From: Elizabeth Childs (echilds@linex.com)
Date: Wed Dec 22 1999 - 02:38:43 MST
Eliezer Yudkowsky said:
> Selling at a loss to build market share? What market share? Customers
> can move from one seller to another instantaneously, the instant prices
> go up. In an inertialess market, current market share has no causal
> connection to tomorrow's market share.
While I agree with the thrust of your post, I have to point out four things
the current ecommerce ventures are building that are lasting:
1) Data. Companies are building vast databases on individuals that allow
them to do very precise target marketing. In the long run, this will be
enormously valuable. The portal companies could be taking much greater
advantage of this to do automatic customization of sites. For example, most
of the search services show current headlines on their front page. They
could customize those headlines based on previous search results. Even with
the minimal AI-like experience you can get from collaborative filtering,
this can create a very powerful user experience - with better AI on the
backend, this would be the difference between interacting with an entity
that "knows" you versus an entity that doesn't know you.
2) Branding. Why do people drink Coke? Is it "I like the taste" or is it
the years and years they've been getting marketed to drink Coke instead of
generic Cola? How many people have even tried the generic Cola? People
tend to stick with what they know. I agree that this tendency will be
rapidly eroded in the New Economy, but it won't disappear altogether. Also,
sites have learning curves. I know where things are on Yahoo, so I continue
using Yahoo.
People can make a rational choice to pay more to deal with a brand they know
and trust. And with so many new sites competing for attention, the fact
that consumers can even remember the name "yahoo" counts for something. If
the next big thing is the perfect shopping robot, then someone will build
the perfect shopping robot and Yahoo will buy and call it "Yahoo". It's a
lot easier than teaching someone about a whole new word. Teaching consumers
a new word costs tens of millions if not hundreds of millions of dollars.
3) Experience. A company that's been doing something for a long time
becomes a smarter organization than one that's brand new. Of course, the
opposite happens - companies get too big and bureaucratic - but a company is
a group of people that has been working together for a while to solve the
same problem. The longer they keep working at it, the better they get at
solving it. Even if the specific people leave, the company develops
institutions, procedures, and technologies that make the organization
smarter. Thus, when you invest in Amazon - not that I'm recommending it -
you are betting that the Amazon team can innovate faster at attracting and
satisfying customers than a new team that hasn't entered the race yet.
Having seen a company go from 5 to 400 in 3 years flat, I can really vouch
for the importance of this. Every day, 400 people are shaving away at the
problems my company faces. You can come up with the capital to hire 400
people with the right qualifications on paper, but that won't give them the
experience we've got.
4) The best e-commerce sites have excellent back-end distribution networks,
which are a lot harder to build than web sites. I order almost everything
online, and it's very clear to me who has a good back-end network and who
doesn't. Shopping.com, for example, has a really messed up distribution
network, and it has wound up with a terrible rating from Bizrate as a
result.
Good distribution builds customer loyalty and also saves the company money.
Webvan cost billions of dollars to build because (I am guessing) that's how
much it costs to put together a decent distribution system of that level of
sophistication - hiring drivers, scheduling them, buying paper bags in bulk
from the right discounter, knowing the OSHA laws applicable to trucks with
freezers in them, mastering the food safety regulation laws, etc. etc.
Remember, when it comes to e-commerce, it's not all about price. Other
important factors are trust of the site, usability, quick shipping, fast
delivery, having good editorial content that helps people find things they
want to buy, customer service, having good recommendation bots, and so on.
An engineer's approach is to decide what he wants and to find the cheapest
place, and that's fine - I shop like that for some things, it's a perfectly
good approach. And some of the things that people want can be handled with
collaborative filtering - if a company will lose your order, you'll get a
warning from your shopbot.
But most people don't shop like engineers. Shopping is also about theater
and status and hunter/gatherer instincts and finding something you didn't
even know existed. Even if you think everyone *should* shop like engineers,
that doesn't mean they *will* shop like engineers. And that guarantees that
a market in shopping sites will continue beyond the birth of the perfect
shopping robot.
That said, I agree that it will be very hard for the current market leaders
to maintain their status over time, and there have been a few losers
already. I would be very cautious about investing in most of the internet
companies. But I do think that current market leaders have some big
advantages.
All of this assumes human beings much like the human beings of today. I'm
not sure what the shopping habits of an AI would look like, or an SI, and I
feel thoroughly unqualified to speculate.
This archive was generated by hypermail 2.1.5 : Fri Nov 01 2002 - 15:06:10 MST