There shouldn't still be a market for lemons
Why I'm building Slant
The psychology of how people decide on what they buy fascinated me for years. As the resident tech nerd in my family, it was always my job to figure out what laptop or DVD player we should get.
This eventually led to spending countless hours devouring any resource on the subject I could get my hands on; from Dan Ariely’s entertaining book on behavioural economics, Predictably Irrational, on how we rarely make purely rational choices, to the “decision support systems” research at my University.
My first foray into actually building something to help people decide was a side project I worked on with a few friends while I was studying. We built a website that incorporated technology the Australian Department of Defence uses to determine what tanks or submarines to buy, but applied it to consumer goods.
The idea was to have a website for when you wanted to make a significant purchase, something like a fridge or a new TV. You could input your requirements and the website would recommend the best product for you.
It turned out that there is a slight difference between what’s needed to help someone buy an M1 Abrams and a blender…but the project got me obsessed with the problem space. It also set me on a path that would result in quitting my childhood dream job at NASA and starting my first company.
One paper in particular inspired the core problem I wanted to dedicate myself to solving.
Back in the 1970's George Akerlof wrote one of the most cited papers in modern economic theory, “The Market for Lemons”, in which he discussed the market implications of information asymmetry. The classic example of information asymmetry and its effect on buying decisions are salesmen of second hand cars — the salesmen are aware of defects, and this puts them at an advantage over buyers who have no reasonable means of knowing.
Akerlof shows that a market with information asymmetry is bad for both the consumers who end up with poor products, and also for quality sellers who are unable to differentiate their value and are forced to sell at the same price as the inferior options.
The Internet was initially heralded as the solution to information asymmetry. Even the hugely popular book Freakonomics touched on this subject:
Information is the currency of the Internet. As a medium, the Internet is brilliantly efficient at shifting information from the hand of those who have it into the hands of those who do not. Often […] the information existed but in a woefully scattered way. (In such instances, the Internet acts like a gigantic horseshoe magnet waved over an endless sea of haystacks, plucking the needle out of each one.) […] The Internet has accomplished what no consumer advocate could: it has vastly shrunk the gap between the experts and the public.
Unfortunately, the sheer volume of information disseminated on the internet overwhelms our ability to process it. Waving your ‘information magnet’ over the internet when researching a choice leaves you combing through endless amounts of medicore data. We went from information asymmetry due to lack of information, to information overload.
Looking at the Apple App Store for example, the number of apps available far exceeds the unique use-cases we use them for. A Nielsen study showed that while the quantity of available apps has exploded, the number of individual apps used by people per month remained relatively flat.
This results in us having to choose from the thousands of available apps for a given task. But how do you pick a calendar app when there are so many of them available, the majority for free and most with roughly the same star rating?
Star ratings are particularly problematic due to the context of each person rating the product being varied and unknown. Devavrat Shah, a professor at MIT’s Laboratory of Information and Decisions Systems, recently conducted research on the failings of star rating systems:
If my mood is bad today, I might give four stars, but tomorrow I’d give five stars…Your three stars might be my five stars, or vice versa. For that reason, I strongly believe that comparison is the right way to capture this. — Devavrat Shah
The App Store isn't the only flawed system we have in place. Amazon requires you read multiple reviews for every potential product and maintain the tradespace of the various pros/cons in your head. Then there are the editorialized approaches, which are typically out of date and, depending on the author’s interests and bias, might not reflect what’s right for you. Forums are a signal-to-noise ratio nightmare. Searching on Google just surfaces the editorialised pieces, forums or whatever product has the best SEO. None of these systems come close to solving the signal-to-noise ratio problem and we're left digging through a stack of browser tabs.
Another common fault of these systems is that they are focused on products in isolation, not contextualized to the problem the consumer is trying to solve. A programmer looking to buy a new laptop wants to see the information through that lens (“does it run Linux?”), not reviews on individual laptops that don't consider the overall tradespace and the consumer’s use-case.
An unfortunate side-effect of these problems are the most popular products rarely being the best of that class, but instead those with the best marketing. Every advertisement is a nugget of information asymmetry. When’s the last time you saw an ad explaining that for some people the competitor’s product is actually superior? Marketers and salesmen have typically been the mortal enemy of anyone who wants to make an informed choice. Even the nicest and most moral Samsung marketer won’t go out of their way to point out that Nexus devices get OS updates earlier.
So despite all these efforts and the amazing utility of the internet, there is still a market for lemons when there really shouldn't be.
Six years after I first started thinking about how people decide on products I'm at it again with a platform called Slant. We're building the trusted pro-consumer resource that people will use to find the best products for them.
Here are the core philosophies we have built the product around.
Qualitative vs quantitative data
That “slight difference” in what’s needed to help someone pick between a M1 Abrams and a blender, that tanked my first product, turned out to be qualitative data. Comparison websites let you see price or the CPU speed side-by-side for the products you're interested in. This however doesn't come close to covering the data people actually care about. When’s the last time someone told you they love their iPhone 6 due to its 2GHz dual-core 20nm 64-bit A8 CPU?
Quantitative data rarely translates into consumer experiences, so it’s important to capture the qualitative data as well. This is currently done in long lists of plain-text user reviews, which is a big contributor to the poor signal-to-noise of the current systems.
Structured and trustable subjectivity
To solve the signal to noise issue of qualitative data, it needs to be structured into a more compact format. By removing redundant data and abstracting it into a structured format, readers can more efficiently process what was once an endless list of free-text reviews.
This has massive potential. Wikipedia is a “collective human knowledge system” that structures objective knowledge to build arguably the most important resource on the internet. There is a huge opportunity to structure our subjective knowledge to help people make decisions.
The way we tackled this problem at Slant was to structure the data like this:
The claim “OS X is developer-friendly” is put into a format with the claim, detailed information backing up that claim, people that support the claim and trusted third-party sources that support the claim.
Structuring the data in this way solves the signal-to-noise ratio problem with user reviews as it removes redundant text and you can abstract the information to the titles to allow quick skimming.
It also solves another problem with qualitative data in that it’s inherently subjective, so being able to enhance trust by incorporating sources and people that back up the claim is important.
Another exciting thing about data in this format is it enables a mobile experience that long reviews and forums cannot match. Imagine being able to ask Siri what you should buy or being able to quickly check if a product in a store is the best choice.
Data contextualised to the consumer’s problem
This data only becomes truly useful when put into the right context for the person trying to make the decision. Different use-cases have different requirements which impact what products are in scope and what characteristics of those products are relevant.
Q&A is a great way of capturing context and scope. You can ask a question like “What are the best laptops for programmers under $1,500?” and the resulting comparison is with that context and scope in consideration.
Combining the question with the structured qualitative data, presented in the form of pros/cons, form the tradespace of why you should (or shouldn't) pick a certain option for a given use-case.
What this means is that you end up with information specifically relevant to your problem, not general information about each product in isolation. This is a massive time saver. It’s ridiculous that every single person currently has to go through the same pain of researching from scratch just to figure out what one product does and the other doesn’t.
All of this adds up to a platform that has high signal to noise and lets you quickly research what are the best products for you and why. Here are some example Slant questions to check out:
Historically, marketplaces online have competed over price, payments, fulfilment or product availability. Over time, an increasing number of these characteristics have become commoditised. In the case of payments companies like Paypal, Wepay and Stripe have made accepting secure payments easy for any marketplace.
The future of enabling commerce won’t be dependent on fulfilment or breadth of products. Just like payments, the ability to get a product to the doorstep of the consumer will become commoditised. Amazon even offers “Fulfillment by Amazon” that lets anyone do global product fulfillment.
The key will be owning the consumer experience around finding and deciding on the products to buy in the first place. The big winner in the future of e-commerce will be whoever nails that experience.
The current version of Slant barely scratches the surface of what we eventually want to build. We still have a long way to go and a lot of challenges to solve before we can claim to effectively inform people about the best product for them. However, I believe this is something truly needed and I'm incredibly excited at the opportunity to play a small part helping the Internet live up to its potential to solve information asymmetry. Lets take lemons off the market for good.
If you're interested in helping us make the world a more informed place we're hiring! Email me at stuart@slant.co
