Price Comparison Engine Case Study
Glenn Fleishman, Unsolicited Pundit, writes:
Per your column on the brilliant Doc Searls, one of my favorite writers and thinkers: I run ISBN.nu .
I list the prices of nine online bookstores for any given ISBN, plus a link to Alibris.com to search by title/author. Given that
it surprises me that Amazon.com comprises about 30% of my gross revenue from affiliates. Part of this is that Amazon.com continues to offer a 15% commission on many items directly clicked through and purchased titles, and 5% on a variety of other items and on other items purchased from a basket started through an affiliate clickthrough.
- Amazon.com often has the highest price for a book, and often charges full price or a short discount on titles that are discounted 20% to 40% by other booksellers,
- Amazon.com has raised its shipping prices to $4.29 for a single book (from $3.95, which is approximately what all its competitors charge),
- stores like Barnesandnoble.com and Borders.com should have a reasonable amount of market prominence, and,
- people come to my site to perform a price comparison, meaning they are value not brand shoppers (ostensibly),
If price were the only factor, Booksamillion.com would be the leader based on their price. Their site offers reasonable performance and design, and their shipping and order tracking is quite decent.
When I added Varsitybooks.com a few months ago, I noticed that I was getting a tiny number of sales from their links despite the fact that they were often cheaper than all other stores placing them at the top of the results from my site. However, three factors led me to drop them from the site:
Recently, I was emailed a proposal by a guy who I think had a relationship with a bookstore I don't list. His proposal, stated pretty ineptly at first, was that I hook up with him to broker a relationship with this other bookstore to make them the "premium" store on my site. And that no matter what price I found for a book at any other site, I would program my system to show this bookstore with that price at the top - they would match the lowest price.
- customer complaints: my users kept emailing them that during the checkout process they were asked about their college affiliation. Since my site attracts a wide demographic among the thousands of daily users and VarsityBooks.com was (originally) aimed at college students, this was confusing and irritating to them, causing them to abandon their shopping baskets.
- branding: people don't know who VarsityBooks.com is, so they don't order from them.
- loss of sales: the abandoned shopping basket problem and the perception that I was pushing a little-known store (because they showed up at the top) led users to stop buying from me. I could see the dropoff in sales over the few weeks I used VarsityBooks.com.
I declined, after a number of back and forths in which I was more vociferous in my rejection of their concept. It's obvious to me that this approach would fail. The lowest price coupled with an unknown or little known brand equals little or no sales.
Amazon.com has now reached a point where they can "afford" to charge more. I noticed that they claim a 10% profit on shipping last quarter, the same quarter in which they raised their base book shipping price by about 8.5%. Odd coincidence, eh?
Jakob's reply: This is a great example of why a price comparison engine is not enough to help people shop. Not to say that ISBN.nu is not offering a great service, but it would be better if it were matched with a reputation manager that could annotate each price quote with a rating of customer satisfaction with that store.
People will buy from a more expensive fulfillment option as long as they fear the quality of the cheaper alternatives. Once it becomes possible to know which of the alternatives offer good service (ship on time, take returns if needed, and so on) we will get a more level playing field. Many users will indeed want to pay extra for better service, but they won't pay extra if all they get is the same service as a cheaper store.
I have a similar example from my own experience:
The page for my recent Web usability book links to two fulfillment channels where people can get the book: Amazon and FatBrain. Even though FatBrain offers a 10% deeper discount than Amazon, most users ask to get their book shipped from Amazon.
This can be very rational behavior on the part of the users:
- Assume that it takes 5 minutes to figure out an unfamiliar user interface and to complete the additional registration information required the first time you order from a new vendor.
- Furthermore, assume that the user's time is worth at least $60 per hour (either counted as their billable rate or as twice their hourly salary, taking overhead expenses into account).
- Under these assumptions, it will cost the user (or rather, their employer) at least $5 if they order from a new place.
- This is not worth doing to save $4.50.
Now, if it only took one minute to complete the transaction on the cheaper site, users should rationally go there. But how are you to know in advance how fast or slow it is to use a new site? Users have been burned by poor usability many times in the past, so they will fear the worst. In the future, the reputation manager will supply the answer: before you follow a link to any e-commerce site, your browser will connect to the reputation manager and display the average time it takes users to buy things at the destination site.
Branding at a Different Level
Ed Blachman writes:
You've been talking about reputation managers for some years, and here you claim they'll replace branding. I don't see how that will work. Maybe I'll move my trust from "the 'store' I've heard of" to "the 'store' that gets high marks from
In your Sept/5/99 Alertbox on rep managers , you mention six: eBay's reputation ratings, Epinions, Google, Go, Slashdot and Third Voice. To that list, I'd add four more that I know of: Gomez.com (ratings of a variety of ecommerce sites), About.com (guides, seemingly like Go), Deja.com (consumer ratings and reviews) and Consumer Reports. Out of all of these, the only one I truly trust is Consumer Reports. Why? It goes back, I think, to branding (of the rep manager, in this case):
I put a very high level of trust in Consumer Reports. For products where its model works well (typically products that change on yearly or longer cycles), I give its ratings and reviews a lot of weight; I can't imagine a reputation manager in any field I'd trust more. Nonetheless the advent and prominence of Consumer Reports has not eliminated (product) branding by any stretch of the imagination. Why do you think that online reputation managers will do better?
- Consumer Reports trumpets its independence and backs it up loudly and visibly.
- Consumer Reports aggregates user experience in a way that makes it relatively difficult to spoof (you have to subscribe to the magazine, which costs real money, in order to be solicited for your opinion).
- Consumer Reports backs up its user experience aggregation with serious testing by serious experts, and they are open about all aspects of their methodology.
Jakob's reply: I agree. The reputation managers really need a good brand since you are going to place a lot of trust in them.
I think, though, that there is quite a difference because of the number of services that have to be evaluated. For the Web as a whole, there are so many sites that most of them simply cannot get a sufficiently strong brand to live on that alone. VCs are wasting billions on this account, but it is doomed in most cases. The human brain only holds so many concepts and we have to tune out most of the things in the world. On the other hand, there will probably be no more than a handful of big reputation managers.
It is true that Consumer Reports has not eliminated branding in the physical world. But this one magazine still has substantial impact. For example, there are a lot of car brands I would never even look at because of their poor reliability ratings in Consumer Reports. And the last car I bought (Mercedes E420), I selected because Consumer Reports wrote that it was the best car they had ever tested.
For other products such as toilet paper (which they recently reviewed), it is more difficult to follow the advice from Consumer Reports. Who can remember the names of 50 types of toilet paper and their relative ratings when you are standing at the shelf in the supermarket?
In contrast, a computerized reputation manager will always be with you while you are on the Web and you will not have to remember to look up things in it because it will automatically judge all the links and advertisements on the current page.
Who are "Most Users"?
Joseph from The New House writes:
I would watch your use of the phrase "most people" in the abovementioned page. I suspect your experience of web users is nonrepresentative of the globe at large, including all users from the school child to the retired policeman in Kenya, or wherever.
To say for example that "Most people already have their credit card number and shipping info on file at Amazon" is a distorted view of who is using the web. Technically speaking, you don't even limit your scope to Web users, and it will be a frightening day for our economy indeed when Amazon has most people's credit card and shipping info on file.
Just thought you should be careful in establishing a sense of the true spectrum of web users. True most of your readers are not reaching out to eight year olds as target audience, but some might be. And certainly accuracy is important regardless of marketing goals.
Jakob's reply: Good point. It should really say something like "most people who are currently shopping on the Web". Or maybe even "most people who are currently buying English-language books online." Amazon does have a lot of international users, but they have been slow at opening local sites in other countries (currently only having sites in Britain, France, and Germany).
However, from a practical perspective, when you consider how to sell English-language books online, your main area of concern is people who are buying such books.