I am interested in putting together a "personals" website. I would
like to have an educated guess on the following: initial
hardware/sofware costs, development time, types of hardware needed,
initial and on-going labor costs (what type of team would I need),
support costs etc. Base this on a low cost, low end site, middle of
the road, and something like the top end which apparently had around
400,000 paid subscribers 4th quarter last year and who knows how many
visitors. All other insights and opinions would be greatly appreciated
including but definitely not limited to an opinion of the
viability/profitability of this idea.

Thanks,
Joel

Thanks,
Joel

Dear Joel,

Thanks for contacting Google Answers!

As other researchers have stated, there are several different options
for providing a website.  I don't agree that we can't provide a good
idea of what you might spend on such a task.  Let's go over these
requests one by one and then hit a summary.

We know that you will need to store information on personals
advertisements, that you will probably need to keep information
separate between users (e.g., giving each user a login name and
password, and keeping certain information private), and perhaps
providing messaging between users.  These are what I would consider
the core capabilities of a personals site.  Obviously you would want
to be able to post photos, etc.

Depending on the traffic and capabilities, there are several different
ways you can accomplish this service, as researcher arunsahlam
mentions.  You may use a service called "managed hosting", whereby you
essentially rent space and services from a provider.  You do not get
or own a computer with this method, you are simply using resources
which can be "shared" or "dedicated".  Shared resources means that
several people may be using the same machine to provide web service,
dedicated means it is you alone (but there is not always truth in
advertising).  This service can be as inexpensive as $10 a month or as
expensive as $400, depending on the type of service, bandwidth
requirements, etc.  These are not things we can adequately assess,
since we do not know how graphics heavy your site is.  Many hosting
providers provide database software and some "integration" software,
which serves as a framework so you can program your database.

You may also "collocate" a computer at an existing provider, or build
your own data center space.  If you expect high traffic and want to
perform services outside of a basic website, this is probably the way
to go.  These costs can vary widely; generally a company may charge
you power costs, bandwidth, and cooling.  These can run into the
hundreds per month and are rarely below about $200 a month.  Once you
have this computer or computer(s) (which you generally must purchase),
you need to install software on it.  The following software is
required:

   * Operating system (Windows, Mac OS or variants, Linux or variants,
or other UNIX).  Cost varies between systems.  You can get away with
this for free.  Most managed hosting companies use Linux or FreeBSD.
   * Web server software (depends on OS,
http://www.netcraft.com/survey/ has a list).  Apache is generally
regarded as one of the faster solutions and has a significant user
base and competency among engineers.  Microsoft's IIS probably also
ranks highly.  Most managed hosting companies use Apache (which is
free).  The web server does not, except in very special cases, affect
your ability to provide service or features.
   * "Glue" software - A programming language you use to connect the
web server and the database.  Perl or PHP are the common languages for
database connectivity nowadays, and they are most commonly associated
with UNIX solutions but also work on Windows systems.  (Free)
   * Database software - This also varies widely.  PostgreSQL and
MySQL are common variants; you may also go with an enterprise-class
system like Oracle or Microsoft, which can be thousands of dollars
depending on the size of your database.  PostgreSQL and MySQL are
free.  The database server does not, except in very special cases,
affect your ability to provide service or features.  It may, however,
have impact for reliability, redundancy, or health under load.

The choice of each of these above items is usually dictated by a
mixture of personal preference, "intellectual capital"/familiarity of
your staff with the product, and support needs (the free/open-source
stuff usually is harder to get commercial support for, if not
impossible, but applications of this type have generated a
not-insignificant number of professional services companies who
perform these tasks).

If you are going to purchase dedicated hardware, you can probably get
away with a smaller webserver for less than $1500.  Companies today
tend to prefer arrays of smaller webservers as opposed to single large
webservers, but this largely depends on application.  IBM, Dell, and
other major manufacturers have several on-line pricing tools capable
of delivering an accurate price.  If you collocate your server at an
ISP or build your own data center, you should expect your equipment to
be housed in racks or placed on shelves which are mounted in racks. 
Racks are divided by "rack unit", most racks have 42 Units of space
and servers can be as small as 1 rack unit (1U).  A 1U web server can
house quite a large amount of data and service quite a number of
requests. With multiple 1U servers you may need a Server Load Balancer
(SLB), which are available from companies such as Foundry, Cisco, F5,
and alteon (now Nortel). There are no hard numbers, though,
unfortunately.

As far as development time, it depends on the number of staff you
have.  You have "web development" to consider, which is basically your
graphics and layout and text of your site, then "glue" development,
which would be your database and the glue language, and overall
architecture, which is what components go on what server.  Many small
websites put all this stuff on the same machine, and have one or two
developers who can put together a reasonable database-oriented site in
two to three days, maximum.  However, an optimally-architected
solution (highly scalable, multiple backed up databases, tied in with
instant messaging and other solutions) could take up to 45 days to
develop (approximately 1000 hours).  Costs vary widely based on the
skillset of the engineer.  Rough numbers would be:

    [Type of development] - [Small/Mid/Large in number of hours]
    Graphics/web development/text/layout - 8/40/80
    Glue development - 16/40/160

Again, I repeat:  All of this varies very widely.  Some engineers can
code a lot faster than others or have more experience, so they can do
these things much quicker (but might cost more).  I have developed
similar sites in less than a week on the low end.

As far as on-going labor costs, in an ideal world a site like this
more or less runs itself except for hardware failures or engineering
difficulties. A well-architected solution should provide you a good
bit of documentation and an easy way to fix things if they are broken
(such as an automated repair function).  You should know exactly what
you want before you hire your staff.

Speaking of hiring your staff, your type of team should contain either
individuals who have a wide base of knowledge in several different
areas, or individuals who are highly skilled in specific areas. 
Generally, at least for the lower-end stuff, engineers will usually
have experience in all of these areas at once.  For the higher-end
stuff, you would want a more specialized team.  I would say that the
costs add up as follows, then.  I've rated these by points, since
specific numbers aren't available.

   PostgreSQL or MySQL database: +1 points
   Microsoft or Oracle database; +5 points
   Microsoft or Macintosh operating system: +3 points
   Linux or FreeBSD operating system: +2 points
   Own equipment: +5 points
   Dedicated managed hosting: +3 points

If you go with single managed hosting from a company like Dreamhost or
Rackspace, you may find that the database, equipment, OS, etc. are all
included in your price (I pay $29.95 a month!)  The cost I incurred
outside of this was just the time it spent to write the code.  The
code basically consists of adding, modifying, or deleting entries in
the database, sending messages to other people and storing them over
time, and logging people in or registering them for the site.  None of
these are particularly hard for someone who knows databases (and it
stands to reason that if they know databases on a UNIX system they
probably also know how to access them using a glue language).

Here is what I would do for a low-end site:

Monthly managed hosting - $29.95, +$50 setup on average
Database possibly included in base price
Glue language - One engineer 1-2 weeks, 80 hours of time @ $45 an hour
- $3600
Disk space (Usually very cheap, but depends on provider)
Usage charges for database (Also very cheap)
Bandwidth charges

Providers usually include a significant amount of the last three
"varied" charges in the base amount you pay them.

For a mid-size site:

Dedicated hosting $350/month
Database installed and configured by engineer, use free database
Architecture - Three engineers 4 weeks, $45/hour average, $21,600
Disk space included in dedicated price (you own it)
Usage charges for database free, you own it
Bandwidth charges as necessary

Guessing at the larger site all comes down to what the amount of
subscribers/visitors are "per day".  Your large site's number of
personals does not indicate how much adding/changing/deleting is going
on, or how much viewing of entries are.  Viewing of entries takes CPU
power, memory, and disk.

For both of the above potential solutions you will need a web designer
to put the pages together if you don't already have a plan in mind,
and probably a graphics artist to do the graphics.  These very widely,
i've seen students do it for $25 an hour and high end folks do it for
$200.  The quality varies widely, but generally I find that people I
pay $35 an hour to can do it equal or better to the higher end folks.

Interesting to note:  Match.COM runs IIS on Windows 2000, craigslist
runs Apache on Linux.

Viability/profitability - There are quite a few sites out there who do
this task.  Match.com, www.matchmaker.com, www.craigslist.com are all
good examples.  I think that you can make some money at doing this,
since investing in people is always good capital.  However, the
pitfalls of running any Internet business are high right now.  But I
don't think there's any question about your target market.  The thing
is, you may be competing with places that do personals for free and
sell "advertising", and if your personals cost money you may be in
some fierce competition.  These are questions for a business analysis
firm, though - to give you direct market data on the online personals
world would be difficult.  However, there are some interesting
references:

http://www.ecmag.net/Magazine/Columns/money12_01.html talks about
Rufus Griscom's business Nerve.com, which last year generated $500K in
revenues (supposedly on track to do $3 million in 2002, requires 7
FTEs to manage),

http://www.msnbc.com/news/809823.asp talks about online personals in
detail, and discusses unique users since July 2002,

and a uDate market profile, from Yahoo:
http://biz.yahoo.com/p/u/udat.ob.html.

All of these seem to indicate the market may be quite a good one to
enter right now (i'd have loved to make $18 million in revenue last
year)

Good luck in your search, Joel.  Thanks for contacting Google answers!

My search methodology was using information i've learned over the
years in my work as a researcher in this field, and searches on google
using variations of 'online personals market'.

Best,
arimathea-ga

---

Clarification of Answer by arimathea-ga on 17 Oct 2002 18:22 PDT

Joel,

I noticed that you asked another question of a similar type, where you
write:

>I asked this question earlier with costs of a low-end, middle of the
>road, and larger type site. Based on my previous question I was
>shocked at the costs given. I am thinking much bigger than I
indicated
>previously, more on the scale of a match.com, udate, friendfinder, or
>matchmaker. I have no idea how many visitors I would have in the
>beginning but I will need the ability to grow quickly. I believe the
>largest personals site had about 400,000 subscribers in Q4 2001. How
>long should I expect to give the service away (I assume I'll have to
>in order to get people to join) before trying to charge. All other
>insights and opinions would be greatly appreciated since there must
be
>costs that I haven't contemplated. Finally I would like a brief
>evaluation of adding languages other than English as well as the
>ability to pay in multiple currencies. Please give your answer in a
>way that a technical idiot (me) could understand. [snip]

I was concerned that my answer did not meet your needs, so I wanted to
expound somewhat further.

You say here that you were shocked at the costs - I take that to mean
you thought it would be much more than I said it would be.  I want to
reassert again that once your architecture is completed to a high
standard, adding additional machines or capacity will of course add to
the cost but that the core design work you do initially will not
change.

Some of the sites you mention, despite bringing in a huge amount of
revenue, really have very little in the way of development costs to
start up with and they add engineers as needed.  The top-end style
sites really are functionally the same as many of the lower-level
sites; the extra cost comes about in hardware and bandwidth which can
vary so widely.

We find in similar businesses that a website architecture of this sort
is not the overriding cost; your most expensive cost over the course
of the year is the monthly recurring for bandwidth and/or hardware
purchases as you gain in popularity.

A site of the type I describe below can handle a number of visitors
you would be very surprised at (in a positive way).  Good luck in all
your research:  I was just concerned I didn't help you out in the way
you expected and I wanted your service to be good.

Best,
arimathea-ga

there are three ways of web hosting

1) Hire computer space from webhosting service provider and place
information that u would like to share on that rented space.

2)co-locate your computer with ISP(internet service provider). Here
the cost would be cost of equipments that you place with service
provider, cost of its monitoring for performance, security, disater
recovery etc and cost of connectivity.

3) Creating your own data center. Here you own both connectivity and
computer hardware. this involves lots of issues as you have to
maintain everything.

Cost factor varies from total work involvement, kind of work. Web page
is just an interface(GUI) but what do u want to have on backends. Is
it data intensive? Is it network intensive?

Nothing much can be said about the cost of system without knowing much
details about it.

hope it helps.
arun

The least expensive option (if it is a high priority) would be to do
it all yourself, while re-using your current hardware. This is easier
then you think (for a site expecting around 1000 visitors a day or
about 100,000 requests).

Windows 2000 professional comes with a free version of IIS. This
version limits you to 10 consecutive connections. You can live with
this limitation for small sites by disabling HTTP keep-alives and
using the least number of images as possible. Otherwise you'd need
Windows 2000 Server or better, which cost more. Win2K Pro costs about
$300.

You'll need a high-speed internet connection. The cheapest optional is
a consumer connection like Road Runner from Time Warner Cable. AT&T,
Verizon, AOL and other companies also offer broadband services.
Cable-modem services are recommended. These solutions will cost about
$44/Month.

You'll need Access97 or better to create your member database with.
This is part of Office 97 Pro. Later versions will also work. This
will cost about $300 depending on the version you purchase. Access97
XP is $99 by itself which will also work.

You'll need to learn ASP, VBScript (or JScript) and SQL. This is easy
enough but time consuming. It is definitly more cost effective to
learn these skills yourself from the internet then it is to hire
someone else to write your scripts for you. This doesn't cost anything
but prior programming experience is highly recommended.

Alternatively, you could hire rackspace to do your hosting and hire a
programmer to do the coding. Check rackspace.com for thier pricing. A
programmer will cost you about $40 to $50 per hour for at least a
month to get the scripts written and debugged.

And your guess is as good as mine for marketting. Most marketting
companies online are pretty affordable though.

-Robert