I played with Admob a couple of times the last week. Admob displays advertisements on the iPhone (mostly in free apps).

You pay for the number of times someone clicks (taps?) on your ad. Advertisements for iPhone app on the iPhone are restricted to an icon + text, which is great because it keeps everything simple. You bid a price and the more you bid the more likely you ad is to be viewed. The whole system is a simplified version of Google's Adwords.

Anyway, I decided to play with $500 for Strip Designer -- Not exactly a stellar budget for an ad campaign. Campaigns require a budget of at least $50 per day and minimum bids in the US are $0.05 per click.

I started off with a world wide campaign. I set the budget to $125 day and I decided to let it last for 4 days. The $125 was spend in less than 2 hours, but I had set the bid to $0.10. That was way too high. I read somewhere, that Admob is supposed to divide the display of advertisements equally over the day, but I guess it can be a bit difficult to predict how often someone will tap on an advertisement and if the purpose of the bidding system is to show high-bidding ads more often, then doesn't that work against the "distribute over the course of a day"?

After two hours I got what I thought was a good amount of clicks and impressions. 2000 clicks and about 200000 impressions. That is a click through rate of 1%.

After that, I started to think about what I wanted to achieve. The myth on the iPhone app store is that the higher you get on the top-ranking lists the more money you make because you increase your visibility. So I thought it would be best to burn all the money in a short amount of time.

I did various experiments both on the same day as the first campaign, a couple of days later for US only, and again a couple of days after that. In total I got around 1000000 impressions and just under 8000 clicks (CTR 0.8%). I don't really see any reason to optimize the CTR since the money was quickly spend anyway, and I got the clicks I wanted. The total click price was slightly higher than the bid I gave. I would also note that impressions were better divided over the entire day during the second and third runs of the app which could indicate that Admob becomes better at predicting. I guess they need to predict your CTR.

I would say that it is naive to think that you'll get anywhere with a $500 budget. If you pay $0.05 per click (which is minimum) you get around 10000 clicks. Most of those people will not *buy* or download your app even if it is free. They'll quickly click away. Let's assume 2% actually buy it; That is 200 downloads. That will get you to top 25 of the photography section, but probably not top 10 or anywhere near the top 100 paid apps. And that'll only work if you get 200 downloads in a specific region. On top of that, there is probably a fair share of click fraud involved.

How much gain did I see? I saw a few additional downloads in some countries of Europa and Asia that I cannot attribute to anything else, but it was nowhere near an additional 100. Maybe 50. When I restricted the campaign to the US, the extra downloads got completely lost in the noise.

What I learned from this is that you either have to bet a lot more money or you should advertise for a free app, which is more likely to get downloaded. My feel is that typically, 4% of those that download a free app will also buy the full version. So if you can get e.g. 10% of your clickers to get the free app then you end up with 0.4% new buyers. For $500 that will give you 40 new buyers. That is in the same order of magnitude as I think I got.
That is something think about. If you want to get into the top 100 you need to get at least 1000-2000 new buyers so you might be looking at spending $10000-$25000. Of course you can potentially get a lot of that back again, if the app your are advertising for is Admob based.

Basically, I wouldn't recommend doing this unless you have a lot of cash for PR that you want to burn. Perhaps a complete strategy with advertisement(s) + discount offer + free app for a period of 3-7 days might work.