Advertising was always difficult and always involved spending a lot of money without very clear or immediate results, but the dot-com era might have reached new limits of the ridiculous with companies burning money at stupid rates with only minimal results. Of course that sites selling advertising will still claim that they can provide amazing results and that giving them your money is the smartest business decision that you can make, but the truth is far from that and in this article we'll try to present some of the smart things that can be done and we'll let you see why other things are rather dumb. We will focus mostly on the software industry and all of our real-life examples will involve trialware/shareware programs.

It is important to understand that web advertising is a completely different type of media, and what was OK with radio or TV is very likely to be wrong in regard to the entire web philosophy of a "real two-way communication" in which the end-user is in control of the entire experience! It is also worth noting that in the long run this empowerment of the end-user first launched by the Internet will also put a serious pressure on the traditional media.

There are a number of things that make a big difference in how you should approach advertising, and probably one of the first things that might push you on a different path is the price of the products involved - for instance it is easier to categorize the products in maybe 3 classes - "high-end" (where the earnings for one sale are over 100-500 US$), "middle-class" (between 10 and 100 US$) and "low-cost" (under 10 US$, which might also create some additional problems on how the payments can be made on the web). Depending on those earnings you can estimate how much you can and how much you should invest in advertising - since it is obvious that you are simply losing money if each sale of a 10 US$ product will cost you 20 US$ in advertising!

Actually the price alone might be highly misleading - for instance with all the cola-drinks the earnings per sale are quite low (and well into the territory of "web micropayments"), but a second important factor can now be seen - the number of potential users - higher is obviously better and in theory the potential of the web is huge. However we have tricked you a little, the number of potential users is not always very important in itself, (and maybe cola-drinks are again a very good example) - even more important is actually the number of potential sales - meaning that some customers might buy certain products over and over again, which is a very important element for some markets and has certainly transformed cola-drinks into such a huge industry - but it's not always very relevant for the software industry.

Now that we have clarified those initial points it's time to focus on the web advertising part, and we'll do that starting with an important statement that will probably come as a small surprise for some people but as a very obvious observation to others - traditional branding/advertising doesn't work on the web! There are a lot of good reasons for that but since this article is about something else we'll only mention two things - first of all most humans are only able to remember a very limited number of brands (and the good spots are already taken) and second - the web is about the user making his own (very dynamic) choices!

While the previous statement (we'll repeat it - traditional branding/advertising doesn't work on the web! ) might seem benign at first it has a devastating impact on how you should look at web advertising - this actually means that all the companies that are trying to take your money in return for "building your image" might actually be a huge waste of money! However this does not mean that image is not important - it only means that you can not "manipulate" it as easy as some might like you to believe, and (very important) that your image is mostly built by yourself!!! What a surprise, your competence, determination and good service do matter after all :)

So the first obvious result is that you must sell a good product - and by good we not only mean "technically good" but also a product that answers to a real need and which is really appealing to your potential customers!

Another result (but this time one that might be less obvious) of the previous statement is that "web banners" and other methods with "low precision targeting" no longer work - or maybe it's actually more accurate to say that are worth very little, and certainly MUCH LESS than you will normally have to pay for them - 1 US$ for 1000 "impressions" means very often 0.95 US$ lost! You should also remember that "aggressive marketing" (like pop-up ads and unwanted email better known as spam) might also be a very dumb idea - not only will it be ineffective but most users will very likely get angry and your image will be dramatically deteriorated (destroying a brand is much easier than building a brand).

You are maybe asking yourself how do we know all those numbers, how do we measure things that are so difficult to count? Well, while advertising was traditionally more of an art than a precise science, web advertising opens amazing possibilities to evaluate and analyze the results, and that is especially true for the software industry!

When we speak about the link between advertising trialware/shareware on the web and making the final sale, the steps are simple and quite mandatory - the potential customers will first see the ads, a certain percentage will actually click on those ads and reach your website, some of them will then download and try your programs and finally only a small percentage of the users will return to your website to buy the program and become your clients! It is obvious that at each step there is a certain "conversion rate" which is usually low - maybe even as low as 100:1 for each step which might not look very bad until you do the math and see that for one sale you will need 100 downloads, for which you will need 10000 visitors that will be only generated from 1000000 impressions - that is one million ads that you will need to pay for before you will get only one sale!!!

After so many things that do not work you are probably anxious to find what on earth is actually working - and the answer is simple - very high precision targeting can bring very good results, and probably the best way for that is by using search engines!

First a few words on search engines by themselves - it's usually a very good and desirable thing to be ranked very high in the search engines since a very big percentage of the people that visit your site will actually come from such search engines (in certain conditions maybe up to 90%). Being ranked very high by a lot of search engines also means that your page (or site) is highly relevant for that search term - being relevant is good not only since it will raise your position (and the number of visitors) but because will also help you with the second conversion rate - from simple "visitors that are only browsing" to "potential customers that are downloading and installing the trial version for your program(s)"!

Not very far away from search engines are the "web directories" - being listed in such a directory will also improve your position with most search engines and is a very desirable goal but there are only two very big (and global) directories left and unfortunately both have some small problems - Yahoo is the oldest one but is becoming a little too greedy and at some point that is hurting its value and credibility - and Open Directory which is more dynamic but where some editors are more interested in promoting their interests even if that involves some lying - more details here. However the bottom line is that being listed there is good.

Advertising with the search engines also makes a lot of sense - when somebody is using a search engine he expects to get a number of links that might answer his/her problem and then he/she will follow those links - which means that the web philosophy of letting the user control his way is not disrupted - and even more important - the user is aware that he needs something and you can target that something with extreme precision!

Currently the most important sites in web advertising on search engines are Google (which is the bigger player) with their AdWords program and Overture (formerly known as GoTo - which actually is the much better/effective engine for advertisers).

Based on what/how you pay, there are two important categories of advertising - "pay per impression" and "pay per click" - the first method is pretty much a leftover of the dot-com bubble in which you have to pay a certain amount of money each time your ads are displayed (and you are probably right if you think that's pretty much "hype-oriented" since the payment is in no way related to how effective the ads were for you) - while the second is pretty much a "results-oriented" method in which you only pay for those ads that have effectively sent a visitor to you site. Now the bad news are that Google is (still) using the "hype-oriented" method (and as a result will be much less effective) while the good news are that Overture is a "pay per click engine".

Our opinion is that anyway you should do some tests with both of them (especially since the price for the tests is usually rather small) - but our findings below might help you get better and faster results for your money.

The first (and quite obvious) advice is that before you start advertising you need a good web site - that's somehow coupled with our previous result that you need a good product but unfortunately many good products only have poor web sites - which usually means a site that is too complex, too difficult, has too much or too little real/usable information, does not look serious enough or is mostly "under construction".

It is also important to decide early what kind of results you want from your advertising - of course that the ultimate goal is to make a sale, but that can be done in a number of ways - for instance you can target "impulse buyers" and push them directly towards the page where the payment is done, or you can help your visitors towards downloading and trying your product or you can first give them more information in order to decide if the product will answer their needs - the first approach will certainly work better with products that are not expensive or which are for short-term or single use (the best example might be some foods or drinks), the second method will probably work better with addictive products, while for programs where each download can cost you and/or the user a lot of money or time (including time for support questions) the third alternative might be better.

In most cases the best approach is to point each ad precisely to the specific page from your web site dedicated to the corresponding product, but if you have only 2-3 important products that are somehow targeting a similar market it might also work creating a simple page with all of them together and make all your ads point to this simple page (which can also be you home page), the user will then be able to use the links from this page in order to find more information or download the programs.

Now about the search terms and the ads themselves - it is usually not possible to use pictures in your ads on search engines (which actually is a very good thing) - and both the search term and the text of the ad should be highly relevant to your product(s) - that is especially true for Google where you will also pay for the inefficient ads but is also true for Overture since a visitor that only takes a short look and then leaves forever is equally ineffective!

Both engines will tell you that repeating the search term in the text of the ad will improve your results (mostly since the user will believe that the link is more relevant to his search and as a result will more likely click on it) but that can not be a substitute for being really specific and relevant. It is also important to keep the text quite simple and appealing while avoiding "cheap shots", for instance XDESK - the most addictive virtual desktop manager will work much better than XDESK - the best virtual desktop manager - when people see addictive they are intrigued and attracted but best seems more like self-adulation :)

With Overture there are two other aspects to consider - the precise URL for your link and the bidding. The URL should at least be a special hidden link on your webserver (for instance a page named "from_overture.html" which is actually a symbolic link to your real page) or much better - you should build very specific URLs containing both the name of the search engine and the full search term - for instance http://www.xdesksoftware.com/index.html?a=overture&q=virtual%20desktop might not look very friendly but will certainly help you a lot when you want to measure the results! (the full URL will be saved into the log and later a log analyzer will be able to use it).

If the search term is very popular on Overture some bidding might be required - obviously that the first position generates more clicks but the price might be a serious problem - we'll talk more about that when we'll speak about the detailed financial analysis but for the moment it's important to know that sometimes the second or third position might be much more effective - especially when the search term might not be highly relevant for your product - the result will be that only the users that are really interested in your product will click on the ad and as a result your effectiveness will improve!

It is also very important to understand how relevant your product is for a specific search term - for instance it is not unusual to see advertisers with more money than brains getting the top spot for search terms that have no or very little relevance to their products - this only means lost money for them but might also mean some saved money for you if you can use them to filter only the more serious visitors to your own site.

Another fascinating aspect of this entire "relevancy story" is the degree of awareness - it is very interesting to compare the results for two of our products that basically provide the same kind of features - improvements in the MS Windows user interface - SysTrayX will solve the annoying problem of the tray icons (the program makes possible to hide them, and also to recover them after a crash), while XFilesDialog will bring a LOT of improvements and new features to file dialogs - but even if XFilesDialog will actually improve productivity much more than SysTrayX we have seen much better results in targeting ads for potential SysTrayX customers - which we think it is closely related to the fact that people are much more aware that they have a problem (as we said, the tray icons can be annoying) and as a result they will try to fix it by actively searching after terms like "system tray", while much less people will search after "bigger file dialogs" and only very few will click on other search terms loosely related to file dialogs - in other words advertising on search engines works best if the potential customers are aware that they need something - if not, the result is only a more expensive banner ad :(

This is probably a good time to tell you that having access to the detailed logs of your web site is a must when you want to know if your advertising is effective, and you might also need some other small tools in order to analyze all that information - especially if you are selling software (but not only then). The first tool should be any simple "log analyzer" - we have initially used Analog since it is free (and open-source) but for more advanced things we have developed our own tools (XLogAnalyzer) that we'll probably sell as a stand-alone product soon. If you don't have access to your logs (or if you are not very interested in analyzing them) you can also extrapolate from our results below.

There are also differences between Google and Overture - the most obvious one is the price but we'll talk more about that later - and after that the most important is the "advertising approach" - if you don't have an unlimited amount of money to burn, with Google you can only use the most effective ads (and sometimes not even those) while with Overture it is actually better to have as many search terms as possible - you will only pay for the clicks! With Google is much easier to "catch" a huge number of impressions (for instance if you try to use a common word in their "phrase match") while for Overture you have to define pretty much each search term on its own and usually you will need over 100 for very good results! Another difference is that Google has a lot more users that Overture, but this can be both good and bad - you can reach more people but also you will spend money much faster!

Now is the time for some numbers - for the first "conversion rate" (from "viewer" to "clicker/visitor") our good results with Overture are around 10% (on average) while on Google pretty much the same input will only generate around 5-6% (the main difference is that in Google the ads are displayed in a separate corner of the browser window, while with Overture the ads are part of the normal results that are marked as "paid for" but and are much easier to be reached and clicked by the user). It is also worth noting that the results are also very consistent with the policies of the corresponding sites, Google is only interested in displaying your ads (and nothing matters for them after that) while Overture is more interested in people clicking on the ads (and as a result somehow closer to what the advertiser wants). However those numbers can vary greatly and a lot of Google campaigns will start under 0.5% in the initial form - the trick is to keep trying for another 3-4 "improved versions" (or even with other "test campaigns") - try new search terms, try a few (2-3) different texts for your ads and generally try to get a feeling of what works best for your specific market. If after those 3-4 iterations your Google results will not be over 1-2% you should think about dropping them and eventually only use Overture - where your click-ratio doesn't really matter in terms of money!!!

The tricky part starts when the visitors have reached your web site - and our results show that visitors brought by the ads from both search engines have pretty much the same conversion rate to "downloaders" - something around 10-20% - which is certainly better than the average rate from all other "normal" visitors!

In theory it is possible to also directly measure the final (and most important) "conversion rate" - the one from testing the product to actually buying it - but getting 100% accurate results usually is very difficult since it can only be done with a lot of effort from you and with a lot of "potential restrictions" for the customer - probably around 80-90% accuracy be achieved with cookies and minor restrictions but we have only used a simpler indirect approach that might be less accurate (we match the class of the IP address of the people that buy to the IPs of the people that reach our site coming from ads on the search engines) - and our results are close to the top of the average range reported for the shareware industry which is somewhere between 1% and 5% (of course with exceptions outside the range).

Now it's time for a simple calculation - with Overture good results are under 20% * 5% = about 1 buyer for 100 clicks , with more typical results of around 10% * 5% = 1 buyer for 200 clicks - which actually means that you should NEVER bid more than 1/200 from your product price (actually from the product earnings would be more accurate - so for a product that brings you 20 US$ for each sale you should never bid more than 0.10 US$ !!!).

Now the shocking truth - with Google very good results are 5% * 20% * 5% = about 1 buyer for 2000 impressions, with typical/good results of around 1% * 10% * 5% = 1 buyer for 20000 impressions - and since the typical price for 20000 impressions with Google is over 280 US$ this actually means that you should NEVER try a lot of advertising with Google if one customer will bring you less than 300 US$ !!!

So what does this mean?

It means that even the best web advertising, the one with the most precise targeting ever = web advertising on the search engines, the top of the top, even this one might only get very poor results - actually for most shareware authors it is a safe bet to say that Overture is the only option - and even with them a LOT OF WORK has to be done in order to reach profitability!!!

What can be done?

From the shareware authors' part - not much - probably the alternative is a lot of work invested in raising your position in the search engines, but that can be difficult and might also mean less work for your actual product so it is a tough choice... The final trick here seems to be either to have very low bids - which is no longer possible - shame on you Overture! - or to sell more than one product to each customer - having a lot of products (and some bundles) can help, and so do subscriptions!

And of course that Google should review their pricing policy - it might have been "almost normal" during the dot-com bubble but we now live in the real world, in which there aren't many cheap mass products where the average earnings start at 300 US$ / sale :) Probably a hint that something is wrong - even very popular search terms have no AdWords at Google!!!( but have hefty bids on GoTo!!!)

But there are some reports that suggest that web advertising works ?!

The very first thing that you should check in those reports is who is paying for them - and you will be surprised to find that all the reports that claim such things were actually ordered by some companies where over 90% of the revenues are coming from advertising - would you really expect them to admit that they are selling nothing more than "snake oil" and hype? The second but probably more interesting fact about those reports is that they no longer claim (as in previous years) that "web advertising works" but instead that "web advertising might increase brand recognition" - and they generally try to "prove" that with examples like Coca-Cola or Mercedes - that already have a HUGE brand recognition which is the only thing detected in their tests and then incorrectly attributed to "web advertising" :)

And yet I can see a lot of web advertising - is everybody throwing away money?

The shocking answer is "mostly yes" :) - a lot of people are simply not able to measure their returns from web advertising and they are rather hoping that a miracle might take place!

But of course other things might explain some of the ads:
- for huge companies like Coca-Cola the budget for web advertising is only a very small part of the total advertising budget, so they don't need to be very effective on the web; also large companies might get big discounts and the loss might not be so obvious for them;
- a lot of the ads that you might see are not really paid for with "real money", but instead "exchanged" or paid with another service that is equally overpriced;
- finally - a lot of companies that still pay for web advertising really hope for (and some might even get) more than 500 US$ earnings from any new customer - an expensive car or top of the line computer might bring much more than that, and if a bank or credit card company can transform you into their customer they also will probably recover all costs.

Since some things have changed after we first published the article there is an update / extension here