COGBLOG

What does it say about Yahoo!’s sponsored search system that they could not make a profit from the 10MM visits/month that GeoCities sites were receiving before they pulled the plug last week?

I was reading this post yesterday, and it made me want to estimate how much the GeoCities sites were really costing Yahoo.

I tried to make some order-of-magnitude estimates of how much the hosting of these domains was costing Yahoo. Was it really a cash drain?

Hosting:
7.5 million hosted sites.
5 Megabytes of data/site (this is probably way high)

That’s 37.5 TeraBytes of data to store.
Using the highest tier Amazon S3 pricing (we know it would be cheaper than this) of $0.15/Gigabyte, that’s $5,625 in storage/month.

Bandwidth:
10 MM Visits/month.
Assume an average of 3.5 pages/visit (probably high).
Assume 100k transferred/page (also probably high).

That’s 3,500GB transferred/month.

Again using the highest S3 data transfer price of $0.17/GB, we’re looking at $595/month.

Add the per/request fees of $0.012/1000 requests. We’ll assume there were 15 objects/page, 35 million page views, that’s another $6,300 in request fees.

So, looking at storage and bandwith costs of serving the GeoCities requests we’re talking about somewhere on the order of $12,520/month. That doesn’t seem like a lot of money just to keep the geocities pages alive. To cover those costs you’d need to generate just 3.6 cents out of every 1,000 pageviews! It seems like some contextual links could have covered that?

I’m making a lot of assumptions, and to be conservative I’ve tried to estimate on the high side of things. They’ll also need to maintain some of their own infrastructure, or outsource the actual serving of the sites–but I’d think they probably have that down to a science. Yahoo would also likely need at least some staff to manage and support the service, and some staff to plan and manage marketing campaigns to your GeoCities audience. These salaries would far outweigh actual hosting costs—but still, seems like they could have made it work.

Ways they could have made money from that traffic:

1. Cross Sell other Yahoo! properties from the GeoCities pages. Use a small top-of page banner of some kind.
2. Serve their own PPC contextual ads on the Geocities pages
3. Remarket Yahoo’s hosting services to the Geocities site owners (they made a small effort in this regard at the end, but you could continue to market to your 7.5 million site owners month after month.
5. Add AdSense to the GeoCities sites — Heh, heh, just kidding but it would work

So, what do you think? What am I missing?

In the end they probably just decided the revenue upside was just not worth the effort. But it makes me wonder what other Yahoo! properties are in exactly the same situation.