How often do people break downloads?

[BongPig]

Is it a common thing for people to decide to stop a download before its finished?

If 100 visiors clicked to download one of your demos ( assuming you have one ), how many would make it to the end of the download?

Can anybody share any figures if they have any?

We get around 60%+ broken. Ive never really understood why.
We're talking about 4mb files.

[Jack Norton]

They're probably using download manager like Getright, they seems to split or break the download because those programs use resume download or split the download in several parts.
If you have many of those, it means probably that lot of users have dialup connection ;)

[Bluecat]

Yep. Some of the new downloaders open several connections to the file in order to download it as fast as possible. Some programs try and open as many connections as possible. (I think Bittorrent works like this.)

I believe that it is possible to set up your download to limit the number of connections. I'm not sure if you can do it dynamically so that you can provide an 'unlimited' number of connections when your site is quiet, and drop it to one connection when you are experiencing a lot of downloads.

cheers

John

[BongPig]

Right.
I never considered that.
So its not broken downloads at all. Simply multiple hits from the same user grabbing the same file using some kind of download manager.

Cool. Thanks for that guys. :D

[Roulette]

Is it a common thing for people to decide to stop a download before its finished?


I get a lot of these as well...so many that my download reporting software takes it into account. Thus, if I have 5 downloads appear from the same IP within 10 minutes, for example, it's all counted as a single download.

On an interesting note...when I recently had a game hosted on download.com, I noticed that their download numbers were running about 50% higher than these "true" numbers that I was generating (where I discarded multiple downloads from the same IP over a small period of time.) Hence, I suspect that download.com simply tallies each one as a separate download. This of course gives an inflated estimate of how many downloads they are generating, which in turn would help to explain why the sell-through percentage on download.com tends to be considerably lower than normal.

- Roulette

[arcadetown]

Comparing our web log .exe download hits to the actual bandwidth usage we typically see from 30% - 50% turn into completed downloads with smaller games getting higher completion rate. For example yesterday Platypus @ 5.5mb saw approx 934 hits / 1.41gb transfer => 260 completions so about 30%, Betty's Beer Bar @ 3.2mb / 894 hits / 0.94gb => 290 completions so about 33%, while smaller BLOX Forever (online playable so less downloads) @ 0.8mb / 180 hits / 0.076gb => 90 completions so about 50%.

@Roulette - what kind of % decrease do you see reported by your software compared to web log hits? Would guess not too many people use download accelerators?

[Roulette]

@Roulette - what kind of % decrease do you see reported by your software compared to web log hits? Would guess not too many people use download accelerators?

I see roughly 10-20% of downloads occurring more than once by the same IP within a period of a couple of minutes. Also, when this occurs, the number of "downloads" by the same user varies dramatically...today the range was from 2-13. It doesn't take many 7s and 9s and 13s to totally screw your percentages if even a moderate percentage of your users are doing this. If I had not taken these "repeat downloads", or bandwidth accelerators, or whatever you want to call them into account, I would have registered just over 37% more downloads today than I actually received.

- Roulette
http://www.superluminal.us