“The decline of physical bookshops…affects which books have a chance of breaking out : bestsellers flourish, but midlist books that might have been discovered while browsing a bookstore are worse off, because consumers cannot easily stumble upon them while shopping on the internet.” – The Future of the Book, The Economist Oct 11, ’14
“Amazon is ideal for finding the book you know you want. But it is not ideal for finding the book you don’t know you want.” – Henry Mance, Shannon Bond in their article Bridges To Span Amazon’s Dominance, FT Aug 11, ’14
“A large part of the profit in the industry comes from the mid-list – the stuff that’s at the back of the bookshop and sells reliably for years, and this risks being invisible on Amazon. The concern is that as physical book-selling goes away, more and more sales gets concentrated in best-sellers, both for print and e-books” – Benedict Evans in article by Henry Mance, Shannon Bond, Bridges To Span Amazon’s Dominance, FT Aug 11, ’14.
The Discovery Problem
Let us get to the point. Online booksellers have a discovery problem. Unlike physical bookstores where discovering books whilst browsing and buying them as you walk out the store resulted in a tight coupling of discovery and purchase, in the online world, discovery and purchase are largely decoupled. This isn’t the result of any grand designs at business design, but has arisen from the very nature of these mediums (if I may term it so).
Rarely do you come out of an online book-buying experience, having discovered a new title or author (notwithstanding Amazon’s ‘people who bought this also bought that’ widget). Mostly you go online with a clear idea of what you need to buy, and having bought that, you are content to exit the online bookstore.
In contrast, the physical store is visited as much to browse (and perhaps buy, perhaps not) as much as it is visited with the intent to buy a specific book. Discovery is also an objective for a consumer visiting a physical bookstore. What is more, offline browsing is also more efficient. Mike Shatzkin says in a very interesting post “It isn’t hard for somebody in a bookstore to look at hundreds of books in a few minutes. It is nearly impossible online.” With such a wide funnel, the probability of finding a book is higher offline.
Serendipitious discovery of a book you didn’t know you wanted, is also built into the physical bookstore experience, encouraged by the randomness of stacking in the ‘New Books Counter’, or the opportunity to stumble upon a book accidentally stacked next to your favourite writer, or in your preferred genre. Mind you, in many cases the discovery may not be the result of an accident – it may be the result of curation by an enlightened proprietor, or even the recommendation of a particularly observant assistant.
This decoupling of discovery and commerce during online shopping for books has resulted in a concentration of sales in the bestsellers list: ‘books people know they want’ – Mike Shatzkin states that online sales of bestsellers are even more concentrated than physical bookstores (5% of titles account for nearly 70% of sales online, he says; citing a conversation with an executive).
It looks like there is a fat head and a long tail, but there is no chunky middle, or rather the middle is thinning fast as online bookselling squeezes out physical bookstores. And this disappearing mid-list cited in The Economist article and also mentioned by Benedict Evans (both quoted at the start of this article) is of huge concern to Publishers. This is because the Mid-list is more important than Bestsellers when it comes to the bottomline for Publishers. And it matters for online retailers too, as they too benefit from every extra book sold via online. Clearly, discovery or the lack of it online, bothers everyone one.
Attracted by the challenge of cracking discovery, a number of startups have emerged – some prominent ones, though now beyond the startup stage, include Goodreads, Shelfari, Jellybooks. And new ones keep emerging continuously such as Bookbub, Riffle etc.
We can classify book discovery sites primarily on the basis of how the books are curated or not. Curated models include Crowd, Expert or Code.
Examples of code-curated models are algorithm such as Amazon’s “people who bought this book also bought”, whatshouldireadnext.com’s “if you like this you will also like” model, whichbook etc
Human-curated models can be curated by a Joe Nobody or an expert : an influencer, a celebrity, a well-known author etc. Pure expert-driven models are rare. Aggregation is a real challenge and these sites are not easily scalable. I could only arrive at three sites – Favobooks, Bookish and The Books Project.
On the other hand, non-expert driven models are numerous –Goodreads, Shelfari, Riffle, Librarything etc. Sometimes there are sites that use a mix of experts and non-experts such as My Independent Book Shop.
Lastly there are non-curated sites, such as Jellybooks which throw up a multitude of covers and get the prospective reader to click on one. Here the objective is slightly different. The focus for Jellybooks is sampling, while for Bookbub, it is discounted e-books.
However none of these have come anywhere close to addressing the challenge of discovery entirely. Peter Hildick-Smith, of Codex Group, a consultancy, states “Something is really, chronically missing in online retail discovery, But what might that something be? It’s not as if book buyers aren’t using online sites like Pinterest, Google and Goodreads — they are, but those sites simply aren’t converting to actual book purchases.” He continues, “Sixty-one percent of book purchases by frequent book buyers take place online, but only seven percent of those buyers said they discovered that book online, while physical book stores account for 39 percent of units sold and 20 percent of discovery share”. Why is this so?
Discovery is a particularly knotty problem – it has challenges at multiple levels for the consumer. It has broadly three components –
- The Expertise Problem – Who is recommending (or curating)? Anybody or Somebody knowledgeable?
- The Relevance Problem – Is the product relevant for me at this stage / context? (people don’t always buy for themselves or to read)
- The Trigger Problem – How do you make it easy for the consumer to translate the suggestion into action?
Let us take an example (hypothetical) of how perfect discovery could proceed – Clayton Christensen recommends a book on Innovation to say a CEO looking to set up an Innovation unit in his organization (assume he knows the CEO well and understands what the CEO’s present information needs are. Clay does this in a bookstore (where the book is available and is discounted). All three challenges – expertise, relevance and trigger are addressed. This is the ultimate example of ‘hand-selling’.
It is extremely hard for any one solution to address all three challenges perfectly. At best, two out of three can be addressed. Perhaps all three can be addressed, but it is unlikely that the solution will be scalable – the typical example here is how the proprietor of an independent bookstore recommends a book to a frequent buyer. Hardly scalable.
Let us look at the first challenge – Expertise. On a site such as Goodreads, with its zillions of bookshelves, how do you know whose recommendations to trust? Sure, the fact that you can look at your friends’ digital bookshelves do help, but given that anybody can put up a bookshelf, crowdsourcing models do not necessary lend themselves to filtering for quality.
“It is, in fact, trivially easy to discover books: Just go to any book store, online or off, and you’ll discover more books than you can shake a stick at. The problem is not one of finding books, it is one of discrimination, of deciding which books are worth reading, and it’s a problem is compounded by the vast choice on offer. – Suw Charman Anderson”
The models that do enable some degree of discrimination, such as by sourcing recommendations from experts such as My Independent Book Shop or even a blog such as Farnam Street, struggle to populate content at the scale that a Goodreads or LibraryThing can. This brings us to the key tradeoff that discovery solutions need to make : Scalability vs Expertise. A scalable expert-led model seems an impossibility at this stage.
Given the difficulty for any one solution to address these challenges equally, what we therefore need is a diversity of approaches – each trying to address one or two of the challenges and moving us closer to the goal post.
We may also see solutions emerging from entirely different approaches. Let us relook at the Expertise vs Scalability problem.
First let us relook at how we approach Expertise. The challenge for book discovery companies is in aggregating experts at a large enough scale to matter. Rather than them attempting to set up these themselves, why not look at existing platforms that aggregate experts such as LinkedIn’s Influencers or even Twitter. A young financial analyst on twitter following senior execs in his industry is conferring expert status on the people he follows. Can LinkedIn or Twitter leverage this by encouraging people to tweet out favourite books or movies (or bags or whatever) for a slice of affiliate revenue?
I don’t know if LinkedIn or Twitter will do this – monies via this route may not be meaty enough for them – but there could be opportunities for a book discovery startup to think through how they can leverage Twitter, Linkedin and other platforms that have already aggregated experts of some kind.
Second, let us take up scalability. In the relentless quest towards scalable models — which force discovery startups to look at online bookshelving / crowd-sourced or code-driven models – are we missing solutions that are perhaps not scalable but enable enhanced discovery? Here, I cant but help recall Paul Graham’s advice – do things that do not scale – which worked very well for airbnb. I will address non-scalable approaches in detail in a later post. At this stage, I do think that it is by playing around with alternate approaches to aggregating expertise and scalability that the next big breakthrough in book discovery will emerge.
Implications for Physical Bookstores
Given how critical discovery is to the book trade, and given how well-integrated discovery is with the physical bookstore experience, I have always wondered why physical bookstores have not looked at leveraging this strength.
What if a bookstore said they would charge for discovery – essentially discovery via serendipity or even charge like a museum / gallery does for a curated section in their shop? What if they got a Jonathan Ive or Dieter Rams to design a part of the shelf / shop to look like it was their library / shelf, and charged people $10 per hour. I am sure enough designers would be keen to visit the exhibit and find out which were Ive’s / Rams’ favourite books. Similiarly an executive in a venture capital co might be happy to look at what a Marc Andressen has put together. (Incidentally you can look at Marc Andreessen’s bookshelf here.) Discovery is clearly a ‘Jobs to be done‘ that a customer seeks out in a physical bookstore. Bookstores would do well to exploit this, than look at adding unrelated products such as toys and pens in their stores.
To reiterate, we could see physical bookstores as selling not just books but also discovery time and space. In the discovery time-space offering, they are selling consumers an opportunity to explore and identify books that they otherwise would not have chanced upon. This is why we like certain second-hand bookstores or bookstores like Lotus House Books (now sadly defunct, but at one time considered to be the best bookstore in Mumbai) which had a strong curated feel. On the other hand, big box booksellers such as Crossword have no surprise factors in their collections. In some senses, the future of the physical bookstore is in becoming something akin to a museum or (an even better example) an art gallery.