An important part of the work that we carry out for our clients involves getting their software listed on "the important" shareware and software sites. The problem is identifying the sites that are worth submitting to.

Some believe that there is no such thing as a site not worth submitting to. Others believe that only a very small number of the sites are worthwhile. The fact is, however, that while there is no shortage of sites to submit to, the time required to do so is another matter altogether.

The research.

In August 2003, we decided that due to the ever-increasing number of shareware and software sites, we needed some way to ascertain the value of each.

We then decided on a fairly simple means of doing so.

We combined the raw web logs from one single month of some fifteen of our clients, and then wrote a basic application that would search for the names of referral sites, then group and count them accordingly. The software also searched for all the names of known shareware and software sites. If there were no referrals whatsoever, these were listed as zero.

The main drawback of this technique was that a site generating a high number of referral counts might not indicate any visitors at all. Some sites display data (for example icons) that are pulled from the developer's website, and this would then be counted as a hit. Our method would not allow for differentiation between a hit and a visitor.

Other issues to take into account included the way that some sites refer to other sites, the fact that some sites may use a different referring domain name, the fact that some sites host their own copies of the files, the sites that only list a small number of specific applications and so on. The list of exceptions proved to be extensive to put it mildly, but these were then removed from the list altogether. This process was carried out manually, as the last thing we wanted was to stop working with sites that can and do generate a

reasonable level of traffic.

Once complete, we still couldn't make any assumptions about the top end of the list. The bottom end of the list, however was a different story altogether.

The results.

We discovered that some of the sites that our clients' products were listed on were being counted with very low figures; for example many were in the 1-5 range.

What did this mean? For the sake of diplomacy, let's create a fictitious site - www.superblycoolsoftware.com, which generated a total of 5 referrals. Let's also assume that this site listed a total of twenty of our clients' applications.

This means that when we combed the web logs from fifteen different clients, www.superblycoolsoftware.com generated a total of 5 referrals; which may have been displayed images or actual visitors. Even if all 5 were actual referrals, resulting in 5 visitors, a total of 5 visitors to 15 different sites over a period of one month is a little on the low side.

This now presented us with some extremely useful information, and once we'd excluded some of the sites with particular circumstances, we could clearly draw the following conclusions:

(1) That a large number of the sites listing our clients' software generated an incredibly low number of referrals.

The obvious counter-argument to this is that if a site generates even one visitor per month to a client's website, then wouldn't this be worthwhile?

The answer is no, not at all. If one site generates one visitor per month, a small fraction of these will explore the website to any level of detail. Of these, a small fraction will download the software. Of these, a small fraction will actually purchase the software. When you look at a small fraction of a small fraction of a very small number over a long period of time, you're looking at possibly waiting for many years (or longer) to generate a single sale, if at all!

(2) That blindly submitting to hundreds of shareware sites is a waste of time.

Even if sites are quick and easy to submit software too, time is still required to do so. We have seen software that claims to submit software to around 400 sites, but aside from anything else, this still takes time. The time taken to submit to (possibly) hundreds of sites that generate almost no visitors or downloads is time wasted. Most companies can find better things to do with their time than waste it.

At this point we reached a difficult decision; what to do with this information.

We realised that although one end of the list was highly questionable, the lower end of the list pointed to a fair number of websites that simply do not generate traffic of any sort. It wouldn't be too difficult to argue that submitting to these sites is completely pointless.

However, we realised that by releasing the details of this information to the public, we would be pointing a very damning finger at a fairly large number of websites. For reasons of conscience, liability and peace of mind, we decided to keep this information within our company.

We ourselves can now choose where to submit our clients' software with far greater precision, and as well as optimising the actual submission process, this also allows us to more accurately utilise one of our most important assets - time.

So how many sites are worthwhile?

Drawing a line on how many sites are worth submitting to wasn't easy. The results of our analysis suggested that submitting to the top 30 or so sites was critical, and if we wanted to maximise our efforts, then the top 50 or so sites appeared to be worthwhile.

For our long term clients, in the interests of being certain, we then decided to add the next 40-50 sites as well.

Our research suggested that beyond the top 100 sites or so, it simply wasn't worthwhile.

Note that we continue to monitor these findings on an ongoing basis on behalf of the companies that we work for.