Kuro5hin.org: technology and culture, from the trenches
create account | help/FAQ | contact | links | search | IRC | site news
[ Everything | Diaries | Technology | Science | Culture | Politics | Media | News | Internet | Op-Ed | Fiction | Meta | MLP ]
We need your support: buy an ad | premium membership

Website profitability: an economic analysis

By Remus Shepherd in Internet
Wed Mar 07, 2001 at 06:25:16 AM EST
Tags: You Know... (all tags)
You Know...

This analysis is being submitted to both Slashdot and kuro5hin, as well as posted on Lum the Mad's forums. All three sites have been discussing the recent disintegration of the banner ad market and how it affects websites. More importantly, how people can continue to survive, let alone make a profit, on making web content.

I'm a scientist, not a MBA. So the first question I asked was, 'Why would anyone think they could make a profit off a web site?' What follows is my analysis on the profitability of advertiser-based websites. Comments are appreciated.

First I went looking for numbers. There are several web hosting options, for personal sites, business sites, and going up to heavy bandwidth use. Some typical numbers I found were:
                        Base Price   Bandwidth   Bandwidth
Personal use website        $20.00      500 Mb    $0.10/Mb
Business website 1          $85.00     3000 Mb    $0.10/Mb
Business website 2         $300.00     7000 Mb    $0.03/Mb
All this is on a monthly basis. So you can pay $20 per month for a personal website, which will serve 500 megabytes per month; if you go over that allotment you're charged $0.10 per megabyte extra. Some hosting services use different price schemes for excess bandwidth cases (forcing you to buy the next higher plan, for example), but in the end you pay for the bandwidth somehow, so let's go with these numbers.

Now let's look at Banner Advertisement rates:
Personal websites, game sites        $1-5 CPM
Targetted Demographics             $20-50 CPM
Highly valuable demographics   up to $100 CPM
'Highly valuable' includes such groups as member-only websites for lawyers or doctors. I read that the average rate was $35 CPM, but I somehow doubt that's current. Note that 'CPM' is 'counts per mille', or the rate for a thousand viewings of the advertisement.

All right. So why are websites going under?

Advertising rates have plummetted, but that doesn't explain it all. Let's look at a fictional personal website, where ads exactly pay for the cost of hosting:
Base hosting cost per month:    $20.00
Ad Rate per thousand:            $2.00
Extra Bandwidth cost per Mb:     $0.10
Initial Bandwidth:                 500 Mb
Content Size                      0.02 Mb
Page Views per month            25,000 views
Ad Revenue                      $50.00
Bandwidth used                     500 Mb
Cost per thousand pages          $2.00
Bandwidth cost                  $20.00
Profit                          $30.00
There's a few things to notice about this. First of all, the number of page views DOES NOT MATTER. With zero page views, the owner only has to pay his $20/month. At 1,000 page views he breaks even. At 25,000 he hits his bandwidth allotment, and makes the most money he'll ever make from his site: $30/month. Any further increase in page views are exactly balanced; his cost per thousand pages exactly matches his cost per thousand page views. Even if he gets 1 billion hits, he will make no more than $30. (Of course, his host will force him to change his hosting options.)

The above is a Flat website. The amount of money it can lose is bounded, at the $20 monthly fee. The amount of money it can possibly make is also bounded, at $30, at the point where its bandwidth allotment is exactly filled. Note that one of the key ingredients of this site is the small page size: 20 kBytes per page, including HTML and images.

Another important thing here is what happens beyond the initial bandwidth allotment. Websites, especially commercial ones, want to succeed. More hits should be better. So unless the breakeven point is ridiculously high (and sometimes it is), I'm going to ignore it and just look at the extremes.

Most websites aren't flat. Most websites are Fat. A Fat website loses money when its hits go up. For example, if the above website added content to its pages:
Extra Bandwidth cost per Mb:       $0.03
Content Size                        0.04
Cost per thousand pages            $4.00
Revenue per thousand pages (CPM)   $2.00
Income per thousand pages         -$2.00
The site loses money for every hit above it's bandwidth allotment.
The maximum profit doesn't change; it's still locked at $30.00 per month. But there is no limit to the losses this website can have.

That's a Fat website. There is no limit to the amount of money a fat website can lose. There is a limit to the amount of money a fat website can make; that limit is when the bandwidth exactly equals its allotment. The maximum profit of a fat website is Profit = CPM * Bandwidth/Pagesize - (fixed monthly fee).

Let's look at some of my favorite websites.
 Front page size (HTML and images)
Slashdot         68 kB
kuro5hin         85 kB
Lum the Mad      98 kB
Something Awful  146 kB
CNN              140 kB
These numbers include the HTML and all images except for ads; the ads are generally served by the advertiser's server, not the website. The exceptions are Slashdot and CNN, who both appear to serve their own ads, so I included those in the totals.

Some sites (like CNN and Slashdot) may get breaks on bandwith cost because they're owned by server companies. But there is a bandwidth cost, and it doesn't change. The CPM from ads might be significantly better.

Making some assumptions about their bandwidth cost and CPM rate, here's what we have.
Extra Bandwidth cost per Mb:       $0.03
Content Size                        0.068
Cost per thousand pages            $2.04
Revenue per thousand pages (CPM)   $5.00
Income per thousand pages          $2.96
Slashdot's a trim website. It's probably Flat, or better than flat, if its advertisers recognize that it has a specific demographic and pay accordingly. If their advertisers are screwing them, or if it pays more for bandwidth, then it could be fat.

But it looks like Slashdot may be a Thin website. There is a limit to how much a thin website can lose, and that limit is the monthly fee. There is no limit to how much money a thin website can make. The more subscribers the better. Unless the owners are independently wealthy, all websites should strive to be thin.

Note that there are only three variables controlling a website's profitability: Content Size, CPM, and Bandwidth cost. The owner of the site only has control over the content size. If the advertisers change their rates (or refuse to pay), or their host changes their rates, a thin website can turn into a fat one overnight.
Extra Bandwidth cost per Mb:       $0.10
Content Size                        0.085
Cost per thousand pages            $8.50
Revenue per thousand pages (CPM)   $5.00
Income per thousand pages         -$3.50
Kuro5hin needs a CPM of $8.50, or a good deal on bandwidth, in order to ever turn a profit. Otherwise it's fat; the more popular it gets, the more it loses money. Note that kuro5hin has two ads on their front page, so they might be making more money from ads. A lot of online magazines use this tactic, by spraying every page with dozens of different ads. I somehow doubt it's a strategy that works well, as it should only prompt the advertisers to lower their rates.
Lum the Mad
Extra Bandwidth cost per Mb:       $0.10
Content Size                        0.098
Cost per thousand pages            $9.80
Revenue per thousand pages (CPM)   $5.00
Income per thousand pages         -$4.80
Pretty fat, Lum. Er, nothing personal. As a side note the site looks very clean, with only small images...but the front page has thousands of links and acres of text. Lum's bandwidth is being chewed up by pure HTML.
Something Awful
Extra Bandwidth cost per Mb:       $0.10
Content Size                        0.148
Cost per thousand pages           $14.80
Revenue per thousand pages (CPM)   $5.00
Income per thousand pages         -$9.80
Poor, poor Lowtax. On the other hand, the quality of content he provides is worth it; of the sites listed, this is the only one I'd subscribe to.

An aside on subscriptions. With a subscriber base, a website's income is suddenly fixed. But its bandwidth cost continues to fluctuate, as subscribers load the page multiple times and as visitors come to the front page. Let's assume the visitor pages are flat; then we can look at SA if Lowtax instituted a subscriber charge.

Revenue = (No. Subscribers) x (Subscription Fee)
Costs = (No. Subscribers) x (Revisit Rate) x (Content Size) x (Bandwidth Cost)

Profitability = Revenue/Costs = (Fee) / (Revisit Rate) x (Content Size) x (Bandwidth Cost)

Where what I'm calling 'Profitability' is a ratio that shows whether a website is thin, flat, or fat. Flat websites have P = 1; Fat is P < 1, Thin is P > 1. The main difference between them, remember, is whether their profit or losses are bounded, and thus whether increasing numbers of hits are good or bad for them.

Note that the number of subscribers would drop out of this equation. The number of subscribers DOES NOT MATTER, because it drives up bandwidth costs at the same time it drives up revenue. Changing the number of subscribers will make the situation better or worse, but it will not turn a fat website into a thin one or vice versa. Only the revisit rate -- the number of pages each subscriber views per month -- the fee, and the old Content Size and Bandwidth Cost matter. Or:
Extra Bandwidth cost per Mb:                 $0.10
Content Size                                  0.150
Subscriber revisit per month                150.0
(5 page views per day per subscriber)
Cost per thousand pages per subscriber       $2.25
Revenue per thousand pages per subscriber    $2.25
(Which is the Monthly subscriber fee)
It can work. But god help you if you get it wrong, because if your bandwidth is too high there's no balancing feedback in your income, and costs can become immense quickly. Note that the breakeven fee is equal to the (Content Size in Mb) x (Bandwidth Cost) x (Revisit Rate). And a change in any of those can suddently turn a website from thin to fat; from profitable to a money pit.
Extra Bandwidth cost per Mb:       $0.10
Content Size                        0.14
Cost per thousand pages           $14.00
Revenue per thousand pages (CPM)   $5.00
Income per thousand pages         -$9.00
CNN's front page is chock-ful of ads, all of which they appear to serve, though routed through Akeami. But even stranger, all of the ads are for other products of AOL/Time Warner. They are their own client, they don't have any actual revenue (although they do have actual costs), and thus CNN.com exists by AOL/TW's whim. Hey, they can afford it. Note that if CNN gets enough of a break on bandwidth, they might be thin after all. (I don't expect this to be the case, as they also serve streaming video and other content at the drop of a hat. Their average content size is probably an order of magnitude higher than above.)

Please remember that this is just a back-of-the-envelope analysis, and I'm sure all these sites are somehow at least marginally profitable, whether through good CPM contracts, good bandwidth costs, or just not having enough users to exceed their bandwidth allotment more than occassionally. As an abstract analysis, though, the lessons we've learned are valid. What are they?

Trim Your Website. The content size is the only parameter you can control to make your site profitable. Especially important is to make your front page, and the pages most often viewed by visitors, as lean as possible. The Drudge Report's front page is only 35 kB...we will never be rid of Matt Drudge. Lum, get those links off your front page.

Large Content Can Only Be Delivered To Small Audiences. Hobbyists can afford to provide huge images and other content if they only get a few hundred hits. Of course, if a site like this gets Slashdotted and its bandwidth balloons, the owner can get into debt fast.

Subscriber Models Can Work. But they're fickle, and I expect to see a few rotting corpses of websites who misjudged their needs.

Pay-For-Content May Not Always Work. You can make people whip out their credit card and charge them for content. But visitors to your site might still ruin your profit margin if they drive up your bandwidth.

We, The People, Own The Internet. Under the current pricing schemes, it is impossible to turn a profit by 'providing content' over the web based on advertising revenue. The more content you provide, the more money you lose. Corporations don't like that. Hobbyists and low-traffic web pages can get away with it.

...Except For Time Warner. If your bandwidth is free you can provide anything you damn well please. Expect to see more mergers and mega-corporations, with part of each corporation being an internet backbone. Or maybe AOL will become all.

So that's the World Wide Web of the future. Megacorporation sites, and some sites with content available by credit card access. Their competition will be hobbyists -- with or without advertiser support -- who have specific, narrowly targetted offerings. And then there will be room for a few advertiser-supported websites, who have learned to be thin and lean enough to beat the system.


Voxel dot net
o Managed Hosting
o VoxCAST Content Delivery
o Raw Infrastructure


Related Links
o Slashdot
o Kuro5hin
o Slashdot [2]
o kuro5hin
o Lum the Mad
o Something Awful
o Also by Remus Shepherd

Display: Sort:
Website profitability: an economic analysis | 48 comments (48 topical, editorial, 0 hidden)
And there you have it, I suppose. (2.60 / 5) (#1)
by regeya on Wed Mar 07, 2001 at 12:01:35 AM EST

I wonder what would have happened to print publications if paper manufacturers, ink manufacturers, etc. had charged flat rates instead of giving volume discounts all along...

[ yokelpunk | kuro5hin diary ]

Problems with your numbers (4.80 / 21) (#2)
by rusty on Wed Mar 07, 2001 at 12:15:44 AM EST

This is a really interesting analysis, and I'm glad someone took the time to actually do the math. Nevertheless, there is at least one (and probably two) serious flaws with your numbers.

Bandwidth costs a lot less than you think it does. Any site large enough to attempt to survive on advertising income is running on it's own machines at a colocation facility (server farm). K5 for example, is colo'ed at VHosting. Now, we pay for our bandwidth with that little ad over there on the right, so we do actually have that "very good deal" you mention we'd need. But the thing is, we don't actually need it anyway.

Your numbers for bandwidth appear to be based on virtual hosting costs. $300/mo for 7Gb of bandwidth is outlandish for a colo provider. Look at VHosting's colo charges page. K5 uses over 100Gb/mo in bandwidth (probably more like 150 Gb). Configure that form for 2U of rackspace, and 150Gb/mo, and it'll cost you $460.00. $420.00 of that is for bandwidth. Which means that bandwidth on a colo scale actually costs roughly $0.0028 per Mb. That's 0.28 cents per meg.

Assume we make $5.00 per thousand pageviews. You calculate based on the front page, but my stats for the past few months show us us serving between 45 and 55Kb per page, on average. Let's be really generous, and call it 70Kb per page. So:

0.07 Mb/page * 0.0028 dollars/Mb = $0.000196 dollars/page, or roughly $0.20 per thousand pages.

An income of $5.00 per thousand pages, minus expenses of $0.20 per thousand pages, means we should be making a profit of $4.80 per thousand pages, not the loss of $3.50 that you calculate. Don't forget, that's with a padded size estimate.

In fact, our pages would have to be about 1.79 Mb each *on average* before costs would even equal income at $5.00 CPM. The idea of 1.79 Mb pages is pretty ludicrous. I doubt our longest nested comment page here is over 800Kb.

So, bandwidth costs are problem one. Problem two is with your estimates on CPM. A small personal site can't get $2.00 CPM anywhere. Sign up with someone like Burstnet, that allows low-traffic sites to join, and you're lucky if you make $0.01 CPM. The business just doesn't scale linearly like you'd expect.

$5.00 CPM is a really good deal, I doubt SA or Lum the Mad make that. I'm positive Slashdot makes much more than that. That's the thing with estimating income: you don't necessarily know who's making what. It varies wildly, and CPM rates are a very closely guarded secret.

Despite all this, I think the way you went about these calculations is admirably clearheaded, and I'm very glad to see it. I hope this helps clarify the case a little.

Not the real rusty

The equations are still great (4.00 / 3) (#6)
by RandomPeon on Wed Mar 07, 2001 at 01:24:18 AM EST

I'll take your word for all of it. But the equations still model the business well. It's refreshing to see someone attempting to apply real, hard-headed economics to anything with .com or .net attached to it. I've always wondered if it was even remotely possible to make money off web advertising - it looks like it is.

And it does point out that a link from Slashdot (or even Kuro5hin) could beat some people into the ground profit-wise. Maybe this is a new argument for them to cache pages instead of linking to them. If a website doesn't have the bandwith to survive being slashdotted, would slashdot be justified in assuming they don't have a business model that supports sudden and massive ramp-ups in traffic?

Finally, you have to remember certain costs which may or may not be independent of the number of page views. Somebody has to pay people to generate content, at least at sites besides Slashdot, which relies on ZDNET and CNet, and Kuro5hin, which relies on us. And there's the admin costs, the support costs, and other overhead....

[ Parent ]
Absolutely (4.50 / 4) (#11)
by rusty on Wed Mar 07, 2001 at 02:08:34 AM EST

But the equations still model the business well. It's refreshing to see someone attempting to apply real, hard-headed economics to anything with .com or .net attached to it. I've always wondered if it was even remotely possible to make money off web advertising - it looks like it is.

This was what I really liked about the article. The specific numbers may be off, but the calculations are a pretty sound base, and a very clear analysis of how the business works.

Finally, you have to remember certain costs which may or may not be independent of the number of page views. Somebody has to pay people to generate content, at least at sites besides Slashdot, which relies on ZDNET and CNet, and Kuro5hin, which relies on us. And there's the admin costs, the support costs, and other overhead....

That's the other big thing that's missing here. I could probably do a back-of-the-envelope estimate of how much that would drain out of the available cash pool, if you were to run a site with wholly paid staff and contributors. As it stands though, this is a pretty good model for how most of the sites on the web run: with one or two paid admins, and the rest volunteers.

Not the real rusty
[ Parent ]

bandwidth (4.00 / 1) (#22)
by alprazolam on Wed Mar 07, 2001 at 11:40:36 AM EST

if you're a good paying customer for say 6 months and all of a sudden one day you're on slashdot and you exceed your allocation or whatever, are you responsible for paying it exactly? what if you claimed it was a dos attack?

[ Parent ]
This is probably true (none / 0) (#26)
by weirdling on Wed Mar 07, 2001 at 01:39:11 PM EST

I doubt they'll hit you up for it immediately, at the very least. They'll probably simply tell you that you can pay it off over time or forgive the thing entirely. I'm pretty certain that if you bought, say, 5G of bandwidth, that you are limited to 10G total by the router, or some other multiplier, because the ISP definately has a vested interest in you not chewing up all their bandwidth that is in use by other customers.

I'm not doing this again; last time no one believed it.
[ Parent ]
CPM? (3.75 / 4) (#7)
by BigZaphod on Wed Mar 07, 2001 at 01:32:12 AM EST

Not that it's really our business, but does K5 actually make in the $5.00/CPM range? My site (bebits.com) gets nearly 60,000 impressions daily (usually only 40k will be recorded by the ad agency for various reasons (BeOS browsers don't like rich media ads, blockers, etc)). We have a CPM of about $0.22. That's killing us right now. We simply don't get the kind of traffic we need (apparently) to confortably pay for the server colo let alone pay for our time. And since we're a files site it just kills me that we can't actually mirror the stuff we list because there is no possible way we could afford it.

Anyway, just wondering how close to reality your examples were. :-)

"We're all patients, there are no doctors, our meds ran out a long time ago and nobody loves us." - skyknight
[ Parent ]
Any CPM is hypothetical (2.33 / 3) (#9)
by cp on Wed Mar 07, 2001 at 01:57:35 AM EST

Since we all know OSDN compensates rusty with sexual favors and astute fashion advice.

Sadly, editing is still strictly a volunteer job....

[ Parent ]

Did I say "top secret"? (3.50 / 4) (#10)
by rusty on Wed Mar 07, 2001 at 02:04:27 AM EST

What cp points out elsewhere in this thread is certainly no secret. However, the actual terms of my deal with OSDN are confidential, and so I can't really say. The rest of the numbers in my calculation are real. The CPM may or may not be, I was just going with what Remus posited. An interesting thing to note is that our break-even CPM is about $0.20 (or would be, if we actually paid for bandwidth).

What do you pay for hosting? In theory, you ought to be able to break even with a colo deal, and perhaps you can get someone to offer you bandwidth in exchange for a permanent ad, as we did.

Not the real rusty
[ Parent ]

Working on it.. (3.75 / 4) (#12)
by BigZaphod on Wed Mar 07, 2001 at 02:09:03 AM EST

Actually, we're working on such a deal right now. It's not a 100% trade, but it will cut our bandwidth costs in half, so we'll be back in black soon. So I'm happy about that. My main reason for bringing it up, though, is that I wanted to make sure we weren't doing something horribly wrong by not being able to make it with the kind of traffic we get. I'm glad to know that even the big and popular sites have to have some help on this front. :-) Thanks Rusty!

"We're all patients, there are no doctors, our meds ran out a long time ago and nobody loves us." - skyknight
[ Parent ]
OK ... the had question. (none / 0) (#31)
by jann on Wed Mar 07, 2001 at 06:25:01 PM EST

I presume that K5 pays for itself ... but does it manage to pay rusty and Inoshiro's salary? could it? if not how far away is it from reaching this point?

[ Parent ]
Think I used an old advertiser's ratesheet... (4.50 / 2) (#15)
by Remus Shepherd on Wed Mar 07, 2001 at 09:34:04 AM EST

Yeah, people who know seem to think that the numbers are off by quite a bit. I think the advertiser's ratesheet I found must have been from before the dot.com crash. I'm glad that your bandwidth is so cheap, though; you could deliver three times the content you're doing now. ;) I just found it interesting that, whatever the actual numbers are, the business operates in two modes. Thin websites have unbounded profit, fat websites have unbounded losses. Again I'm no economist, but I've never heard of any business like that... And I certainly don't know any business where the more customers you get, the deeper you sink into debt.
Remus Shepherd <remus@panix.com>
Creator and holder of many Indefensible Positions.
[ Parent ]
Numbers, and conclusions (5.00 / 4) (#17)
by rusty on Wed Mar 07, 2001 at 09:49:05 AM EST

While your numbers are off, here and there, I also thought the overall conclusions were very interesting. There definitely seems to be a threshold of profitability, which depends on the ration of your bandwidth expenses (cost * page size) to your ad income.

The thing is, it's a result of having a fixed cost and income per page view, and some initial monthly fee on top of that. For a "flat" site, your model posits an initial fee which is lower than revenue to start off with, but at a certain point, income and expense become balanced. Imagine I run a lemonade stand (Thank you, Apple ][ for providing the foundation of my economic knowlege), and every month I buy a "lemonade stand super-pack" which includes 10 cups and enough lemonade mix to fill them. This costs me $5.00. I sell my lemonade at $0.75 a cup, because it's all the market will bear. So, after my first ten cups, I'm ahead of the game by $2.50. But when I go back to the store to get another super-pack, they're out, and I have to buy mix ($3.75 for ten cups worth) and cups ($3.75 for ten cups) separately. From now on, I'll break even forever. It's the exact same scenario.

The case of the fat website is analogous to my selling my lemonade for $0.50 a cup. I still break even after ten bups, but then I start losing money on every cup. Gee whiz, I should raise my prices, or find a cheaper supermarket. Selling below cost is never a good business plan (ask the former employees of pets.com: they're the ones on the streets of LA with the signs that say "will operate stupid hand puppet for food").

A thin website is selling it's pages above cost. This, traditionally speaking, is a good business practice. I don't think I need to eleborate much on that. :-)

The only real problem with this article is, you should probably have written it in about 1998. :-)

Not the real rusty
[ Parent ]

Duncan Yo-yo's (5.00 / 1) (#35)
by nuntius on Wed Mar 07, 2001 at 08:50:43 PM EST

I don't have the references handy, but Duncan went bankrupt during the height of their yo-yo sales due to a phenomenon similar to the one you describe.

They discovered that sales were going much faster than production could keep up. Not wanting to lose out on this huge market opportunity, they ramped up production ASAP. Paid overtime, purchases of higher-quality wood, factory construction and other costs finally raised the price of the yo-yo until Duncan was selling each one at a loss.

By the end, their operation was so overpriced that they had a hard time selling their assets after declaring bankruptcy. The only thing of value they had remaining was the trademark; even "Duncan Yo-yo" sold for a charm...

[ Parent ]
Free bandwidth for me :) (2.66 / 3) (#3)
by delmoi on Wed Mar 07, 2001 at 12:16:08 AM EST

Actually, bandwidth here at ISU costs me exactly $7 a semester, and that's 200 megs outbound a day.

Of course, when I leave collage, it's going to suck. But for right now, I can host whatever website I want, for as long as I want. (Of course, I can't use it for 'commercial gain', which is why picture-rate.com doesn't have any advertisements.) Having your own bandwidth, and being independent from an ISP is a wonderful thing.

If, when I leave here it is at all possible, I'm going to try to get my own t1/t3 (who knows, maybe I'll be a stock-option millionare). A t1 would be slower then a cable modem, and a t3 would cost a shitload of money, but It would be well worth it to me not to have to deal with the brain-dead restrictions placed on users by commercial ISPs. I want to be able to do whatever I want with my connection, and I'll pay good money to be able to.
"'argumentation' is not a word, idiot." -- thelizman
Economic distortion (3.50 / 4) (#4)
by kmself on Wed Mar 07, 2001 at 01:03:57 AM EST

Your university is essentially sponsoring your Internet connection. The rates you're seeing aren't indicative of real costs as experienced by USI, but are some artificially established value. Though interesting to you, your situation isn't generally applicable to Internet business opportunities as a whole (though it does suggest that college is a good place to get an Internet business off the ground, as does the track record).

What you've got there is a clear-cut case of an economic distortion introduced by artificial subsidy.

Karsten M. Self
SCO -- backgrounder on Caldera/SCO vs IBM
Support the EFF!!
There is no K5 cabal.
[ Parent ]

you need to look at your site on a Mac (none / 0) (#18)
by imperium on Wed Mar 07, 2001 at 10:27:33 AM EST

Not even iCab could make sense of your slider, and it's all SNAFU on Navigator...

how about some check-box options instead?

[ Parent ]

sorry :( (none / 0) (#23)
by delmoi on Wed Mar 07, 2001 at 12:28:40 PM EST

it's only in beta, and it only works in IE (and I was only barely able to get ie5 and ie5.5 to do the same thing). I do have a 'simplified' page that people could use, but there's no bounds-checking on it, so people could enter scores of 10,000 or -223,234,590,565.2342342454 if they wanted to...
"'argumentation' is not a word, idiot." -- thelizman
[ Parent ]
I'd fix it if I were you (none / 0) (#28)
by regeya on Wed Mar 07, 2001 at 04:20:54 PM EST

and no, this is not a flame or a rant. Maybe a little flamey. :-)

It should be possible to do such a thing in Not IE. Or at least, I'm assuming so, as I saw something resembling a slider, but it was non-functioning. I tried both Netscape 4.76 and Galeon (a Mozilla derivative) and it worked in neither. If you're having trouble getting it to work in different revisions of IE5, I'd suggest a rewrite. :-)

[ yokelpunk | kuro5hin diary ]
[ Parent ]

well, it works *now* (none / 0) (#43)
by delmoi on Sun Mar 11, 2001 at 12:34:14 PM EST

I do plan on making sure that it works in all browsers before the time I 'go live'. I actualy started development in mozilla, but I couldn't figure out how to find the location of the mouse, so I switch to IE. Right now I'm going to have a really simple HTML2.0/no JS whatsoever version of the page and use either client side or server-side redirection depending on what browser their working on

Right now, I'm more intrested in getting the backend doing things correctly (It dosn't use a database now, everything is just stored in hashtables and vectors in memory), and imlementing the features I want.
"'argumentation' is not a word, idiot." -- thelizman
[ Parent ]
That slider thing is annoying (none / 0) (#32)
by skim123 on Wed Mar 07, 2001 at 07:25:41 PM EST

Just my two cents... I much prefer AmIHotOrNot.com's radiobutton clicking thingie... easier to use, standard HTML, blah blah blah.

Money is in some respects like fire; it is a very excellent servant but a terrible master.
PT Barnum

[ Parent ]
A couple things you forgot.. (3.50 / 4) (#5)
by BigZaphod on Wed Mar 07, 2001 at 01:24:00 AM EST

I really liked the write up and I was glad to see someone trying to find a real solution to all this that is somewhat based in reality. However, there are a couple points you missed. One is that often times mid to large sites colocate. That means they often have to buy hardware up front and possibly even pay some extra rack fees on top of bandwidth (I don't, but someplaces do). So there's a large up-front investment cost right there that needs to get paid back. But the big thing that is missed is cost of labor. A small site is done mostly as a hobby and the owners really don't expect much in return. However, larger sites eventually begin to take so much time to maintain and update that it could almost make up a full time (or at least part time) job. Your numbers do not take that into account. It's hard to run a website when you have to keep a real job as well (plus avoid failing in school :-).

Anyway, great article and this is a subject that needs MUCH discussion.

"We're all patients, there are no doctors, our meds ran out a long time ago and nobody loves us." - skyknight
Atta Boy! (2.20 / 5) (#8)
by slimy_snake on Wed Mar 07, 2001 at 01:33:50 AM EST

Dayum good, is all what I would say. I'm no expert on the business side of running a site, but I guess the lack of meaningful and deep analysis like this is the main reason why most dots have gone as El Reg would say "titsup."

Yup, the figures are a wee bit arbritary and it is unfair to make such generalised assumptions on hosting options. But it is a good start anyway and well thought out.

I'm so tired, of playing
Playing with this bow and arrow - Portishead

Excellently written, but maybe inaccurate (3.00 / 4) (#13)
by Aquarius on Wed Mar 07, 2001 at 02:52:22 AM EST

This article is very well written; it's a remarkably clear and easy-to-follow discussion of the topic. Unfortunately, whether it's useful in itself depends on the accuracy of the figures, a lot of which Remus frankly admits are guesstimates. The first thing I thought once I'd read it was "I hope Rusty chimes in with an indication of whether or not the figures bear resemblance to reality", and it seems that they don't necessarily do so. That aside, it will spark interesting discussions, as more people who already know about this fom personal experience flesh out the figures with more detail. I'd like to see, when the discussion is over, this article turn into an essay on the topic somewhere so it's available for permanent reference by people thinking of funding through advertising.

Good work.


"The grand plan that is Aquarius proceeds apace" -- Ronin, Frank Miller
You forgot one thing. (3.25 / 4) (#14)
by farmgeek on Wed Mar 07, 2001 at 08:47:33 AM EST

Labor. If you do not start out figuring in the labor, your numbers will be artificially inflated. The labor spent on a site costs something. Granted, you may receive intangible benefits from running a site, such as contacts and experience, but if you are planning to make a go commercially, don't forget the hours you spent getting everything up and running and the time spent on maintenance.

Now, the question is where can I get $1-$5 cpm for a small site? Most of the rates I've seen have been more in the $0.10 - $0.50 range for lower trafic sites (< 1 mil ad views per month).

Profit first, pay employees second. ;) (5.00 / 1) (#21)
by Remus Shepherd on Wed Mar 07, 2001 at 11:19:22 AM EST

I tend to think of labor as something you pay out of your profits; you have to have a business plan that gives you a revenue first, then you can allocate money to employees. But I'm well aware that the real world doesn't always work that way. ;) This makes the analysis even less relevant to large commercial sites that have payrolls to meet.
Remus Shepherd <remus@panix.com>
Creator and holder of many Indefensible Positions.
[ Parent ]
Exactly (3.50 / 2) (#16)
by LaNMaN2000 on Wed Mar 07, 2001 at 09:42:36 AM EST

The author hit on one of the most important consequences of the advertising market fallout. People no longer look only at the CPM rate that a site can obtain; they also examine the bandwidth consumed by each pageview. Sites like /. and Kuro5hin, that serve enormous pages, will need to get enough pageviews to support a "bulk" hosting plan to remain profitable.

While many commercial websites with large staffs are using large Flash animations, you hardly ever see small content sites using them. The cost to serve the Flash animation could easily exceed the amount of money that will be made from a given visitor throughout his visit. I think the advertising fallout will prevent sites from using design enhancements that increase the bandwidth/pageview ratio.


Lenny Grover -- link-spamming to make Google give me my name back!
the new web business model (3.00 / 3) (#19)
by pustulate on Wed Mar 07, 2001 at 10:44:54 AM EST

Is going to be blackmailing websites -not- to get slashdotted! Just imagine: "Pay be $20 and I won't post that story." :)

Actually, are there any tools available to monitor web costs? Just curious.

CPM Rates (4.00 / 4) (#20)
by briggsb on Wed Mar 07, 2001 at 10:52:01 AM EST

Just to add another data point to the whole CPM discussion. My site BBspot gets about a 10 to 15 cent CPM. A $2 CPM would be fantasyland for me. I have thin pages (my host offers unlimited bandwidth anyway) so I don't have huge bandwidth costs.

If my site needed $400 bucks a month in coloc services I can guarantee you that it would disappear.


Numbers are off... (4.57 / 7) (#24)
by selkirk on Wed Mar 07, 2001 at 12:35:42 PM EST

I am almost an MBA, so let me give it a try...
On the cost side:
Hurricane electric charges $25/mo for 10GB of traffic in a virtual host. Extra bandwidth with this account is $100 per GB. Rackspace offers a dedicated server (750 Mhz, 256MB RAM) with no bandwidth for about $300/mo. Bandwidth is then $30 per GB with discounts at higher quantities. ($250 for 100GB). The question with this host is how many pages could you serve from a single server, as there is a significant fixed cost here.
on the revenue side:
ContentZone guarentees a fixed CPM of .20 for sites of size < 250,000 hits/mo (The virtual host option above) and .40 for size >250,000 hits/mo (The dedicated server option).
So, the virtual host breaks even at 125,000 hits/mo, assumimg a page size of 100K.
The dedicated host breaks even at around 2,000,000 hits/mo, assuming a page size of 100K.
ContentZone should be about the bottom of the barrel. It should be possible to get more per CPM by offering a better targeted audiance. However, since it costs money to buy and sell specific quanitites of impressions, only sites that can offer larger numbers of hits can acheive the economies of scale to participate in this market.
Oh, one detail...
Now, include one full time employee at $5,000/mo. Assume that employee can generate 4,000,000 hits per mo on the dedicated host. (!?) They must get $1.57 CPM to break even. Also, assume that a single server can handle 4,000,000 hits/mo.
What were they thinking?
So, when you are talking less than millions of hits, you are talking about a hobby. I have to laugh every time I hear about a dot com laying off hundreds of employees. What were they thinking? How many hits (billions?) and what CPM (hundreds?) did they think they were going to get?
Simple economics
Here is the deal: simple economics dictates that the CPM for ads will fall to nearly equal the costs of hosting and generating an audiance. (perfect comptetition, low barriers to entry, etc.) The ad market will continue to "disintigrate" until it reaches this level. The specific costs of hosting and CPM are irrellevant as they will eventually be equal.
Making money
The sites that will make money are those that can beat market averages: Generate more hits per employee, charge more per impression, host cheaper, or charge more for a given quantity of hosting. Running a business in this market is not going to be easy or "free money." However, most businesspeople are used to this.

Methods for .com Revenue Generation (4.00 / 1) (#25)
by flowers on Wed Mar 07, 2001 at 01:18:35 PM EST

It is not unreasonable for a .com to have hired (or fired) hundreds of employees. Advertising is only one way for a .com to get revenue. For a content provider, yes, it's probably the most significant way, but not all .coms are just content providers. There are also application service providers, and the plethora of 'shopping cart' web sites (for example). e-commerce and e-business sites are .coms, too.

[ Parent ]
Rackspace bandwidth charge (none / 0) (#41)
by phinance on Fri Mar 09, 2001 at 09:21:12 AM EST

Small correction: The highest charge per GB at Rackspace is $30/20Gb = $1.5/GB, not $30/GB.

Read, annotate, and discuss open source documentation.
Andamooka: Open support for open content.
[ Parent ]

Irony... (3.40 / 5) (#27)
by Giant Space Hamster on Wed Mar 07, 2001 at 02:09:50 PM EST

It's ironic that as individual users move to broadband, content providers are moving away from the stuff that makes broadband worthwhile.


The whole problem with the world is that fools and fanatics are always so certain of themselves, but wiser people so full of doubts.
-- Bertrand Russell

Maybe similar economics hold for SPAM ? (3.50 / 2) (#29)
by redelm on Wed Mar 07, 2001 at 04:35:21 PM EST

A very interesting analysis. Perhaps similar economics hold for SPAM (Unsolicited Commercial Email)? To wit:

On the high side, if SPAM pays $5.00 CPM and costs $0.10/MB to send, then the breakeven email size is 50 kB. On the low side, if SPAM has 1.00 CPM, but is colo'd costing only $0.003/MB, then breakeven size is 333 kB. Ouch.

I've noticed that SPAM is rarely very large, 10-30 kB seeming average. Perhaps this sort of bandwidth economics is the limitation. We can only hope that SPAM doesn't pay well (low CPM), and is so unpleasant for ISPs that they charge a high bandwidth fee.

GZip Compression? (4.33 / 3) (#30)
by nebby on Wed Mar 07, 2001 at 04:46:36 PM EST

Maybe I'm crazy, but since I implemented GZip compression on my site my front page (at least, for IE4/5 and Netscape 6 users, maybe NS4 I'm not sure) is like 6kb (down from 60kb)

Maybe on a large scale site like /. it would be too CPU intensive to compress the page, but it certainly is a nice way to chop the bandwidth.
Half-Empty: A global community of thoughts ideas and knowledge.

how? (none / 0) (#33)
by timmyd on Wed Mar 07, 2001 at 08:03:59 PM EST

how do you get the clients to decompress the page? i didn't think windows people have gzip compression.

[ Parent ]
use content-transfer-encoding: gzip (none / 0) (#46)
by phr on Tue Mar 13, 2001 at 03:09:18 AM EST

This is supported in NS and IE 4.x and later browsers. There's a mod_gzip apache module floating around that compresses outgoing html. I'm not sure if it's on modules.apache.org but you should be able to find it on google.

[ Parent ]
PHP and gzip (none / 0) (#47)
by Peej on Tue Mar 13, 2001 at 08:44:49 AM EST

PHP also has gzip support. I haven't used it and I think it requires you to cache your response, run it through the gzip function, and then send it to the client with the correct heading all yourself.

[ Parent ]
Re: gzip (none / 0) (#34)
by agentk on Wed Mar 07, 2001 at 08:04:33 PM EST

That brings the size of your data down on the server, but its uncompressed before being sent out; your bandwidth usage is the same.

[ Parent ]
Client Side GZip (none / 0) (#44)
by nebby on Mon Mar 12, 2001 at 03:25:03 AM EST

The latest browsers can decompress gzipped content if you send them the right headers. I'm not talking about file storage, I'm talking about compressing the HTML on the way out the door so it's 10x smaller going over the pipe.

Half-Empty: A global community of thoughts ideas and knowledge.
[ Parent ]

Banner ads suck (2.00 / 1) (#36)
by k5er on Wed Mar 07, 2001 at 09:20:23 PM EST

They are extremely annoying. Thats one of the things I used to like about K5, but I guess its all about the money...sort of.
Long live k5, down with CNN.
Another perspective on economies of scale (4.50 / 2) (#37)
by Scrymarch on Wed Mar 07, 2001 at 10:49:50 PM EST

Not sure if this has been posted before, but a column at Dan's Data, a hardware review site, had a few interesting things to say about economies of scale when it comes to online content. Basically his premise is that sites that are small businesses are more profitable on the web because of reduced barriers to entry. Note that he gets by mainly on referrer fees, not an advertising, and referrer fees are left out of the above "back-of-the-envelope" analysis.

A few things to remember. (4.50 / 2) (#38)
by gromm on Thu Mar 08, 2001 at 04:02:03 AM EST

I work for a computer consultancy that does work for small to medium-sized websites, just so you know that I'm not talking out of my ass.

Let's take Slashdot as a good example...

Slashdot does not pay for its hosting in the terms you state here. They have an internet connection right to their office, with their web server sitting on it. This would cost probably at least $4000 or more per month for a 10 gigabit/s connection. (which they would need... they *are* slashdotted every day after all) They pay for bandwidth, but at probably a lower rate than they would at a hosting company; less than $20 a gig is my guess.

They also have a staff. Around six or seven people, I think. The cost of this staff would easily dwarf their connection costs, at at least 3500 or 4000 a month each. (that's quite conservative though, assuming they're the dedicated revolutionaries we like to think of them as) This is an estimate... their "about us" page doesn't give me much detail I'm afraid.

Thankfully, they also lie in the "Targeted Demographics" pay scale for their advertisements. This is stated up front on their web page... "news for nerds, stuff that matters." Not only is their advertising targeted to a particular demographic, but so is their content. As a result, they can charge much more money for their banner ads than your average "personal webpage."
Deus ex frigerifero
not quite (5.00 / 2) (#42)
by god on Fri Mar 09, 2001 at 06:15:03 PM EST

I wouldn't have bothered to respond except that you prefaced your comments with the warning about how you aren't talking out of your ass. Clearly you are and I find in most cases when people go out of their way to tell you of their credentials they are usually not what they claim. That said slashdot hosts their site in an 7'x8' cage in an exodus datacenter in waltham massachussetts. It's a dinky little place actually. At exodus you get an ethernet line run to your cage in exchange for paying a for a minimum amount of bandwidth plus burstability. For instance you can pay for a 5mb base on a fast ethernet link (note not 10GB, 10GB is faster than the backplane of anything but a $50,000 LAN switch). Anyhow this setup costs around $5600 plus $1200 per mb of sustained bursting(95%) on a normalized curve. Details of this are listed in an old article of theirs where they talk about their install of the cat6500. If you total up the cost of leasing the cage, paying for power(only some of which is usually included) and other stuff theyprobably shell out 8K per month to host their site(s). just spreading the word....

[ Parent ]
Is it just me...? (4.00 / 2) (#39)
by insomnia on Thu Mar 08, 2001 at 08:33:46 AM EST

...or does all this number crunching seem a bit wanky? It seems to advocate developing websites without substance, on the basis that they're more affordable to host.

Except, of course, that content is what drives people to your site... and dynamic content is what keeps them coming back.

About the only good advice I got from all of this is that it strongly argues for a lean, mean main page for your website... but then again, so does usability.

As for subscriber models being fickle, if you really want to see something fickle, why don't you take a look at all those dotcom websites that based their business model on advertising. Based on my experience running a site that is completely member supported (LiveJournal.com), subscriber models are a far more stable and profitable source of revenue.

People will contribute money to your site if they feel that what you are supplying them is:
1> of value to them 2> empowers them in some way 3> provides a sense of community

Personally, I can't think of a good reason why any subscriber-based site shouldn't average at least $15 revenue per 100 users per year. We only seriously started on a member-based business model around November, and we're already at about $40 per hundred users. I suspect we will top out at around $80 per hundred users per year.

Here is a partial list of just some of the ways I can think of off the top of my head that you can fund a website...

- Micropayments
- Memberships
- Merchandising (t-shirts, cups, etc.)
- Offering internet connectivity through 3rd parties
- webhosting
- web-based e-mail
- Offering financial services
- Selling relevant books, etc. through Amazon, etc.

Many of these revenue sources can be implemented without making your site look like one big ad, too. All finances aside, the real trick is making a site that your users love, can actively participate in, and will gladly support. If you have that, bandwidth costs are the least of your worries.

'cause I said so, that's why!

hmmmm, let's try this on my site (3.00 / 1) (#40)
by danny on Thu Mar 08, 2001 at 09:55:55 PM EST


Cost to run - about $30/month (for high-quality hosting at pair.com). Running at around 140 000 page impressions a month. (It's a text-only site, so that's well within my bandwidth quota of 400MB/day.) Yeah, I could probably recover costs from that - if it weren't for the fact that I loathe advertising.

But bookshop affiliate programs would make more sense for a massive book review collection... does anyone know of a decent affiliate program run by people with ethical business practices? (ie, not patent-abusers such as Amazon)

[900 book reviews and other stuff]

How much can you charge per CPM? (3.00 / 1) (#45)
by wangchung on Mon Mar 12, 2001 at 11:31:07 AM EST

There is an article here that discusses how much internet advertising is really worth.

Browser Text Compression (3.00 / 1) (#48)
by EriKZ on Wed Mar 14, 2001 at 06:49:50 PM EST

What I don't understand.
Compression utilities are great with text. The last time I loaded a page from slashdot it was 360k! Why don't they have it so that the file is compressed on the server site, the browser gets the file, uncompresses it and shows the user?

Should be faster download times, bandwidth cut down to a a quarter of what it used to be.


Website profitability: an economic analysis | 48 comments (48 topical, 0 editorial, 0 hidden)
Display: Sort:


All trademarks and copyrights on this page are owned by their respective companies. The Rest 2000 - Present Kuro5hin.org Inc.
See our legalese page for copyright policies. Please also read our Privacy Policy.
Kuro5hin.org is powered by Free Software, including Apache, Perl, and Linux, The Scoop Engine that runs this site is freely available, under the terms of the GPL.
Need some help? Email help@kuro5hin.org.
My heart's the long stairs.

Powered by Scoop create account | help/FAQ | mission | links | search | IRC | YOU choose the stories!