebook, Apple, Amazon, Uber did. They built platforms that were truly scalable and capable of providing opportunities to millions of more people and businesses.

I think the first situation is an easier decision. The second situation tests your judgement. You could say that, at the time of starting, you'll not know the true transformative potential of your idea, e.g., a lot of people didn't realize the potential of the idea of Facebook in its early days. But it's also easy to fail to question your judgement that what you're working on truly has unidentified potential.

Start up landscape is a mix of good and not-so-good in everything - ideas, business models, people. While some are fixing an important problem, others are getting influenced by the trend.

I am not perfect and I don't always find it easy to follow what deep down I believe is the right thing. But for this one my answer is focus on impact.

9 Nov 2016

The informal world

I was a democratic supporter because I thought Donald doesn't have the temperament to be a world leader. Yet Donald won the election by a good margin and swung many dicy states in his favour. What struck the chord with the public at large? Perhaps a part of it was the hope of Donald changing the trajectory of the financial prospects of a common American whereas with Hillary they could hope only the continuation of the current trajectory. But perhaps there was another subtle part at play.

Why did Donald's volatile temperament and inflammable words not swing the public opinion against him? The answer could be in the informal world we live in today. Remember when we were kids, our school teacher taught us extensive templates in which to write news reports, headlines, applications, letters, telegrams, etc.! But the communication media of today emphasise short, to-the-point, and direct language even at the expense of grammar and with complete disregard to whatever we were taught about syntax. People today have little time and energy for formality, for keeping two sides of their personality - a professional and a personal one. That's why you see people cheering to Barack Obama going to The Tonight Show and being himself. Deep within, who doesn't have thoughts and ideas that would make public scoff at them yet only few have the guts to come out in the open and put these in words. Donald did exactly that. He portrayed his true self. He didn't mince any words. He said what came to his mind. He said what he thought.

And this didn't happen suddenly. Barack Obama had already given a bridge from a typical formal behaviour to this completely informal behaviour. He was somewhere in between, folding sleeves while wearing a tie, appearing on prime time shows, recording an appearance with Jerry Seinfeld, orchestrating light moments with world leaders, etc. What Donald did was taking it further afar. He was completely informal. He didn't behave like a typical socially accepted 'leader'. He became one of them. Perhaps that's what stuck a chord with the public. What do you think?

14 Sep 2016

The world of driverless cars

When driverless technology starts maturing and is proven to be safe, the cars will not require a manual override. So steering wheel will be taken out. The car will essentially have a navigation screen in which you fill the destination coordinates, choose the path, and press 'Go'.

Urged by regulators and governments, car companies will leverage this technology to reduce pollution, reduce energy consumption, and reduce road congestion. The cars will no longer look like the present day integrated vehicles with two seats in front and typically at least two seats in back (except in few high end luxury cars). The idea will be that if one person is travelling, why does he need to carry along an empty compartment! So the design of future will be modular. You would have a driving unit which will have one seat and will attach as many additional passenger units as required. Various flexible layouts of modular units will be possible depending upon how many units are travelling. So if only two people are travelling you can either connect the units side by side or front to back. If three people are travelling, you can either connect the units linearly back to back or like a current three wheeler with one unit in the front and two in the back. The software will automatically detect the layout and align the transmission and coordination between wheels of various units accordingly. The most visible impact of the modular design will be that it will significantly reduce traffic congestion.

Cars will be electric and will run on batteries. Batteries will be chargeable by either electric or solar power. You will not need a new extensive infrastructure of charging stations to support your hybrid driverless car. Batteries will be light and detachable. You will charge them at home by plugging into one of the sockets. There will also be a small inbuilt battery to store solar power in the interest of further fuel economy, energy conservation and environmental protection. The cars will also have an emergency mode of running directly on solar power though with significantly reduced performance. You will usually have spare battery banks like you have for mobile phones today. They will be working on various futuristic technologies like controlled fission of atoms and using water or air as a fuel. Perhaps you will see some trial runs of these technologies in your lifespan.

Driverless driving will improve inter-vehicle coordination on the road substantially so accidents will be rare and car bodies will mostly be scratch free. Reduced risk of damage will encourage higher borrowing of passenger units or even driver units from your neighbour, friend, or relative. If your friend lives some distance away from your home the driving unit can potentially autonomously go and return the passenger unit to the owner and then come back home itself. The cars will have embedded security technology to prevent theft when the car is travelling without you.

Insurance premiums will reduce significantly because of reduced risk of vehicular damage. To compensate the per unit realisation (i.e., premiums per policy), insurance companies will count on higher volumes (units sold) as cheaper technology becomes accessible to lower income groups also either through private ownership or through taxi services.

Various fleet management companies will operate driverless taxies and use a fleet management software commercially available from leading software companies. This software will act as an interface between various driving units and will help them talk to each other. If you booked a pick up for two people then the nearby driving unit accepting your request will either have the second unit on itself or will coordinate with another driving unit nearby which has a spare passenger unit. If no other nearby unit has a spare passenger unit then this driving unit will request additional passenger unit from the closest stand. Another driving unit will directly come to deliver it to your place while your booked driving unit comes to you from another direction around the same time. The time taken for such coordination will reflect in your pre-boarding waiting period for which you will of course not be charged. In case the driving unit that accepts your request has an extra passenger unit which is not required for your trip, it can either hand it over to another passing driving unit which may need it or leave it at the closest stand on the way either before or after you board.

For those who enjoy driving for pleasure, expensive convertibles will be available from a few car companies. They will look more like present day's integrated vehicles with multiple seats but will come with advanced automation and networking features. They will have both manual and driverless modes of operation. However, if you flout traffic rules intentionally or unintentionally, the traffic police's GPS driven network monitoring software will detect you and take appropriate action. You will not only be penalised for disturbing traffic but also be sued by insurance companies for any damage to other vehicles.

The changes will be easier for developed countries to adopt. However, developing countries will need to completely overhaul their traffic rules education and enforcement system to enable the driverless cars to drive without running over careless pedestrians or ramming into manual drivers with no knowledge of traffic rules. This will take time. For a while the divide between developed and developing countries will increase.

13 Sep 2016

What's with this personal data security!

A few days ago, there was this huge outcry about Whatsapp asking you to share your phone number with its parent, Facebook. A lot of informed people not only opted out of it but took it as their duty to educate the general public about the same. What right does a business have to access our personal information, even if for a service!

It's astonishing how valuable Whatsapp has been to our lives. It's a simple piece of software to which even our illiterate parents/grand-parents have adopted naturally. It has helped large families, alumni groups, teams, etc. keep connected with each other every minute of the day. It has filled a lot of available time in the everyday life of our retired parents with jokes, funny videos, motivational messages, and well with its fair share of hoax. It has united classmates from the previous generation. In ways unique to each individual context, it has added a lot of value to our lives. It has almost become an indispensable part of our lives. And it has done all this for free.

How do we price such an added value to our lives? How much are we willing to pay for this value? Perhaps nothing if you ask millions of people using it. Not until the last such service goes out of business not being able to pay for itself and we're left isolated, unconnected, undiscoverable. And it takes time after any such existing service goes out of business and before new one gains a critical mass to offer the same level of connectivity. If this happens 2-3 times, people will get tired of switching and perhaps the world will not be as connected much before the last service dies.

So here is a company asking for no money but just for you to share your phone number with its parent platform. A number which you've perhaps already put on your profile page on its parent platform, whether visible or invisible to the public. A number which you've perhaps shared with it so that it can allow you to reset your password using your phone number in case you forget it. A number which you happily give to every retail chain on the street just to earn 1% purchase value worth of loyalty points. Let's face it, you're anyways going to see ads on your Facebook wall. What will happen if this company is able to get a better picture of your preferences by linking Whatsapp? You'll get more relevant ads. It may actually end up saving you some time, improving your experience with ads, or adding to your gratification!

Cute little babies!

I went to a tea lounge the other day in a market. After a few minutes, a small family came and sat on the adjacent table. They had a very little cute baby girl with them. Her mother put her up on the table beside me and she kept looking at me after that without much of a comprehension. It was very cute. But I couldn't help but think, however nonsensically, that...

...if babies were to be compared to a computer at that point of time, they would be a state-of-the-art hardware built to very high specifications but with a primitive operating system, like an MS DOS on a 2016 Dell machine!

Well, on second thoughts there is an important patch of code to this version of MS DOS, it supports machine learning!

Sorry folks, random meanderings of a wasted mind!

9 Sep 2016

Economics of e-commerce

E-commerce is an exciting new industry in India. A large number of horizontal and vertical players are battling it out to win. Everybody is bleeding. Everybody is experimenting with new ways of winning over customers' loyalty. Leadership is changing hands every few months. Employees aren't sure who to bet their careers on. Investors are hedging their bets by putting money in multiple competing players and watching with nervousness as newer players with deeper pockets keep entering and stakes keep getting higher. Leadership of traditional industries is confused whether to jump into it to protect their market share at high cost or wait for clarity at the risk of being too late. What happens eventually is sure to serve as an extremely important learning experience for businesses and the current generation of young leaders on how to do business, and not just in the digital world but also in any new worlds of the future.

But the fundamentals of doing businesses still remain. Business exists to generate a higher RoI than the opportunity cost of capital. If a business cannot generate a short term RoI higher than the interest of, say, a fixed deposit, yours truly as a retail investor would rather keep his money in a FD rather than buy the stock of that business. But if yours truly however invested for a living then he would perhaps be willing take slightly longer term bets and say that the business should give him a higher 8-10 year RoI than the opportunity cost of capital. He would perhaps not try to look beyond 10 years because the world these days changes so fast that the assumptions made in earlier years may become invalid by the 10th year.

But to make an informed call over an 8-10 year period, he would try to model some assumptions of how the business would do on some fundamental drivers. This will help him set clear targets for the business on these assumptions and from time to time take any hard calls required to get the business back on track on these fundamentals. If the business still doesn't get its act together then it will help him to pack his bags and exit while the losses are still low. But if yours truly was getting paid a sure x% of the funds under management every year, he will need to summon all his ethics not to be creative with these assumptions and not put money where the assumptions for a reasonable RoI are unrealistic. [Blame it on his principles but yours truly will also not hand over his dough to anyone who is starting up to be sold or make money. He would rather give his dough to people who have a real purpose and a set of guiding values for the business].

Yours truly deconstructed the economics of e-commerce to understand the underlying components. As an outsider if you want to roughly assess the likely success of a business, you could do the following steps:

  1. Try to find out the annual losses so far.
  2. Estimate the future annual cash flows based on your assumptions of how the business will do on each of the fundamental components as explained later in this post
  3. Use your opportunity cost of capital as the discount rate and calculate the current value of all cash flows
  4. Try to see how many years it will take for the business to give a positive NPV. Do you think the assumptions can be considered valid for these many years, given the pace of change? Is the NPV positive in your reasonable investment horizon?
To try to understand the fundamental components, let's deconstruct the economics of the e-commerce, i.e., the equation of annual profitability:

Profit = Contribution margin - Marketing expenses - Fixed costs
          = [Unique visitors X Conversion rate (Orders/UV) X Average order value X Unit economics (Gross margin per order)] - Marketing expenses - Fixed costs
          = [Unique visitors X Repeat rate (Visits/unique-visitor) X Conversion rate (Orders/Visit) X Average order value X Unit economics (Gross margin per order)] - Marketing expenses - Fixed costs

As an investor, you want the above equation to return a positive value as soon as possible. For that you could set operational targets on each of the components for the management of the company and do periodic review to identify where the management needs help or where you need to take any extraordinary steps.

Here is a look at each of the components in detail:

Unique visitors:
E-commerce business is usually characterised by heavy network effects, i.e., the more unique visitors you have, the more sellers want to sell on your platform, and the more sellers you have, the more is the likelihood of visitors converting (i.e., doing a transaction on the platform). So a very important thing in building a good e-commerce platform is acquiring more and more unique visitors and increasing the value of your platform. Another way of thinking about this is that because of low barriers to entry, e-commerce is a low margin business so a high volume is required to make money. Hence, more the number of transacting users the better it is for the platform profitability. That's why businesses spend so much in customer acquisition early on even at the risk of sustainability. 

It's because of this that e-commerce is expected to constitute almost 25% of the TV spending and 15% of the total digital spending in India in 2016. As a result of this competition to acquire the same customer, the cost of acquisition of customers has gone through the roof. Both digital and offline are so expensive that no sane assumptions around drivers of customer lifetime value (LTV) seem to justify the cost of customer acquisition (CAC) for e-commerce platforms. It takes a lot of creative accounting on the part of start-ups to convince the investors of the sustainability of their customer acquisition economics and business sustainability. Why are platforms spending so much then? Because if they don't do so they're anyway doomed to die so they might as well go all in! Everybody hopes eventually they will be able to build a loyal customer base that they can then mine.

A good customer acquisition strategy will segment the target customers into various groups each of which can be reached in its unique ways based on its behaviour, geography, etc. A very rough example is that digital channels will typically give access to urban, highly educated, mid to higher income group individuals whereas mass media (TV/Radio/Newspaper/Bill boards) will also give access to broader segments, e.g., tier -2/3 towns, low to moderately educated, and lower income group individuals.

In India the fight for urban customers in the top 12 odd cities is intense because they are the early adopters of e-commerce and also have higher purchasing power. However, the next phase of growth is expected to come from high volumes of lower income customers in tier 2/3 towns. That's where non-digital channels will become more important.

And when you reach a potential customer, the next important thing is what story you tell. That's why one needs to identify what they stand for and what differentiates them so that the customer will remember them. In Seth Godin's words, this is where business needs to create itself a purple cow that the customer will remember.

To assess company's performance on this dimension, an investor will need to keep a tab on the market share (Daily or Monthly active visitors on the platform as a percentage of the total Daily or Monthly active visitors on all platforms or as a percentage of the total internet users or as a percentage of Facebook's daily/monthly active users for a proxy of internet users). There is no right target for a market share. Whatever makes your overall equation profitable becomes the lowest threshold. Sustaining that market share will impact the other components of your economics which may lead you to revise your market share target. This will act as a continuous feedback loop.

Repeat rates:
Once you have acquired the unique visitors, you want to make them visit your platform multiple times, i.e., have high repeat rates, so that the life time value of your customers increases. You could do so either by push or pull.

Push typically works through digital marketing, e.g., contextual advertising, in-market advertising, re-targeting, emails, notifications, etc. The first three involve additional financial cost every time you bring the same customer back whereas the latter two have a reputational cost involved because if you overuse them you could be seen as a spammer. Companies also dole out attractive offers to make the customers repeat. In early days, this makes the economics unfavourable but businesses do it nonetheless in the hope of building a loyal base. However, with newer competition springing up every now and then with low barriers to entry, the road to loyalty seems to be long and tough.

Pull, on the other hand, works by creating a high level of brand awareness, a purple cow of the business. Yours truly hasn't seen a shortcut to building brand awareness. Companies he knows have built brand with extensive investments in mass media advertisements and positive associations with celebrities, all the time communicating the purple cow about themselves . Most companies are still trying to find out how to make content marketing work for them but rarely any has been able to do a good job in the current Indian e-commerce industry by mid 2016. Content marketing ideally needs to be pull based but a lot of current experiments border on blatant push.

An investor will quickly need to assess how much of this repeat rate is dependent upon the direct cash burn in terms of various offers and how much upon the sustained indirect brand spend. The idea is to create such a purple cow that a business doesn't need to give loss making offers to make the customer repeat.

Conversion rates:
A registered user is good only if he/she transacts. You want them to transact as much as possible. your economics is highly sensitive to this metric. As mentioned earlier, this metric benefits from your market share because of the network effects. Apart from that this is a function of typical product-market match, quick and easy discovery of right products on your platform, price competitiveness, quality of information you give, trust you build through product quality and service, and creating cross-sell opportunities.

An investor could gather market intelligence around the conversion rates of various platforms to set a practical target for this metric. Or could also set a target based on where the economics of the equation works out. It helps to look at an e-commerce business in terms of different categories (e.g., mobile phones, women's bags, etc.) because then you could also do internal benchmarking of your conversion rate metrics across different categories and share best practices across the various category teams.

For product-market match, a good way is customer segmentation into different groups with unique category, style, price point, service, freshness, etc. needs. There can be multiple dimensions and one has to make judgement calls in creating the most logical segments which are homogenous within and heterogenous across. Then you've to ensure your selection composition on the above parameters reflects the composition of your target customer segments in the market. A simple example is that in apparel e-commerce, if 70% of your target customer can purchase only a sub-500 price point product then your selection composition has to reflect that. Please note that I am using the phrase 'target-segment'. If you want to build a business targeted only at higher price point customers, you don't need to build a sub-500 price point selection.

Quick and easy discovery of products on your platform is important because in the digital age the customer doesn't have a lot of patience of sifting through multiple pages or filters to find the right products. So personalisation based on past customer behaviour becomes important. This can be based on the individual's past behaviour if your technology is strong enough or based on the overall segment's affinity behaviour. A lot of platforms claim to be using personalisation algorithms but actually their technology is fairly nascent in India in 2016. Personalisation can happen around the specific category, price point, or brand, etc. preference of the customer based on their historical behaviour. Platforms can also leverage statistical methods over large customer sets to predict based on the historical large group dynamics what brand/category/style is the specific customer is likely to prefer.

Other than that you have to continuously experiment with your product navigation to get the customer quickly to where he/she wants to be. This essentially makes product discovery easy on your platform. Identify where customers click the most when they get to your site/app, which pages they visit the most, and use these insights to put the strategically important links in those locations. Strategically important links could be of high conversion rate products, high margin products, or something else based on what your company strategy is at that point of time. Also experiment with new product features which encourage customers to engage more with your platform and discover new products. A very popular technique used by companies these days is the quick A-B testing and iteration to quickly identify what works and what doesn't. This is essentially experimenting a new feature on a test group vs. a control group to assess how the new feature works. Another aspect of aiding discovery is understanding the preferences of your customer groups and making the relevant products easily discoverable. For example, if your low price point buyers prefer some sort of value deals then you could create a page for such value deals on your platform and make that link easily discoverable on your website/app.

Price competitiveness is extremely important in e-commerce because it's very easy for customers to open multiple websites/apps and compare the prices of products. This comparison is easy in concentrated categories, e.g., electronics and electrical appliances, where a few products constitute a major share of category sales but is difficult in fragmented categories, e.g., apparels where a large number of brands and small sellers exist and within each brand/seller the selection is large. Nonetheless, even in fragmented industries there are some core lines which don't see too many design variations, e.g., innerwear or some standard colors/styles in apparels, and are easy to be compared for price range across platforms. It's important for a platform to be price competitive on such sub-categories. Platforms do it in three broad ways: First, develop an automated crawling engine which keeps track of prices of similar products across platforms and keeps changing price of those products on own platform to remain price competitive all the time. Leading platforms have also segmented products into very commonly searched/purchased products vs others and on the former they ensure they have the lowest prices even if on the latter they are expensive and make money. This method works better in concentrated categories. Second, they develop some sort of algorithm which estimates the expected price of a product based on its input materials and tries to see if the product price is broadly in range. This method is vague and less accurate so better suits when you deploy it on a sub-set of selection in fragmented categories. Third, they look at the conversion rates of various products with a hypothesis that lower conversion rate for a sustained period may indicate price uncompetitiveness. Once a platform finds that a product is not competitive on price, it can either take a hit on its margins to become price competitive or it can go back to the seller and ask them to be more price competitive. The former is quick and easy but negatively impacts the unit economics whereas the latter requires manually intervention and takes time but doesn't hurt the platform's unit economics. However, sellers do not always agree to make the products price competitive because the economics doesn't work out for them. In such cases the platform either punishes the vendors in some way or takes the hit itself. Doing business is not so easy after all!

Price competitiveness is also a function of your business model, i.e., inventory based vs. marketplace. An inventory model adds fixed cost to operations but in categories like apparel it has clear advantages in price competitiveness over marketplace. An inventory based apparel e-commerce can procure bulk merchandise at attractive prices which allows it to offer lower prices than regular marketplace distributors while still making some gross margin. Any left over, so long as it's not fast fashion, can be offered at steep discount in next season. Many inventory led players in fact use this model as an advantage to create a perception of higher discounts on branded merchandise which marketplace distributors find difficult to match because they have smaller and usually current season inventories.

Vertical integration is another way of being more price competitive because it reduces the number of parties involved in the transaction and because the platform is willing to accept lower in-house/contracted manufacturing margins compared to pure 3rd party manufacturer-sellers to gain scale through own e-commerce platform. The idea is to make a reasonable margin over the end-to-end value chain and generate cash by higher volumes.

Quality of information you provide becomes important because customer cannot touch and feel the product on your platform. It's important to have a strong product on-boarding process which includes strong quality check of information about the product for comprehensive and accurate information. Companies currently do it in two ways - One, by throwing more people at the problem who review the product information bottom up, Second, by using a carrot and stick mechanism in which they push down the listings of vendors who have poor conversion rates or poor ratings or in which they show extensive customer reviews about the product so that any new customer can make an informed decision. Customer reviews also serve the customer as important peer feedback about the product.

Building trust in your product quality and service is a no brainer in the interest of a long term business. It helps to have a clear set of values and purpose statements of your business early on which act as clear guiding principles to your employees in their day to day decisions. After all, it's day to day decisions which build customer's trust in your products and service. In emerging markets, heavy communication and innovation is required to build trust on the new offering. Some examples of communication are using stamps of quality assurance, return guarantees, secure payments, etc. on your platform pages. An example of innovation is cash-on-delivery service.

Quality control becomes difficult with increasing scale of marketplaces. In such situations, many platforms identify a small, strategic group of sellers who constitute most of the sales and get the team to work closely with them in proactive initiatives in all areas of business, e.g., selection management, right pricing, reducing return rates, etc. This is easy in case of concentrated categories which are anyway dominated by big, well known brands but in fragmented categories the smaller set of vendors can be identified based on the scale of their business or some other parameter that indicates their strategic importance. The other sellers are left to be managed by self service technology and customer feedback.

Typically, companies find it easier to maintain high customer experience by doing things in-house which enables a tighter integration into various processes. E.g., Apple's in-house hardware, software, and services enable a high level of customer experience by better performance through tighter integration. Similarly, in e-commerce, leading players have found that an inventory model helps create a better customer experience than a pure marketplace model. This is because you can tightly integrate logistics with your front end platform and reduce the dispatch and delivery time. This is how Amazon offers one day delivery. You can also do better product quality checks in an inventory model.

Vertical integration, in which companies get into private labelling, also helps higher level of quality control compared to a pure marketplace model where the platform can't always control what's offered by 3rd party sellers.

Once you've invested in acquiring a user, you want to maximise their lifetime value by creating more and more opportunities for cross-sell while remaining aligned with your purpose of existence so that your brand remains consistent in a customer's mind. There is enough literature on how to expand your product offering while maintaining your differentiation in the minds of customers. For example, Gillette expanded from just Men's shaving products into Women's hair removal products by interpreting their purpose as creating quality shaving products. They further expanded the interpretation to overall grooming products and expanded into Men's grooming products. Similarly, an Indian e-commerce platform that stands for enabling discovery of Indian traditional/Ethnic products finds sense in expanding into traditional pickles and eatables segment but not in Western wear segment, you know who I'am talking about! The higher opportunities of cross-sell make the economics of horizontal players (cross-category players) more attractive than vertical ones (single-category players). Due to this reason the lifetime value of the same customer tends to be higher for horizontal platforms than that for vertical platform. This means the horizontal platforms can have a higher customer acquisition cost than vertical platforms. This is another argument in favour of better economics for horizontal platforms vs. vertical platforms. However, the fundamentals of competitive dynamics ensure that in a crowded horizontal market the competitors bid up the cost of customer acquisition so much that it one by one everybody exits the game until eventually only the player with the deepest pockets, the maximum customer engagement, or the least cost remains.

Average order value:
This has proven to be a tough metric to crack for most platforms. It can be broken into the following equation:
Average order value = Units per order X Average selling price per unit

Units per order can be influenced by recommendations at the time of product views or when products are added to the cart. Amazon does it very well by showing 'frequently bought together' or 'Customers who bought this also bought...". Higher trust in your product quality and platform service also favourably impact this metric. Two ways platforms have built trust in their product qualities is either by building a brand for themselves or by listing strong brands as part of their selection. In either case, a brand name usually increases trust.

Average selling price per unit tends to be a function of composition of your visiting user base and their purchasing power and to that extent tends to be beyond control. However, higher trust in your product quality and platform service favourably impact this metric.

Unit economics:
This means that you make a positive gross margin on each of your orders. You could be cross subsidising products within an order in the interest of increasing your overall GMV and keeping your sellers' growth broad based as long as you make the required margin on an order to cover your marketing and fixed costs and leave the required profit for a reasonable RoA.

Private labelling through vertical integration improves the gross margin across the combination of manufacturing and selling because both activities now come under the same company. In this case the combined manufacturing and selling margin of the platform is lower than the sum of manufacturing-selling margin of the 3rd party seller and the marketplace fee of the platform. The platform hopes to gain volume by passing on part of the margin of combined operations to the customer and being price competitive. The whole idea is to generate reasonable cash at large volumes even if at lower percentage margin. That's the whole game of businesses characterised by network effects.

In recent times and especially in business characterised by heavy network effects, businesses are disregarding unit profitability in the hope that once they have clearly established themselves as a preferred go-to destination in the minds of customers, they will make reasonable profit. The success of this approach has been proven in developed markets, e.g., by Amazon in the US. However, there is always a risk of taking it too far amid intense competition in a developing market characterised by low per capita purchasing power. There are two key risks: One, you get so much into the red during the investment period that recovering the investment becomes extremely difficult in a predictable time frame and Second, the still-low purchasing power of the customers in future makes it difficult for you to recover your initial investment.

As I said earlier, what will happen is a matter of debate but once the outcome becomes clear it will serve as an extremely important learning experience for businesses. Investors sometimes try to hedge their risks by keeping a clear timeframe in mind within which they expect the platform to break-even on a customer and if the platform misses this target for a few quarters then they wouldn't fund the platform anymore. These days (mid-2016) the thumb rule for investors seems to be that a platform should break even on a customer in 12-18 months because the repeat rates of customers tend to be really weak post that.

A new area which leading platforms have started using from last one year is advertising. Earning from advertisement goes directly to the bottomline of the platform hence is very attractive to otherwise unprofitable platforms. Advertising on e-commerce platforms happens in the following key ways: One, the sellers pay the platform to show their products as sponsored products in product search results, Two, some unethical platforms may also artificially push the advertisers results automatically in organic category listings, Three, sellers pay the platform to show their banners/links on several pages or on various assets of the platform. However, platforms will find out that marketing objectives from advertisements tend to be different for large brands vs. smaller sellers. Whereas large brands spend money from their marketing budget with an objective of strengthening brand awareness, smaller sellers do not have well defined business practices and do not set aside any marketing budget. In fact, the very fact that small sellers are not brands means that they also do not command enough margins that could fund large advertising campaigns. This means that small sellers tend to expect 1-to-1 attribution of any advertising money to increased sales. This is a huge ask on the part of e-commerce platforms which are otherwise occupied with many larger issues of their own. All of this leads to such advertising being primarily led by few well known brands and not a large number of small sellers.

E-commerce is a complex and exciting industry and no one seems to have cracked the winning formula yet in India. As the various players fight a bloody battle out, I hope they remember that the real skill is in making profitable businesses. Anybody can put a desk outside and pay people hundred rupees to come buy a eighty rupee handkerchief.

Are your large acquisitions creating or destroying value?

A McKinsey research says large acquisitions destroy value more often than not. That research recommends a programmatic acquisition which is regular, multiple small acquisitions meant to fill specific gaps in an organization's capabilities, offerings, or geographical presence. IBM has been a poster child of such a program in recent two decades.

When I read such pieces of research, most examples tend to be older, preceding the research. I wanted to start documenting more recent examples that offer some learning on value creation/destruction through inorganic pursuits. Perhaps more failures will get highlighted but I will try to find successes too. I will keep adding new examples as they come up to my notice. If you know any new examples, pls share with me. So, here we go:

HP Enterprises's divestment of its software portfolio:

  • Acquisition: Britain's Autonomy Corp Plc in 2011 for $10.3 bn. Integrated into HP as a software business.
  • Divestment: To Britain's Micro Focus International Plc on 8-Sep-2016 in a deal worth $8.8 bn ($2.5 bn cash and 50.1% equity in the combined company). Additional expense of $700 mn in one time costs related to the separation of the assets.
  • There will also be some accumulated profits/losses for the duration of ownership which I haven't tried to find.
Intel's divestment of McAfee:
  • Acquisition: McAfee (internet security software company) for $7.7 bn in 2011 and renamed as Intel Security Group in 2014
  • Divestment: Sold 51% stake to investment firm TPG for $3.1 bn in cash on 8-Sep-2016. Brand name to be reverted to McAfee. TPG can leverage synergies with its other two investments in security start-up Tanium and Zscaler
  • Intel's investment thesis:
    • Integrate McAfee security technology into Intel chips but no progress made
    • Get a piece of the emerging business of protecting corporations from sophisticated espionage, but newer players such as Mandiant, came to dominate the business
    • Gain from market share increase in PC security but PC growth itself slowed