Ben's Website

Serious musings

Competing with Amazon: If I were eBay

Efficient markets are a choice!

Over the course of 2019 Amazon came under increasing fire for the scope of its marketplace and the market power that conferred. The recent revelation that it used internal, private seller information to gauge markets and design competing, white-label goods is not going to help. At the same time, the covid cessation has significantly increased Amazon’s sales and our reliance on them, leaving many with a newfound-yet-sickly appreciation for their scale and platform. With this as a motivation, I’d like to focus on Amazon’s integration of 3rd party sellers alongside the first-party goods they directly distribute. (See Stratechery for a deeper discussion on Amazon's model.)

I don’t like the inclusion of 3rd party sellers in Amazon because it is anti-consumer. To see this, consider the story from typical retail: shelf space is limited, leading a competition between the number of products that can be sold and the amount of shelf space devoted to that. The finite resource of shelf space limits the number of products that may be sold, driving the retailer to select which brands they are willing to stock. While name-brands enjoy brand affiliation and command higher prices, retailers realize that consumers often desire a generic competitor to keep the name-brand prices in check, and so they choose certain generic brands to stock next to the name-brands. Retailers leverage their limited shelf space to encourage name-brand competition while also delivering value to customers by stocking generics.

In contrast, Amazon's shelf space is unlimited. While carrying the major brands, it is also able to care every knockoff and imitation, reducing the signaling power of a brand name. That is, while choosing between two known brands (HP and Dell, say) invites consumers to bring their prior experiences with the brand into the purchasing decision, a multitude of brands undermines the value of experience, begging the question of why the Dell costs more than a white-label while having the same feature list.

At the end of the day, Amazon wants purchases; they structure their 3rd party marketplace to be as profitable as their own inventory, so that it doesn't matter to Amazon whether you're purchasing from them or a 3rd party sellers, despite the difference in quality. As long as the purchase happens on Amazon, you'll renew Prime. With this equivalency, Amazon has an incentive to decrease the quality of the product information shown in the search and browsing results, to focus the customer on price while minimizing the value of brand or other measurable dimensions of performance.

Most people can’t taste a difference between Kellogg’s Froot Loops and Malt-O-Meal’s Tootie Frooties, the difference is entirely in the packaging and branding. So if the retailer can remove the value applied to the brand and packaging, they can save their customer money which allows the customer to spend elsewhere. Notice that this savings is entirely dependent on whether people can taste a difference between brands, if you remove their ability to recognize the brand or packaging you can still save them money without actually competing on taste. This is just an example of how Amazon benefits from commoditization on the consumer side; they benefit from the total volume of sales and, in their control of the Amazon.com search window, have the ability to distort and hide actual product differences, harming the customer. For these reasons I've opposed and complained about the Amazon experience, and wish there were a better way. What would it take to compete with Amazon?

At this stage it’s basically impossible to compete directly, toe-for-toe with Amazon’s warehouses, supplier relations, web platform, loyal consumers and unrelated business lines. Actual, consumer-benefitting competition requires a different approach, one that severs the harmful link between Amazon as a consumer-facing retailer and distributor, and Amazon's 3rd party market.

Competing with Amazon

eBay and other auction sites have this differential approach, despite malinvestment in their core experience. They already provide a mechanism for the sale of single units and can thus access smaller sellers than Amazon's 3rd party stores. While eBay has direct sales of new goods, a significant fraction of their sales is resale of rare or random (non-standardized) items. This is a key differentiator: where Amazon encourages purchase of brand-new, active production items (and their quick disposal), virtually all of eBay is resale of items that have residual value in their original function or some subset thereof.

An aside: healthy markets have resale:

It can never be cheaper to produce a new thing than to use one that already exists within some measure of local. To the extent that it is cheaper to buy a new thing than to use the existing, that is a distorted market. As I've mentioned before, I once paid 99c for a tiny package of clover seeds from India (total price cheaper than postage to send a letter to India). Amazon's 3rd party seller included a packet of Saffron seeds so that (I believe) the entire envelope could be treated as a 'gift' for customs purposes. Even if the whole 99c made it to the seller, it is cannot be worth their time to bag, label, ship, and manage the whole affair. It only makes sense if the market is distorted, possibly in this case by the Universal Postal Union and Amazon-related seller volume incentives.

eBay has lots of sellers providing access to a potentially unlimited selection of items, with a default pricing mechanism that is per-sale optimal (auctions, by definition, achieve the optimal seller/buyer price because the item would have sold earlier to another buyer if that lower price was correct).

So, eBay should be able to satisfy any demand and compete in significant ways with Amazon: while eBay lacks free shipping and supply chain efficiencies, it has increased product selection and longer (more planned) sales decisions. Amazon still has an advantage for disposable or convenience goods, where the primary value is to free the customer from having to remember to make that purchase—one-click* and the decision is over and the item is on its way to you. These regular purchases have zero purchase risk: the customer is already well-acquainted with the product and has already decided to buy it, the only variable is whether it is available and at the remembered or expected price. Hence, Amazon's Dash buttons and subscribe-and-save.

But higher-value purchases are accompanied by increased purchase risk due to: the customer having not purchased this item before (uncertain on the brand, category, model, vendor, etc.) or uncertainty that it will work as they intend, as there is more marketing and (dis)information on higher-valued items. Adding to that, the purchase risk is magnified by the overall purchase price, so there is more to lose, even with the fine-printed free returns. To compete with Amazon, eBay needs to create the infrastructure for sellers to maximally communicate product attributes. (I'd like to see a study showing that, for a common product, more complete product descriptions do lead to faster/more sales because the purchase risk is less...perhaps comparing the number of pictures in a listing vs. the time until sale.)

To this, I am a strong proponent of total product specification. This is the coding of product features and specifications into standard formats that allow comparisons to be made. Getting these details right is often regarded as a wasted effort, because manufacturers lack the incentive to copy/paste from their internal specification or datasheet into the platform’s product database, they often can’t agree on standardized representations of items, and they're afraid to overburden customers with all the differences. eBay can solve these issues in multiple ways.

First, by publishing and regularly updating the sales-vs-information study I described above. Information is plentiful and this isn’t a very hard thing to do at scale. eBay has various seller analytics but doesn’t appear to get beyond the search engine and advertising perspectives (when should I list my product, what is a good seller page design, what ads should I buy), to actually informing sellers how unique their listing is relative to the others in-category, and what chain of filters the potential customer used to arrive at their listing. (This is some of the information that Amazon is alleged to have used in commissioning it's knockoff products.)

Once sellers have the ability to understand their sales, they can appreciate the value of fully-specified products. As a general principle information should not be intentionally withheld or suppressed, this can only decrease market efficiency and limit how fully the buyer can use the product. For instance, we recently bought a refurbished monitor for the home office; the seller didn’t say what was wrong with it so when it arrived we pulled it out and saw a huge crack down the middle of the screen and a chipped corner bezel. Was either of these the original flaw that motivated the first return and then ours? We likely would have accepted the damaged bezel but obviously rejected the cracked screen. We don't know what was 'refurbished' and because this information was not communicated to us, potentially leading us to waste two weeks and while the seller paid twice for shipping. We both lost something.

And this is a key point: If you're in a price war and don't want to be, you must find some other way to give the customers value. If you can't lower prices and must charge 5$ more than your competitor for similar items, you have to give your customers 5$ or more value in non-monetary terms. Local bookstores do this through cozy decor, personal connection, and recommendation. Car dealers advertise their prompt maintenance, loaner cars, and support of the local pee-wee football team.

eBay sellers can do this by decreasing purchase risk, providing more information to buyers and enabling them to use the product in the widest possible set of scenarios. Take, for instance, selling used (removed-from-equipment) AC motors on eBay. This is a commodity product with purchases driven by the total cost (unit + shipping + time) and some qualitative estimate of future utility. The motor’s base specifications are quite readable off the motor plate, so sellers can very quickly describe motors as 1Hp 3000 rpm 3/4” shaft. But, if they can tell me its former use, I can imagine its past and project its future life: use in a continuous duty application (pumping/fan) leads me to look at the bearings and how easily they might be replaced; use in process control implies lower lifetime revolutions, but lots of stop/starts potentially leading to heat-related winding damage. So, specifying something that the seller knows allows the average-to-sophisticated user to get more value out of the purchase at little cost to the seller. It’s not like the seller had some other use for this information (the only weakness is if the buyer misinterprets the information, but that is always a risk).

eBay's burden here is to 1) develop the seller interface and tools to allow them to easily and efficiently import this data (say making the eBay seller app able to take pictures, OCR the data, and direct that data into the listing fields), 2) create powerful buying interfaces that exploit the data including enabling user-created and recommended-market filters, and 3) publicize information about the health of this market.

Second, the world is complex and it can be challenging to measure some product attributes. When the market's needs are unknown, manufacturers will inevitably come to different prescriptions for the same need. This means that interchangeable products can be designed and sold under different operational and market models, challenging direct comparison. This is where fully-specified products can unlock positive, customer-serving dynamics: the best way to compete is to show how your competitor's specifications compare to your own (playing on your home-turf, as it were), showing how your preferred product measures more accurately communicate the performance.

Third, there is a fear of overburdening customers with too much information and overcomplicating a decision. This is largely a false-problem, and where it does apply it indicts the poor buying interface. The example that is often brought up here is from traditional retail, where, say, 30 varieties of peanut butter lead to less total sales than 3 varieties, that some fraction of peanut butter buyers were so flummoxed by the varieties on offer that they reject any purchase. This should not limit online marketplaces for a few reasons. The first is that eBay has control over product presentation and can, for instance, choose where the brand logo appears, where the price is located, the prominence of the unit cost, etc. Thus eBay can focus buyer attention on relevant characteristics, and not overload them in processing a gaggle of different label designs spread across shelves.

The online marketplace can add sorting and filtering options that are impossible in-store, allowing the user to structure the decision according to their preference. And since sorting and filtering are inherently attempts to apply order to a disordered system, they create value, and this value can be shared between customers so that the most popular sortings are the default, saving most customers time. (Indeed, the common default presentation of 'featured' and 'best-selling' is the demonstration of an anti-pattern, something intending to short-circuit the customer's preferred sorting.) These are all free features of online commerce and, rather than overburdening the customer, are actually ways to decrease customer purchase risk while also learning more about the market dynamics.

This is a very crucial point: the customer's frustration in composing search queries is mirrored by the seller's frustration in understanding why those customers eventually bought what they did. Markets are complex, it is eBay's job to help both seller and buyer understand each other and come to terms. To the degree that customers are unable to evaluate and understand product risk on Amazon, there is room to compete.

So, if I were eBay I would focus on users, explaining market dynamics and telling them how we've engineered our marketplace to maximize their satisfaction.