Market-making in two-sided on-demand markets

Every on-demand market has “liquidity providers” and “liquidity takers”. In fact, the market is “on demand” only for the liquidity takers, for when they arrive at the market there are liquidity providers who are providing this liquidity. In other words, the liquidity takers’ demand is instantaneously supplied by the liquidity providers, who make sure that the market is on demand.

Now, providing liquidity is risky business. Let us look at it from the perspective of stock markets, where the concept of market making is most well established. Market makers in a stock market are responsible for keeping “live quotes” (on both the buy and sell side) at all points in time, so that whenever a trader enters the market, there is ready liquidity available for the trader to execute his trade “on demand”.

Let’s say I’m a market maker and have put out a bid (offer to buy) on a particular stock. Now, if there is a massive downward price movement in the market, my bid gets “taken out” (i.e. a counterparty takes me up on my offer to buy and sells me the stock at the price I’ve quoted), and the market continues hurtling down. Notice now that the moment my bid got “taken out”, I have a long position in the market (since I have now purchased the stock), and the continued downward movement of the market leads to a loss. It is similar if I’m making a market on the ask side and there’s a massive upward price movement.

To put it another way, the market makers (i.e. liquidity providers) are effectively “short optionality”, as they constantly need to write the option of trading at a particular price at any given instant. The “liquidity taking” side is effectively “long optionality”, since they have the option to trade at that particular instant at the price quoted by the market maker.

The question is how I as a “liquidity provider” (as a market maker I’m providing liquidity) need to be compensated for the liquidity I’m thus providing. In the American stock exchanges, for example, market makers are charged a much lower exchange fees than the liquidity takers (the opposite side which trades with the liquidity provided by the market maker). Sometimes the exchange even provides a fee to the market maker in exchange for “always being there” in the market. This is a recognition of the costs being borne by the market maker in order to provide liquidity. The question is how this can be replicated in other “on demand” markets.

It is well known that taxi providers such as Ola and Uber provide attractive “non-linear” (not directly tied to rides) incentives to drivers for being “switched on”. The economic rationale behind that is to compensate these drivers for the cost and risk associated with being “switched on” (there is opportunity cost in being switched on in terms of the driver’s time, risk of getting an unattractive ride, etc.). These incentives, are in other words, compensation to the drivers for “providing liquidity” to the market. In the long run, this gets paid out of the fees charged from the liquidity takers.

Similarly, in the electricity market (though we seldom look at it that way, electricity is the classic “on demand” market, since electricity cannot be stored and needs to be generated in response to demand), grids typically pay power producers a fixed cost to just “be there”. This is again in addition to the per unit cost of power that the producers are paid when power is actually drawn from them (this is the classic “two part tariff” that is prevalent in the electricity sector). On similar lines, think of a lower power tariff to customers who are willing to accept frequent power cuts!

In “traditional, one-sided” markets, where the seller of the good/service is also the owner of the “market” where the sale takes place, the seller usually absorbs this cost of providing liquidity. For example, your neighbourhood grocer (note that this is again “on demand” since the grocer sells you groceries whenever you demand it) provides liquidity by way of keeping his shop open (even when there are no customers) and maintaining inventory. A car mechanic provides “on demand” service by (again) keeping his shop open and having spare mechanics who can service spot demand immediately.

In a platform business (or “two sided market”, or a market where the owner of the marketplace is not a participant), however, the owner of the market cannot provide liquidity himself since he is not a participant. Thus, in order to maintain it “on demand”, he should be able to incentivise a set of participants who are willing to provide liquidity in the market. And in return for such liquidity provided, these providers need to be paid a fee in exchange for the liquidity thus provided.

Thus, the key to the success of an “on demand” marketplace is to be able to incentivise enough market participants who are willing to provide liquidity – who are willing to forego instantaneous matching, and take risk of not being matched so that other market participants can enjoy the liquidity thus provided. This process also involves the clever structuring of incentives such that such compensation to the liquidity provider is paid for by the liquidity taker – this is nontrivial since this may not be consistent with the way in which the marketplace actually charges participants!

Cross posted from LinkedIN

Rating systems need to be designed carefully

Different people use the same rating scale in different ways. Hence, nuance is required while aggregating ratings taking decisions based on them

During the recent Times Lit Fest in Bangalore, I was talking to some acquaintances regarding the recent Uber rape case (where a car driver hired though the Uber app in Delhi allegedly raped a woman). We were talking about what Uber can potentially do to prevent bad behaviour from drivers (which results in loss of reputation, and consequently business, for Uber), when one of them mentioned that the driver accused of rape had an earlier complaint against him within the Uber system, but because the complainant in that case had given him “three stars”, Uber had not pulled him up.

Now, Uber has a system of rating both drivers and passengers after each ride – you are prompted to give the rating as soon as the ride is done, and you are unable to proceed to your next booking unless you’ve rated the previous ride. What this ensures is that there is no selection bias in rating – typically you leave a rating only when the product/service has been exceptionally good or bad, leading to skewed ratings. Uber’s prompts imply that there is no opportunity for such bias and ratings are usually fair.

Except for one problem – different people have different norms for rating. For example, i believe that there is nothing “exceptional” that an Uber driver can do for me, and hence my default rating for all “satisfactory” rides is a 5, with lower scores being used progressively for different levels of infractions. For another user, for example, the default might be 1, with 2 to 5 being used for various levels of good service. Yet another user might use only half the provided scale, with 3 being “pathetic”, for example. I once worked for a firm where annual employee ratings came out on a similar five-point scale. Over the years so much “rating inflation” had happened that back when I worked there anything marginally lower than 4 on 5 was enough to get you sacked.

What this means is that arithmetically averaging ratings across raters, and devising policies based on particular levels of ratings is clearly wrong. For example, when in the earlier case (as mentioned by my acquaintance) a user rated the offending driver a 3, Uber should not have looked at the rating in isolation, but in relation to other ratings given by that particular user (assuming she had used the service before).

It is a similar case with any other rating system – a rating looked at in isolation tells you nothing. What you need to do is to look at it in relation to other ratings by the user. It is also not enough to look at a rating in relation to just the “average” rating given by a user – variance also matters. Consider, for example, two users. Ramu uses 3 for average service, 4 for exceptional and 2 for pathetic. Shamu also uses 3 for average, but he instead uses the “full scale”, using 5 for exceptional service and 1 for pathetic. Now, if a particular product/service is rated 1 by both Ramu and Shamu, it means different things – in Shamu’s case it is “simply pathetic”, for that is both the lowest score he has given in the past and the lowest he can give. In Ramu’s case, on the other hand, a rating of 1 can only be described as “exceptionally pathetic”, for his variance is low and hence he almost never rates someone below 2!

Thus, while a rating system is a necessity in ensuring good service in a two-sided market, it needs to be designed and implemented in a careful manner. Lack of nuance in designing a rating system can result in undermining the system and rendering it effectively useless!

Meru’s pricing strategy

Let’s assume I’m writing this post two weeks back when Uber, Ola and TaxiForSure were still running successfully in most places in India. Since then, they’ve been banned to various degrees and it’s gotten harder for customers to get them and for drivers there to find customers leading to a sharp drop in volumes.

Thanks to the entry of app-based taxi booking services such as Uber, Ola and TaxiForSure, entrenched players such as Meru Cabs and Easy Cabs started losing business. This is not unexpected, for the former operated at around Rs. 13-15 per km range (depending on discounts, time of day, etc.) while the latter operated around the Rs. 20 per km price point. This meant that for immediate trips and mostly intra-city movement consumers eschewed the likes of Meru and embraced the likes of Ola.

In the last few weeks I’ve spoken to taxi drivers (mostly Uber; Ola drivers don’t inspire much confidence and so I don’t indulge them in conversation; and I’ve never got a cab via TaxiForSure) who have been affiliated to more than one aggregator, and from that I get what the problem with Meru’s pricing is.

What sets apart Meru, KSTDC and Mega Cabs is that the three are the only operators with a license to pick up passengers from the taxi rank at the Bangalore Airport. Any other taxi that you might book (Ola or Uber or a local cabwallah) don’t have the rights to pick up passengers there and park in the airport’s taxi parking zone. They instead have to park in the space allocated to private cars, paying the parking fees there, and  there is usually a delay from the time when the driver meets the customer at the arrival gate to the customer actually getting into the car. This distinction means that the likes of Meru and Mega offer superior service to the other operators at the airport and thus can command a premium price. Getting into anecdata territory but I always prefer to get a cab from the taxi rank (though the queue occasionally gets long) than to book a cab for which I’ve to wait.

At the city end, the difference between Meru and Uber (Ola is in an intermediate state) is that you can pre-book a Meru, while Uber only accepts “spot bookings”. This difference in service levels means that you can never be assured of getting an Uber at the time you want to leave for the airport – there is a statistically high chance of getting one but you don’t want to take the risk, and thus prefer to pre-book a Meru or a Mega, which lets you know at the time of booking if they are able to service you.

Now, this guarantee from a Meru or a Mega comes at a cost. An Uber cabbie who also drove for Easycabs told me that Easycabs would allocate his trip an hour before it was scheduled to start. Since Easycabs would have assured the customer of a cab reaching his place at the appointed time, this means that they need to account for a sufficient buffer to ensure that the cab does reach on time. Thus the allocation an hour in advance. This cabbie told me that from his point of view that was inefficient, for in the one hour of buffer that EasyCabs would add, he could complete one additional trip through Uber!

So it is clear as to why Meru is more expensive than Uber/Ola – their pre-booking provision means that they have to potentially ground your cab for an hour before pickup, and there is a license fee they have paid the airport for the right to pick up passengers from the taxi rank there. Notice that both these factors also result in increased convenience for passengers. So effectively, Meru is justified in charging a premium. The question is if the current structure is optimal.

The problem with Meru is that their fare structure doesn’t appropriately represent cost. A pre-booked taxi costs as much as a taxi hailed at the time of demand. A taxi from the airport (where they have paid license fee) costs as much as a taxi from anywhere else. So while their cost structure might be optimal for travel to and from the airport, the structure simply doesn’t work out for other rides. And they are getting priced out of non-airport rides.

Assuming that they want to get more non-airport rides for their fleet, how do they do it? The answer is rather simple – let the fare structure reflect cost. Rather than tacking on every piece of cost to the per kilometer fare, they can have a multi-part fare structure which is possibly more “fair”.

A typical trip from the airport to the city is about 40 km, and costs around Rs. 800 (excluding service tax). Instead of charging Rs. 20 per trip, how about charging Rs. 16 (Ola’s rate) per kilometer and an additional Rs. 200 “airport charge”? At the other end, how about charging an additional Rs. 100 or Rs. 200 as pre-booking charge in order to account for driver’s idle time on account of the pre-booking? If they were to charge this way, they will both make as much money as they currently do on airport trips, and also compete with Ola and Uber on intra-city immediate-ride trips.

To take an extreme analogy, this is like asset-liability management – prudent banking dictates that the term structure of your assets reflects that of your liabilities. Similarly, prudent pricing (to the extent it is practically implementable) dictates that your price structure reflects on your cost structure!

Offline marketing of online services

Using snail-mail for marketing is an effective strategy for it grabs more of your attention. But messages need to be more personalised to have effect.

This came in the mail yesterday. If you are an old-timer like me, you will recognise it as an “inland letter card”. The edges are frayed because it had been so long since I’d received one such card that I’ve forgotten how to open them.

bigbasket

You will notice that this inland letter came from Bigbasket, the online grocery shopping firm. At first look, it is bizarre that an e-commerce firm is using snail mail for its marketing. On second thoughts, though, it isn’t that bizarre!

The thing with online modes of communication such as email or SMS is that the cost of sending a message is low, very close to zero. What this leads marketers to do is to bombard you with messages. For example, I bought something from Jabong a couple of weeks back and they’ve since sent me at least an SMS a day. I promptly delete them without reading. On my email, I’ve been unsubscribing wherever possible from promotional lists from which I get messages – for they are too frequent and too “vanilla” (it’s bizarre that even marketers who know much about me refuse to use that information in their communication).

In short, there is too much clutter in online (email/SMS) marketing, and the chances of any promotion really standing out and getting the user’s attention is minuscule.

Sending snail-mail, on the other hand, is expensive. It costs you to buy the paper, print out the letters and then you pay for postage. This means that with the advent of cheaper means of communication, most marketers have moved away from it. What that has done is that you get much lesser snail-mail than you used to a few years ago. Which means that the amount of attention you devote to each snail-mail is actually more!

So with snail-mail being the more expensive form of marketing, it is actually more effective for marketers because it draws your attention! (You can think of it as a multi-player prisoner’s dilemma where the marketer wants to maximise her claim on your attention (relative to her costs), and can do so by either using email or snail-mail. The optimal solution, I believe, is a kind of “mixed strategy” – mostly email, but the odd snail-mail here!)

So an online sales company reaching out to you by snail mail is not that bizarre after all. If only they had customised the mail to put my name on it (not hard to do at all), and made it seem like a personal letter, it would have been even more effective!

There have been two occasions in the last five years when I’ve actually responded to upsell campaigns. One was by Airtel who called and offered me a 3G data plan for almost the same price as what I was then paying for my 2G plan. I had been intending to upgrade and I took it.

The other was by Tata Sky, who sent me a beautifully crafted personalised letter printed on thick A4 paper, indicating I was a “premium subscriber” and asking if I wanted to upgrade to Tata Sky+ HD, and giving me a number of a dedicated call center who I had to call to upgrade. It is likely that had it been email I might have discarded it (or if I were using today’s Inbox, marked it as “Done”). Snail mail drew more attention, and the personalisation made me feel good. And I upgraded.

Fragility of two-sided markets

Two-sided markets are inherently fragile for participation of each side depends on a certain degree of confidence in participation on the other side. Thus, small negative shocks can lead to quick downward spirals.

Following the ill-advised ban on Uber and other taxi aggregators in four Indian states (Delhi, Karnataka, Andhra Pradesh, Telangana), business for drivers who ply their services via such apps has dropped significantly. While on first inspection you might expect it to go to zero (given their services have been banned), the fact that enforcement is tough (there is nothing to identify a cab as “belonging to Uber”) means that apart from Delhi (where Uber has pulled its services) these cabs continue to ply.

In the days after the ban, various news reports have interviewed drivers who ply for Uber who complain about drastically reduced services. While numbers vary from report to report, the general sense is that so far the number of trips per driver per day has fallen by half. And I expect this to fall further unless drastic steps are taken – such as issuance of new regulations or removal of the ban.

In a “normal” market (where the owner of the market is also a participant), when demand for a particular good drops, price is expected to fall and availability is expected to increase. If demand for a particular item that you have in stock drops, you need to take steps to get rid of the excess inventory that you have. You are most likely to indulge in discounting or other such promotional activities, in order to make it more attractive for the buyers to buy, and thus take the inventory off your shelves.

In a “two-sided market” (one where the owner of the market is not a participant), however, things work differently. It is a popular saying that in such markets “demand creates its own supply”. A corollary to that is that “lack of demand creates lack of supply”. Let us take the case of Uber itself. Over the last few days, irrespective of whether the ban on the service is official or not, legal or not, the number of people who have been requesting for the service has dropped.

Now, if you are a driver using the app, you realise that your potential revenues and profits from continuing to use the app are not as high as they used to be. Thus, if there are other avenues for you to make money, you are now more likely to take those avenues rather than logging on to Uber (since the “hurdle rate” for such a switch is now lower thanks to lower Uber revenues). As many of you take the same route, the availability of cabs on Uber also drops – something that I’ve seen anecdotally over the last few days. And when availability of Uber cabs drops beyond a point, I start questioning my trust in the service – a week ago I would be confident that I would be able to hail an Uber from anywhere in Bangalore with very high confidence; that confidence has now dropped. And when my trust in the service drops, I start using it less, and when many of us do that, drivers see less demand and more of them pull away from the market. And this results in a vicious cycle.

Notice that things would work very differently had Uber been a “traditional” taxi service which owned its cabs and employed its drivers. In that case, falling demand would have been met with a response that would have made it easier for customers to buy – price cuts, perks, etc.

The point is that platforms or two-sided markets are inherently fragile, and highly dependent on confident in the system. I leave my car at home only if I have enough trust in the taxi platforms that I’ll be able to get a cab when I need one. A driver will forsake other trips and switch on his Uber app only if he is confident that he can get enough rides through the app.

The same network effects that can lead to a rapid ramp-up in two-sided markets can also lead to its downfall. All it takes is a small trigger that leads to loss of confidence in the service from one side. Unless that loss of confidence is quickly addressed, the “positive feedback” from it can quickly escalate and the market grinds to a halt!

Another good example of lack of confidence killing two-sided markets is in the market for CDOs and associated derivatives in 2007-08. There were standardised pricing models for such products and a vibrant market existed (between sophisticated financial institutions) in 2007. When house prices started coming down, some people started expressing doubts in such models. Soon, this led to massive loss of trust in the pricing models that underpinned such markets and people stopped trading. This meant companies were unable to mark their securities to market or rationalise their portfolios, and this led to the full-blown 2008 financial crisis!

So when you build a platform, you need to make sure that both sides of the market retain confidence in your platform. For in the platforms business loss of confidence can lead to a much quicker fall than in “traditional” markets. This dependence on confidence thus makes such markets fragile.

Practo and rating systems

The lack of a rating system means Practo is unlikely to take off like other similar platforms

So yesterday I found a dermatologist via Practo, a website that provides listing services for doctors in India. I visited him today and have been thoroughly disappointed with the quality of service (he subjected me to a random battery of blood tests – to be done in his own lab; and seemed more intent on cross-selling moisturising liquid soap rather than looking at the rash on my hand). Hoping to leave a bad review I went back to the Practo website but there seems to be no such mechanism.

This is not surprising since doctors won’t want bad reviews about them to be public information. In the medical profession, reputational risk is massive and if bad word gets around about you, your career is doomed. Thus even if Practo were to implement a rating system, any doctors who were to get bad ratings (even the best doctors have off-days and that can lead to nasty ratings) would want to delist from the service for such ratings would do them much harm. This would in turn affect Practo’s business (since the more the doctors listed the more the searches and appointments), so they don’t have a rating system.

The question is if the lack of a rating system is going to hinder Practo’s growth as a platform. One of the reasons I would go to a website like Practo is when I don’t know any reliable doctors of the specialisation that I’m looking for. Now, Practo puts out some “objective” statistics about every doctor on its website – like their qualifications, number of years of experience and for some, the number of people who clicked through (like the doctor I went to today was a “most clicked” doctor, whatever that means), but none of them are really correlated with quality.

And healthcare is a sector where as Sangeet Paul Chaudary of Platform Thinking puts it, “sampling costs are high”. To quote him:

There are scenarios where sampling costs can be so high as to discourage sampling. Healthcare, for example, has extremely high sampling costs. Going to the wrong doctor could cost you your life. In such cases, some form of expert or editorial discretion needs to add the first layer of input to a curation system.

So the lack of a rating system means that Practo will end up at best as a directory listing service rather than as a recommendation service. Every time people find a “sub-optimal” doctor via Practo, their faith in the “platform” goes down and they become less likely to use the platform in the future for recommendation and curation. I expect Practo to reach the asymptotic state as a software platform for doctors to manage their appointments, where you can go to request an appointment after you’ve decided which doctor you want to visit!

Potential investors would do well to keep this in mind.

Update

Today I got an SMS from Practo asking me if I was happy with my experience. I voted by giving a missed call to one of the two given numbers. I don’t know how they’ll use it, though. The page only says how many upvotes each doctor got (for my search it was all in the low single digits), so is again of little use to the user.

Categorisation and tagging

Tagging offers an efficient method to both searching and for identifying customer preferences on the axis most appropriate for the customer

The traditional way to organise a retail catalogue is by means of hierarchical categorisation. If you’re selling clothes, for example, you first divide it into men’s and women’s, then into formal and casual, and then into different items of clothing and so on. With a good categorisation, each SKU will have a unique “path” down the category tree. For traditional management purposes, this kind of categorisation might be useful, but it doesn’t lend itself well to both searching and pattern recognition.

To take a personal example (note that I’m going into anecdata territory here), I’m in the market for a hooded sweatshirt, and it has been extremely hard to find. Having given up on a number of “traditional retail” stores in the “High Street” (11th Main Road, 4th Block, Jayanagar, Bangalore) close to where I stay, I decided to check online sources and they’ve left me disappointed, too.

To be more precise, I’m looking for a grey sweatshirt made with a mix of cotton and wool (“traditional sweatshirt material”) with a zipper down the front, pockets large enough to keep my hands and a hood. Of size 42. This description is as specific as it gets and I don’t imagine any brand having more than a small number of SKUs that fit this specification.

In case I were shopping offline in a well-stocked store (perhaps a “well stocked offline store” is entering mythical territory nowadays), I would  repeat the above paragraph to a store attendant (good store attendants are also very hard to find nowadays) and he/she would pick out the sweatshirts that would conform to these specifications and I would buy one of them. The question is how one can replicate this experience in online shopping.

In other words, how can we set up our online customer catalog such that it becomes easy for shoppers to search specifically for what they’re looking for. Currently, most online stores follow a “categorisation” format, where you step into two or three levels of categorisation, where you’re shown a large assortment. This, however, doesn’t allow for efficient search. Let me illustrate by my own experience this morning.

1. Amazon.in : I hit “hoodies” in the search bar, and got shown a large assortment of hoodies. I can drill deeper in terms of sleeve length, material, colour and brand. My choice of material (which I’m particular about) is not there in the given list. There are too many colour choices and I can’t simply say “grey” and be shown all greys. There is no option to say i want a zip-open front, or a cotton-wool mix. My search ends there.

2. Jabong (rumoured to be bought by Amazon shortly): I hover over “Men’s”, click on “winter wear” and then on “hoodies”. There is a large assortment of both material (cotton-wool mix not here) and brand. There are several colours available, but no way for me to tell the system I’m looking for a zip-down hoodie. I can set my price-range and size, though. Search ends at a point when there’s too much choice.

3. Flipkart: Hover over “men’s”, click “winter wear” and then sweatshirt. Price, size and brand are the only axes on which I can drill down further. The least impressive of all the sites I’ve seen. Too much choice again at a point when I end search.

4. Myntra (recently bought by Flipkart, but not yet merged): The most impressive of all sites. I hover over “Men’s” and click on sweaters and sweatshirts (one less click than Jabong or Flipkart). After I click on “sweatshirts” it gives me a “closure” option (this is the part that impresses me) where I can say I want a zippered front. No option to indicate hood or material, though.

In each of the above, it seems like the catalog has been thought up in a hierarchical format, with little attention paid to tagging. There might be some tags attached such as “brand” but these are tags that are available to every item. The key to tagging is that not all tags need to be applicable for all items. For example, “closure” (zippered or buttoned or open) is applicable only to sweatshirts. Sleeve length is applicable only to tops.

In addition to search (as illustrated above), the purpose of tagging is to identify patterns in purchases and know more about customers. The basic idea is that people’s preferences could be along several axes, and at the time of segmentation and bucketing you are not sure which axis describes the person’s preferences best. So by having a large number of tags that you assign to each SKU (this sadly is a highly manual process), you give yourself a much superior chance of getting to know the customer.

In terms of technological capability, things have advanced much in terms of getting to know the customer. For example, it is now really quick to do a Market Basket Analysis based on large numbers of bills, which helps you identify patterns in purchase. With the technology bit being easy, the key to learning more about your customers is the framework you employ to “encase” the technology. And without efficient tagging, you are giving yourself a lesser chance of categorising the customer on the right axis.

Of course for someone used to relational databases, tagging requires non-trivial methods of storage. Firstly the number of tags varies widely by item. Secondly, tags can themselves have a hierarchy, and items might not necessarily be associated with the lowest level of tag. Thirdly, tagging is useless without efficient searching, at various levels, and it is a non-trivial technological problem to solve. But while the problems are non-trivial, the solutions are well-known and advantages large enough that whether to use tags or not is a no-brainer for an organisation that wants to use data in its decision-making.