Adventures in Online Advertising

This article summarises what I have learnt from introduction of advertising onto El’s Extreme Anglin’, a guide to fishing in the World of Warcraft (WoW). It introduces internet advertising with discussion of the earning potential, cashflow and ethics. The article then provides a series of case studies on specific topics, such as iteratively improving revenue, altering placement, cloaking, use of text or image adverts, and seasonal variations over Christmas. It should offer a useful introduction for those attempting to monetarise medium-sized websites.

On this page:

Internet Advertising

Conventional internet advertising allows publishers (that’s me) to earn revenue from several models:

  • Display – Cost Per Mile (CPM), the cost per thousand page impressions.
  • Click – Cost Per Click (CPC), the cost per user that clicks on the advert.
  • Action – such as completing a form on the advertiser’s website.
  • Sale – commonly via an affiliate model, where the publisher earns a proportion of the sale price of the goods/services they advertise.

The advert itself typically takes the form of images or pure text, with a link to a product or service site.

Earning Potential

Earning potential most obviously varies with the audience of the site. A venture capital blog commands a rate (to the advertiser) of $30 per thousand displays: The audience tends to be wealthy and influential. A popular online comic earns just a few cents per thousand: The audience tends to be rather young, with little or not spending power.

Gaming sites are typically at the lower end of that scale, although WoW sites attract better rates than other games sites, probably because the game tends to attract an adult, somewhat educated audience, who have a reasonable income to spend. The downside with WoW gamers is they spend a lot of time online (WoW is an online game), so become rapidly over-saturated by online advertising: Websites that can target users that otherwise don’t browse the internet much, are logically more attractive to those running advertising campaigns.

Note that online advertising rates are currently much better than they were after the dot-com collapse. For example, Hot-Or-Not was down to 3 cents per thousand displays at one point.

There are other factors beyond the audience.

The balance between:

  1. many visitors that visit infrequently, and
  2. fewer visitors that visit regularly,

is crucial to understand.

In the first case, advertising for affiliate products (and probably also Real Money Trading – see below) can be highly successful: Lots of potential customers might only see the product advertised a few times on the site, but then they will not buy the product more than once, however many adverts they see. Where there are fewer, but more loyal people visiting, the overall number of product sales you can is far lower. Sites in the second category should look to larger advertising networks, if only to gain variety of ever-changing adverts.

WoW attracts Real Money Trading (RMT) businesses, which operate outside the game operator’s terms of use, and so have limited means to advertise their services. Websites about WoW can opt to advertise RMT services (exchange of in-game currency for real-world currency), or related services such as power-levelling (third party advancement of a character through the introductory and middle stages of the game).

This is a very attractive proposition financially: Using Google’s Adwords traffic estimator as a benchmark, the phrase “wow” averages $0.50 per click (to the advertiser), while “wow gold” averages $5 per click. Simplistically for WoW audiences, RMT advertising is worth ten times other advertising. Those figures are from August 2007, and the factor of ten may be biased by prices for search result adverts, rather than adverts on websites. However, it is indicative of significant differences for RMT and non-RMT advertising.

RMT advertising is regarded negatively by many players: Those that influence others to visit the site will rally against it; Blizzard, the game’s operators, will want nothing to do with the site; and generally all the free viral referrals and goodwill that generate the site’s long term audience will fade away. Sadly RMT advertising remains be a great exit strategy for many webmasters who want to “cash in” quickly.

Contextual and behavioral marketing is not yet common in mainstream display advertising. However, on specialist websites it may be easy to target adverts to the audience. You don’t need advanced Google Adwords technology to conclude that a site about a specific video game (such as WoW) might be suited to advertising products related to that game.

Typically at least a third of the amount paid by display advertisers will be taken by an advertising network. In contrast, affiliate networks typically take less than 10% of the revenue from sales. This reflects the balance of risk and cost: Using affiliate advertising, the publisher carries all the risk that the product being advertised may not sell. In contrast, most advertising networks manage to fill most of the publisher’s advertising space (inventory) with revenue-generating business, and take the difficult decisions about how to display advertising.

Placement of advertising is also key to the success of adverts, as discussed later in the article.

The Cashflow Catch

Most affiliate and advertising networks only pay out once a certain amount has been earnt, commonly $100. If you don’t reach the threshold, you won’t see any money. That is a problem on low-traffic websites, and for higher-traffic websites that spread their advertising out across too many networks or channels. Particular care should be taken on affiliate sites with low margins: Amazon’s 4% affiliate cut (on sales worth around $20 each) takes a very long time to reach the $100 threshold for writing a cheque (for those without a US bank account).

If you spread advertising across many products or networks, it is easy to accumulate hundreds of dollars of advertising-related revenue that may never reach your bank account.

For this reason, if a site has the potential to earn the publisher $1 per thousand page views (a realistic expectation for many sites), that site really needs to be serving thousands of page views each month (to people, not robots) to make advertising viable, and hundreds of thousands before generating a serious flow of cash.

Most networks operate in US Dollars. This can be a headache for whose living outside the US, particularly if payment is made by cheque: Physical post is slow, and in the United Kingdom (at least) the process of banking a cheque written out in dollars transpires to be arcane and expensive. Even on a network that writes cheques out monthly, it may take up to three months from the point of revenue generation, to the point that revenue is sitting in the publisher’s bank account.

Read the terms related to payment very carefully. The adverts may be distributed instantly around the globe; but the revenue from them isn’t.

Advertising Ethics

Don’t apologise for advertising! Publishers that make excuses for their decision to display advertising on their sites seem to misunderstand the nature of the partnership they’ve entered into: The advertiser is paying (indirectly) the publisher money – surely there should be a little bit of respect for what the advertiser is trying to do?

Personally I see advertising revenue as a positive thing: It means that I can dedicate a lot more time to maintaining and building the site (and write long articles about it all), while still managing to pay the rent.

Should publishers accept any advertising? Accepting adverts that readers will dislike or are upset by serves nobody: The advertiser is probably wasting their money, and readers will be less likely to return to the site in future.

Should publishers feel like they are endorsing the products/service advertised? I’m neutral to the adverts I run: Consequently, I aim to keep a clear separation between advertising areas of the page, and content areas of the page, to maintain that neutrality.

Sadly there is a conflict here between the effectiveness of the advert, and its similarity to the content of the site (as discussed below). It is exceptionally tempting to blur the line between advertising and content. When that happens, I’d argue the author of the page starts endorsing the advertised material.

Endorsement could go much further: I could, for example, include affiliate guide links in the content of the text, and make them look precisely the same as links to other pages. Perhaps I’d sell a lot more paid guides. But perhaps I’d also loose the trust of readers. I think that trust is more important in the long term, than a revenue gain in the short term.

Case Study: Iterative Advertising Performance Improvements

Regular readers may already know that advertising on El’s Extreme Anglin’ is (currently) affiliate-based, selling guides to playing certain aspects of World of Warcraft. I call it Learn2Play. It took two months to appreciate the full potential of these for generating revenue.


El’s Anglin’ had been online for just under a year when I started to implement advertising. At the outset, I wanted to use the website as a sandbox to learn more about advertising, and to provide a benchmark for what the advertising space might be worth. So inevitably I found myself looking at affiliate advertising, where the risk is taken by the website owner. I wanted to keep RMT advertising out, which is not possible on all advertising networks (it is very hard using Google Adwords/Adsense, for example). The topic of the website was very specific, and naturally steered my choice towards gaming-related products.

I initially selected two types of affiliate product to advertise:

  1. The WoW leveling guide – Kopp and Joana, which seemed the most popular pair.
  2. Another Massively Multiplayer Online Game (MMOG) – Eve Online, which again seemed popular.

There was no tangible backlash against adverts. Nobody mailed me about it. Site traffic didn’t drop. My conclusion: People just accept advertising.

In the last three weeks of July 2007, 170,000 advert views generated just 6 sales, but earning almost $90. Those headline figures hide a lot of patterns:

  1. Leveling guides performed a lot better than the “other MMOG” adverts. Click through rates were twice as high for leveling guides (about 0.6% – arguably still on the low side). And conversion rates were also better. Plus the payout from leveling guides was higher. Overall, the Eve adverts were earning the equivalent of less than $0.10 per thousand displays, while the leveling guides were generated about $1 per thousand.
  2. Index pages – notably the front page – didn’t generate as many clicks on adverts as other pages. My conclusion was that the amount of time a reader spend on a page influences whether they click on adverts.

Improving on Guesswork

Learning from July, the approach in August altered:

  • Adverts for the “other MMOG” were phased out.
  • Adverts for leveling guides were increased. Selling a game guide guide from a game guide clearly works.
  • Adverts for WoW “gold-making” guides were added. WoW-related content was clearly popular.
  • Index pages started to host adverts for other parts of the website, rather than commercial advertising. Perhaps this would encourage people to look at the adverts on other pages a little closer?

Gold-making guides transpired to generate about twice as much revenue per page view as leveling guides: In less than two months of initial experimentation my best idea for advertising earned 20 times as much as my worst idea. The magnitude of that difference is the reason why iterative improvement and experimentation is so important.

Not everything I tried worked. For example, I had originally focused gold-making guides adverts on pages about topics more likely to read by players with high-level characters, while leveling guide adverts tended to appear on pages that I thought would interest players with lower-level characters. My logic may or may not have been sound – certainly there are still a lot of players that don’t start to learn about fishing until they reach the highest level in the game. But I could find no evidence that targeting adverts to pages specifically created sales leads.

Finding a Balance

Of course that doesn’t mean that the entire site should be turned over the gold-making guides. And this is where the real fun starts! The aim is to optimise the proportion of total advertising space allocated to different affiliate adverts, such that it maximises overall revenue. Visitors rapidly tire of seeing the same advert on every page, so this is not simply a case of plastering the site with the best performing advert.

Clearly the process is very specific to individual websites, and far from scientific. It is very important to identify a few key metrics and monitor them month-to-month (more frequently on high-traffic websites). For example, I particularly watch:

  • Revenue per Mile (thousand page views) – Could a “proper” advertising network make more money than me from this advertising space?
  • Clicks per Page View – A particularly useful metric for determining the optimum advert position on the page.
  • Conversion Rate (at the target site) – How well is the product’s website doing at turning referrals into business?
  • Revenue per Click – The absolute value of those conversions is still important. A site that converts at a high percentage, but only pays out a few cents, may be contributing a lot less than a site with poor conversion on a product that generates a lot of revenue for the affiliate.

Remember that as an affiliate advertiser, you can control the type of advert displayed, the position, and so on. But if the target website converts poorly, there is probably nothing you can do about it.

It seems to be the case that adverts decline in popularity naturally over time. However, I can’t yet clearly demonstrate that, because:

  • The products I’m advertising are also changing in popularity over time – as game balance alters, and layers characters develop to higher levels with differing needs.
  • The volume of repeat visits to the site is changing, and repeat visitors will have already seen the advert.
  • During 2007, a lot of other websites (and search advertisers) noticed the revenue to be made from advertising these guides, so there is more competition for customers from other sources. (These are products each person will only ever want to buy once.)

Case Study: Advert Placement

Does the placement of adverts make a difference? The wisdom of the web says yes.

Jakob Nielsen’s conclusions on banner blindness summarise the situation pretty well: If it looks like an advert it will tend to be ignored – the most successful advertising is that which looks like the content of the site. That, or course, raises an ethical dilemma.

Heatmaps have been used to show that visitors tend to look first at the top of the page and the left-hand column of the page (since this is where the navigation and links to content tends to be). Consequently adverts on El’s Extreme Anglin’ have always been placed in those two positions.

Adverts at the top and left of the page are not equally good performers.

For example, I started advertising Team iDemise’s leveling guide using a large banner advert at the top of the page. The click through rate (over 35,000 page impressions) was 0.22%. The advert was then shifted to a wider skyscraper design, on the left-hand side of the page. The click through rate rose to 0.34%. Conversion rates remained high, giving higher revenue/mile for the left-hand position.

What’s most interesting about that example is that many of those left-hand adverts were displayed against the part of the page not initially visible on page load. Advertising networks will often tell you that all advertising should be close to the top of the page. But it seems left-hand adverts remain effective further down the page where there is content that will cause visitors to scroll down and read.

Again this highlights the importance of experimentation with advertising.

It also highlights the importance of a flexible visual design: The ability to add and remove advert positions without making the page look messy should be part of the original design specification. Too many websites try and “shoehorn” advertising in to a design template that was never intended to hold advertising.

Cloaking affiliate links involves creating a link to a page on the same domain as the content page the advert is shown on. When opened, the linked page immediately redirects to the affiliate site. Although I term this cloaking, I have never tried to mask the fact the advert is an advert. (There are implementations of the approach in use on other websites that are more likely to confuse users.)

Originally all links to affiliate sites were coded such that the affiliate tracking information was visible to the reader clicking the advert. For example, takes one through Clickbank, who track the affiliate referral, and then on to Luke’s Gold Making Guide. At first glance, the Clickbank aspect (the affiliate network) is confusing to the user.

Partly to avoid that confusion, and partly to make tracking and administering advert clicks easier, I linked all the adverts to redirect pages. So, redirects to the Clickbank address above, which in turn redirects to Luke’s Gold Making Guide.

The result: Cloaking affiliate links doesn’t make any difference to click-through or conversion rates.

The logic: Users know they are clicking on an advert, because it looks like an advert. They expect the click to take them to another domain, and for that to be tracked, because that’s how online advertising works!

Case Study: Text vs Image Adverts

There is a popular misconception that image-based advertising gives inferior results to text-based adverts. To test this, I experimented with adverts for Derek’s Gold Mastery guide.

Derek’s website converts visitors quite effectively (2-4% of visitors to the site buy the guide), however click-through rates were lower than for some other “gold-making” guides: It was an obvious target for me to try and improve. During the months of September and October 2007, image-based adverts were in use – almost entirely “wide skyscrapers” on left hand side of the page (this image). In November I switched to using text-based adverts, occupying the same left-hand position. The text used was precisely the same as that on the image adverts. Just over 100,000 advert impressions were served each month for this affiliate product.

The table below shows how some key metrics varied between months for these adverts.

Month Advert Type Clicks per View Conversion Rate Revenue per Mile
September Image 0.44% 2.5% $2.30
October Image 0.41% 2.2% $1.90
November Text 0.31% 1.9% $1.30

The conversion rate is the proportion of visitors referred to the affiliate that go on to buy. The revenue per mile is the money generated for every thousand advert views. One might reasonably conclude that all these metrics were already in decline month-to-month, and so the further decline in November was not caused by text advertising, and would have happened with any form of advertising.

In December image adverts were re-introduced. This raised the clicks per view back up to 0.49%. However, due to the Christmas Hiatus (see below), revenue fell sharply.

While I cannot conclusively say image advertising is more effective than text, there is no evidence to support the argument that text is superior. Text adverts do reduce bandwidth consumption, and occupy less of the page area. Text adverts may also allow pages to be loaded with more advertising, so could still contribute generate more revenue overall. Perhaps that will be tested in future.

Case Study: Christmas Hiatus

December is a curious month in western societies, when patterns of human activity and consumption become skewed away from the norm: Time away from work or school, unusual travel and shopping patterns, and more parties and gatherings.

The impact on affiliate sales transpires to be very significant. From the beginning of the second week in December until Christmas, and again over the new-year period, transactions drop to zero. The propensity to visitors to click on adverts remained constant. But those visitors became far less likely to buy anything. This isn’t a small change: It is a shift from a situation where every day there are one or two sales, to a two week period where no sales occur at all.

Whether such dramatic changes occur for other websites, or whether this is a quirk of the topic or product sold, is not known. Observation of in-game player activity suggests a huge drop in the numbers playing the game once universities and colleges finish for the Christmas break.

The Christmas hiatus tells a cautionary tale for those inclined to base their revenue assumptions on their “best” month. Those webmasters will not have a happy Christmas.

Concluding Tips

It is rather too easy just to signup with Google Adwords/Adsense, and never think about how to optimise advertising on a website. For that reason alone, I’d recommend starting with affiliate advertising, purely because you carry the risk, so you feel responsible for making sure your advertising space works effectively.

I’m still experimenting and learning, but these final tips might still help those new to the topic.

  • Advertising is not the only way to generate revenue from a website (that is merely what this article discusses). And if you are “struggling” to cover $100 of annual server hosting costs, you might find it easier to move to a free host (WordPress, Sourceforge, Google, whatever is appropriate), rather than spending time arranging advertising, or feeling like you need to apologise for displaying advertising.
  • Design pages to fit advertising, even if you don’t use adverts right away. Sites where adverts have been dropped in as an after-thought, rarely look good.
  • Wait for traffic to build up before adding advertising: Administering it all does take time, and it will be wasted if none of your initial ten readers click anything. Although unproven, I suspect sites without advertising are more likely to be regarded as hobby projects, which naturally attract viral referrals.
  • Decide what type of advertising to accept, and how much you are prepared to integrate advertising into content. I don’t believe it is worth sacrificing long term sustainability and user trust, for a short-term dollar.
  • If using an advertising or affiliate network, check the terms. Most will only pay out once you’ve earned $100 – and that might take forever.
  • Expect to take up to 3 months from the time revenue is generated to the time it is available for you to spend.
  • Experiment with different forms of advertising. There’s a huge (factor of 20) difference between my best and worst performers, and I could not predict that difference. The great thing about websites is you can constantly change and improve them, without having to start over.
  • If you’re serious, avoid video games. It’s the Warren Buffet rule: It is fundamentally tough territory to make money from – highly competitive, yet not much money to be made from advertising. You may do better for less effort in a richer market. Of course it changes the whole model too. For example, El gets a lot free promotion – prime spots on Google for most common terms, and I’m constantly amazed by the “viral” effect.

Lastly, people using the affiliate model to sell their products need to keep in mind two things:

  1. Focus your activity on improving the conversion rate of your site, because that’s the one part of the equation your affiliate advertisers have no control over. Free banners and morale support are nice, but all other things being equal, a product or service that doesn’t convert effectively, will not get advertised.
  2. You are competing with other affiliates and other forms of advertising for advertising space on the internet. That means you have to price your product high enough to be able to offer the affiliate a sufficient return for them to advertise your product and not something else. I see this a lot with new World of Warcraft guides: Their authors prices them at a few dollars, possibly because that’s what they perceive the guides cost to write, or what their friends tell them they are worth. However, affiliate advertising typically dictates a click-through rate below 1% and a conversion rate in the order of 1-2%. And if every sale requires 10,000 advert impressions, you have to offer the affiliate double-figure dollars on each sale, or they will go elsewhere. If your product only sells for $5, you can’t pay your affiliates $20.

3 thoughts on “Adventures in Online Advertising”

Comments are closed.