Launching & Funding an Amazon-proof Startup

This is an edited version of a longer post I wrote recently. The complete version 'In the Shadow of Amazon' is available on my website.

Phase 1 

It is now just over 20 years since Amazon had its IPO, setting it off on the path to e-commerce dominance and well over $500b in market cap. In these 20 years, Amazon, and its peers expanded across categories, from books and electronics to fashion and even grocery. And in parallel, a ruthless focus on shaving off delivery times and cost ensued, helping reduce buying friction and making online buying even more attractive to the average consumer.

I like to think of these 20-odd years since Amazon launched as the 1st phase of Online Retail. Amazon won the 1st phase handily, reshaping retail in the process. With this reshaped retail however come entirely new challenges for Amazon and its online peers. And in this 2nd phase of online retail, the rules for winning are somewhat different from what they were before.

And these new rules now make it possible for a whole new set of players such as Stitch Fix and Wish to emerge and attempt to shape this 2nd phase. So who are these emergent players? And why are they succeeding? And more importantly what lessons do they hold for the next wave of startups being conceived? 

By attempting an answer to these questions, we can help offer a startup entering the e-commerce space some glimpses of a strategy on competing effectively with Amazon? Similarly they could afford a VC exploring e-commerce opportunities a better handle on evaluating them. 

Phase 2

In Phase 1, we saw Amazon and its peers, driving improvement on three axes – expanding the list of their offerings, reducing time for delivery and finally reducing purchase friction. They succeeded brilliantly, and as a result we have the ‘Everything Store’ that can deliver in a day via 1-click shopping (and other friction-free offerings). 

Benedict Evans, a Partner at VC fund A16Z, had a post sometime ago where he described Google as a vast machine learning engine, chewing on projects and spitting these out, if it did not suit its “automation + machine learning model”, much the same as a Great White Shark chewing a boat or a dinghy to see if was edible. This is actually a good analogy to understand Amazon as well. Its online operations, from sourcing to delivery, are essentially a set of math models, which it is looking to refine and optimize continuously.

Thanks to these well-honed optimizing models, Amazon’s sweet spot is around sourcing and delivering commoditized products cheaply and quickly. If you know what you want to buy, Amazon can source it from manufacturers and deliver it to you cheaper and faster than anyone else. But if even one of the phrases from the above line break, then its famed optimization models cannot be blindly employed. Let us understand this better.

What happens when customers who don’t know what they wish to buy approach Amazon? Or if speed of delivery is not an issue? Or perhaps do not wish to buy, but rent? Or even wish to buy secondhand goods. In all these cases, the Amazon engine is sub-optimal. These conditions are a challenge for Amazon, and an opportunity for a startup, or two.

Thus in Phase 2, we are beginning to see new models such as that of Stitch Fix, subscription packs with returnable products that have pioneered a new format of ecommerce that I like to refer to as shipping-then-shopping, as well as startups such as Moda Operandi and Poshmark, where online buying is only subset of an overall experience. Not to forget the Vertical e-commerce plays such as Carvana, Houzz etc with their unique takes on the online retail experience.

Amazon-proof startups

One easy way to better understand the kind of online plays thriving against Amazon is to look at the list of highly funded ecommerce players. Sure, not all of them necessarily compete with Amazon, but many do, and thus this list of highly funded ventures gives us a sense of these Amazon-proof startups. This is thus a good place to begin.

Let us stick with the U.S. for now. There are just over 20 companies that have received over $100m in funding over the past 5 years in the U.S. There are 2 ecommerce enablers (BigCommerce and Magento), which we can remove from the list. Of the rest, these there is only 1 company, Jet.com (acquired by Walmart) that may be considered similar to Amazon, in that it does general purpose ecommerce. The rest fall across various sub-themes in e-commerce, all of which have one thing in common. They are themes that are difficult, though not impossible, for Amazon to execute on, given its DNA.

To elaborate, let us look at what happens when consumers don’t know what they want? Curation and recommendations become critical to the process, both alien to the way Amazon works.

When you don’t know what you want to buy

Subscription boxes across cosmetics (Ipsy) and fashion (Stitch Fix, Techstyle) have emerged precisely to cater to customers who don’t know what they want to buy, but are happy to get a company to do it for them. Ipsy, Stitch Fix all have human intelligence in the form of stylists who recommend products. Stitch Fix in fact has 3,500 of them and combined their taste and intelligence with past data to create a powerful recommendation engine. It is a distinctive and fascinating business model that is worth going in to in some detail.

Stitch Fix’s recommendation engine sends 5 or so garments monthly to its subscribers. If they don’t like any of the outfits they can send them back. This helps Stitch Fix iterate on and improve the recommendation process. This is thus a shipping-then-shopping model as opposed to Amazon’s shopping-then-shipping model. Stitch Fix closed 2016 at revenues of $730m and has been profitable for the past 3 years. Inspired by Stitch Fix, Amazon launched Prime Wardrobe, its try before you buy service. But I am not entirely sure it will succeed.

Stitch Fix’s use of human stylists, its enhanced recommendation engine and its distinctive try and then buy model all form a tightly-integrated set of decisions. Amazon on the other hand has an entirely different model with minimal human inputs, entirely ML-led personalization, focus on commodity products and extracting efficiencies, and thus a shopping-then-shipping model. Amazon has successfully rolled out private labels focused around basics (commoditized products such as White T-shirts, innerwear etc), but I am not sure it will be easy for Amazon to transform itself to compete with Stitch Fix, a fashion and taste focused e-commerce co.

Other themes and businesses

Of the other themes and businesses, I found Wish particularly interesting in that it targets consumers willing to wait a long time – for ultra-cheap chinese goods, shipped from Shenzhen. There is a kind of deliberate inconvenience built into their business model as is for Boxed, where the customer needs to order in large quantities, in order to unlock the lower prices they desire. A similar inconvenience-oriented business is India’s Milkbasket, which only delivers groceries and other smaller consumer goods in the morning between 5-7am.

These inconveniences need to be seen as features, not bugs. They result in a distinctive business model driving logistics costs well below that of a general e-commerce player. Bezos has built Amazon to deliver goods fast through the day in any volumes the customer desires, at the lowest cost. But when these competencies cease to matter – when customers are happy to wait, or buy in ultra-large quantities or at inconvenient times (say early morning), Amazon loses its existing advantages and is forced to compete on level ground.

Also away from Amazon’s comfort point are business models such as secondhand goods (Carvana, Thred Up), renting (Rent the Runway), precommerce (booking luxury fashion in advance as in Moda Operandi) etc. It isn’t that Amazon can’t get into these spaces. They will possibly take a crack at it one day. It is just that the way it has been built – to go after commoditized spaces with low prices and super-efficient logistics makes it difficult to target opportunities that need curation, don’t need quick deliveries, renting etc. Each business model needs an entirely different DNA.

How should Entrepreneurs & VCs pick spaces?

What kind of ideas will work for companies going up against Amazon? Or even better what are the Amazon-proof spaces that exist today that Entrepreneurs can enter safely?

Sriram Krishnan, a Silicon Valley executive had a very interesting post sometime back titled “Building something no one else can measure”. It expanded on a tweet where he described competing with major platform companies by optimizing a metric they can’t measure. I would hazard a corollary to this: compete on the opposite of a metric that the platform is optimizing for. In the case of Amazon, then you would not compete on range, price or delivery time.

To explain this better, let us take the first theme – offering something that the customer didn’t know they wanted to buy – here you are optimizing for better curation, i.e., limited SKUs which is the opposite of what Amazon wants. As ‘The Everything Store’ Amazon has to have / offer every single SKU in the universe. Meanwhile a curated store can just offer 10,000 interesting SKUs and be a relevant proposition. In fact one may well argue the better the curation the smaller the choice set or SKUs it should offer. This explains why vertical players such as Houzz, Warby Parker, Yoox Net a Porter have done well.

Curation itself comes in various flavours – expert-driven ones such as in Houzz and Techstyle (typically associated with vertical plays) or value-driven curation such as in Honest Co or Thrive Market.

Another good example is prescheduled deliveries such as by Stitch Fix or Techstyle, competing at the opposite end of Amazon, which is optimizing for speed of delivery. Prescheduled or subscription products also make pricing comparisons irrelevant – you don’t know what you are getting, so how do you even check the prices – hitting Amazon which is optimizing for the lowest price. Quite cleverly, this takes away the competitive advantage Amazon has in logistics and prices, bringing both to level playing ground.

In a recent article, Alex Evans, a VC, analyzed potential second order effects as ecommerce expanded. One consequence he argues is that as ecommerce becomes frictionless, it will become easier and easier to rent instead of buying. Netflix streaming and Uber hiring have replaced owning movies and cars respectively. He says “the same principle will hold in commerce: the cheaper and more frictionless access becomes, the less stuff we’ll need to own”.

We can already this happening in the fashion space with businesses such as Rent The Runway and Bag, Borrow Or Steal. Interestingly the option of renting as opposed to owning also sparks experimentation. You become open to renting out that adventurous dress for that one-off occasion, whereas earlier you hesitated from buying it because you wouldn’t use it anymore. I wonder if this can happen in India, say with sarees or wedding lehengas. It will also be interesting to see the rental model expand beyond fashion and enter other spaces. Co-owning (similar to timeshares or with private planes, as in NetJets) could be another business model that could emerge in addition to renting.

Lastly, you can look at building in more and more service elements in to your offering. A good example is Carvana or Vroom, both secondhand online car sellers, each with their own distinctive bells and whistles, e.g., Carvana operates gigantic car vending machines where prospective buyers can borrow a car and drive it around for a week before deciding to buy it. There is a huge service component to this business, as well as the need for intermediation between seller and buyer, moving it away from Amazon’s sweet spot.

Thoughts? Feedback? Criticism? Would love to hear back!