Words Of Wisdom

September 30, 2007

What The Heck Is Adbrite’s Auction Mode?

Filed under: AdBrite — S @ 3:01 am

I don’t think I still get it how exactly it works. I have been using it for just a few months and with no more than a handful of clicks and a few interstitials I don’t really have much revenue to analyze the efficiency of using the network. But here is what I was for a big surprise last week. There was a click whose max CPC was 30 cents and I got credited only 4 cents. For all other clicks I got 4 cents, the max CPC is 5 cents. So, till I saw that I got only 4 cents for a 30 cents max CPC ad, I didn’t bother much about the whole “auction mode” idea.

Anyway, so, if I understand it correctly, the reason why I got only 4 cents and not like 20+ cents (subtracting their commission out of the total 30 cents) is, it appears that the way AdBrite charges the advertiser is just 1 cent more than what other competing ads on my website (yes, my website) are willing to pay. As the minimum price for a CPC ad starts with 5 cents anyway, so no matter whether a CPC ad is 5 cents, 40 cents or 10 dollars (yes, there are CPC ads shown on my site that have that much max CPC!), the click is credited at 4 cents for me.

Apart from the fact that AdBrite ads are not contextual and hence, their CTR is typically less than that of Google’’s AdSense or other contextual ads, this seems to be a big bummer. While there seems to be a potential to make a varying amount of money per click on other networks, it seems to be pretty much fixed and based on the collective demand for your website than the individual advertiser is ready to pay for it.

So, while on the surface this seems like a bad deal for the publishers, I guess in the long run, this makes it more appealing for advertisers and hence increases the demand for ads within AdBrite’s network. I think, one thing AdBrite should probably do is instead of saying that the cost is going to be just 1 cent about the second highest bid on the website, it should be a combination of that and the max cost of the ad that got displayed and got the click. So, for my example, a charge of some where between 5 cents and 30 cents would have been much more beneficial to the publisher.

Infact, as most websites will be at 5 cents in the auction mode, advertisers might even choose to bid for higher CPC as they know that their target network might not be receiving that high cpc rates and hence get maximum exposure by paying little. But this is partly solved by AdBrite’s practice of picking the ad that has a better CTR than an ad that’s ready to pay more money. So, the onus is on the advertiser (in addition to the publisher) to make sure that their ad is compelling for enough users to click it at a decent rate.

All in all, for a non-contextual ad network, AdBrite still seems to be the best choice. I have tried a few others such as AdEngage and Bidvertiser, but I seem to get more CTR from AdBrite (for the same content, same website) than the other networks. In addition, I am also able to make a few extra cents from AdBrite’s interstitials which don’t have to be clicked. Ofcourse, with interstitials starting from 0.001 cents, understanding how these are charged with the auction mode concept still beats me. I will write up later if I find more info. I wish AdBrite had a better help with some sample calculations instead of just stating that their ads are always in “auction mode”.

September 27, 2007

Why The Clicks Do Not Tally?

Filed under: CPC, statcounter, web logs — S @ 5:18 am

I see people wondering that the various reports of clicks don’t tally.

A typical observation for people who use Google AdWords to drive traffic to the product/site for which they are affiliates is that the number of clicks in AdWords report don’t tally with the number of clicks in the report from the site offering the affiliate program? So, how can this be?

Another question I have seen is that the clicks don’t tally with their statcounter or some other web statistics counter reports. Why does this happen?

Here is what happens. When a user clicks an ad, the following things happen

1. The link request goes back to the ad server
2. Which then validates, records the click and redirects the user
3. The redirected url then takes the user to the website of the ad
4. Some agencies and other secondary networks might be taking money from you and advertise on your behalf on the primary networks. In this case, there is going to be multiple redirections.
5. After all the intermediate redirections, with always at least one redirection, the request from the user browser finally is made to the web server of the ad
6. The page is then rendered by the browser and along the way, picks up any website counter javascript
7. This javascript code results in another request to a counter website like statcounter
8. The counter website then updates the stats for the counter.

So, as can be seen, in the entire process, there can be several points of failure each of which contributes to a lower counts in the down stream. As long as the moment the ad server records the click and redirects till the stats counter is updated, if there is a) network failure or b) server failure or c) user not letting the page to completely load so that the stats counter’s request is complete, these clicks are not recognized.

One of the way to increase the accuracy is instead of relying on a 3rd party counter, rely on your website logs. Web site logs are more accurate because any kind of failure between 6 and 8 above doesn’t matter. Your website’s access log already would have registered the click by then.

If you are investing some serious dollars into advertising, here are some ways to ensure you have proper accountability

1. Don’t use intermediate agencies/networks that only adds to further redirection and increases the chance of failure. Think of it as having to take multiple flights to reach a destination. If any of the flight is delayed, you won’t reach your destination on time (and in this case, the user won’t reach at all and hence may end up clicking the original ad again or may simply click your competitor’s ad or just leave)

2. Don’t rely on 3rd party counters. Rely on your web server logs.

3. Use special urls for your ads so that they can be uniquely tracked from rest of the accesses. Even if your landing page is static, simply adding extra parameters to the url such as

halloween_costumes.html?channel=google&campaign=halloween&period=october

where the halloween_costumes.html is a static html page still records the parameters in the access logs! You don’t need a dynamic pages website to track your ad clicks!

September 23, 2007

Inline vs IFrame

Filed under: Ad Blocking — S @ 9:17 am

Earlier I wrote about The Future Of Online Advertising in which I mentioned that the ad code can be combined with the rest of the page on the server side. However, the problem with this approach is, it’s possible to use JavaScript and simulate a click. This is not possible with the current practice of using iframe to display ads because, it’s not possible to use a javascript in one domain’s page to inspect and execute against another domain’s page.

September 16, 2007

Reducing The Effort Of Affiliate Programs

Filed under: Affiliate Programs — S @ 9:01 am

One of the problem with affiliate programs such as CJ and LinkShare is the constant to need to login to the site, get the current promotions and then update the site with those snippets of html/javascript code. This is a laborious task and unless there is a good payout not worth the effort for small websites.

An alternative to this is to make use of web services such as those provided by Amazon.com to automate the content generation or use simple RSS feeds like those provided by Buy.com and create a store out of it. For example, check out this RSS feed based Buy.com Affiliate Store that generated html pages out of the Buy.com’s RSS feeds.

The choice of which route to go to pretty much depends on the type of content on your website. If you have a blog on a particular topic, hand-picking the individual products and creating affiliate links to specific links may be worth the effort. But if you are trying to be more generic, then going with a programmatic generation of your affiliate pages is much simple and effortless (except for the coding part).

September 14, 2007

Google AdSense Revenue Decline

Filed under: AdSense — S @ 8:00 am

I have been reading a lot of articles about how changes in Google’s AdSense policies are affecting some of the websites. Mainly MFA, i.e., Made For AdSense sites and Arbitrage, i.e., sites that buy and sell traffic. This is the right step for Google and while it might impact it’s short term revenue potential, it’s good for the long term viability of online advertising.

However, in addition to discussions about the above two types of sites, people also started complaining that their revenue per click has declined or is declining. While Google might certainly be tweaking their algorithms, here are some factors external to Google that might be forcing these lower rates.

1. Increased Supply: CPC is based on bidding and budget. So, essentially supply vs demand comes into picture. Say, someone did research and figured out the high paying keywords and started making money. Over time, more and more people are likely to figure out and enter the same niche and as a result, the supply of page views for a given keyword is likely to increase. So, with more page views for a given keyword, and people clicking the ads, the budget of the higher bidders will be exhausted at some point and will trigger lower bidders. The net result is a decreased EPC (earnings per click).

2. Decreased Demand: MSN (through AdCenter) and Yahoo! (through Panama) have ramped up their Advertising solutions and are constantly trying to increase their own search engine traffic. As the effectiveness of these alternate ad networks improves, more advertisers are likely to diversify their ad spend to multiple networks. As a result, the demand on Google’s network will decrease.

So, using the conventional wisdom of Economics 101, what happens when supply increases and demand decreases? The price of the product decreases! Surprise, surprise.

So, for all those who are thinking of betting their future through advertising based on some of the examples out there on the web, be aware of this phenomenon. If everyone starts jumping into advertising, who is actually going to produce products or provide services and advertise both of these?

Why Amazon Affiliate Reports Are Not Realtime?

Filed under: Affiliate Programs, Earnings Reports — S @ 7:04 am

While AdSense reports are near real-time. This is exactly the same question asked in the comments section.

The technical reason is it’s difficult and expensive to do it.

However, Google does it. So, what’s the big deal? So, the functional reason is, Google is forced to aggregate the raw logs while Amazon is not based on the business model. You see, Google has to make sure it doesn’t overshoot the budgets set by the advertisers after each click. So, they have to keep aggregating the data. However, in case of Amazon, the advertiser equivalent party, the spenders, are the buyers and there is no budget setting for these buyers. They come, they buy what they want, they pay and leave. So, Amazon doesn’t really have to worry about overshooting someone’s budget and as a result, they can choose to run the reports once a day to aggregate the affiliate commissions.

Infact, many ad networks typically have the policy of reflecting the ad revenues in near real-time while the statistics on impressions are delayed a bit (for CPC ads impressions don’t count).

Affiliate Mashups – The Next Generation Affiliate Programs

Filed under: Affiliate Programs, idea — S @ 5:51 am

Many companies including Walmart have out-sourced their affiliate programs to other specialized affiliate management companies like LinkShare. eBay uses CommissionJunction (CJ) and Apple uses LinkShare.

However, Amazon is probably the only company that has the most advanced affiliate program built inhouse. It is because of the years of investing inhouse, that they are able to roll-out so many sophisticated affiliate programs including such ideas as context aware links, web services etc.

Personally I think, this is where a large giant like Walmart with it’s yester-years slogan of EDLP is lost the battle on the online front. I heard a lot of stories about Walmart that they develop a lot of their applications inhouse and they perhaps have a large IT department. However, they ignored/neglected the ecosystem, the affiliates who can bring in a lot of traffic.

I recently wrote about Affiliate Programs And Retargeting. With the ever growing expansion of the web, and the internet-aware youth who are going to be future smart-consumers don’t just trust someone just saying that they offer every day low prices, but do their homework. When comparison shopping is just a click away, why just limit the focus to a single site? This is where the ecosystem of affiliates comes handy.

Having been playing around a bit with Amazon’s affiliate program, here is one idea I think that can take Affiliate Programs to the next level (Amazon already allows multiple tracking ids, break up by various types of affiliate programs such as text links vs web services). The ability to offer affiliate mashups. Here is how it works

Say a company has a great idea that facilitates more people to participate in affiliate program and bring in traffic. This company essentially makes money from the affiliate referral fee but pays a cut of their commission to the people who contributed to that referral fee. Currently this is not possible since, there is no way to track by two affiliate ids. The main affiliate id belonging to the company and a second id (can be the actual affiliate id of the people participating in the program or some unique id given by the company). If this were there, then it’s possible to track exactly how much revenue each of the affiliates participating in the program contributed to the revenue and accordingly get a cut.

Currently, one approach people take for the above issue of not being able to track through two ids is to randomly change the affiliate id with a 90%-10% split (or some other value) where 90% of the time the actual affiliate id is used and 10% of the time the company’s affiliate id is used.

However, with the ability to track by two affiliate ids (one primary and the secondary), where the secondary affiliate id can only see the report but is not directly paid by Amazon (or any other company for that matter), it opens up a wider affiliate mashups possibility.

September 13, 2007

How Many Online Ad Networks Are There?

Filed under: Ad Networks — S @ 7:14 am

While you typically hear just a handful, there are many online ad networks. Check out this list of online ad networks.

So whether you are advertiser or a publisher, the options available are plenty. Do the necessary research and pick the right ad network for you. Publishers should note that not all ad networks accept them into their program. Most of them expect a large size of traffic to the website.

September 12, 2007

The Future Of Online Advertising

Filed under: Ad Blocking — S @ 8:03 am

There is an article on slashdot The Morality of Web Advertisement Blocking that talked about whether it’s right to use ad blocking software. Typical arguments on both sides of the equation.

Personally I don’t buy into the arguments such as ‘waste of bandwidth’ and ’speed’ because, these days people are wasting bandwidth on much more intensive apps like YouTube. Popups, animations and voice are a valid concern and the advertisers should eventually learn to get rid of these.

Anyway, I think, if the problem becomes widespread that no one is able to make revenue out of ads due to ad blocking, two things are likely to emerge.

1. Today, ads are served mainly by using a snippet of javascript that typically creates an iframe. So, the blocking software is actually looking for urls that contain well known ad serving companies. What if the ads are served as part of the content itself and not as separate iframes? This requires fetching content from the ad-network at the server side and combining it with rest of the dynamic content as opposed to letting the client to fetch the ads.

Ofcourse, the actual ad link itself has to go through the ad network. Converting just a link alone doesn’t really get rid of the ad, but will deprive the ad network of it’s revenue. One possibility is to use random ip addresses for this.

2. Even with the above, there is a possibility to look for common patterns and get rid of the ads. So, the second technique that ad networks can try out is to randomize the generated HTML content so that it’s not possible to come up with a pattern that can be easily filtered.

None of the above techniques are robust, but worth exploring. I guess, in the end, the ad-serving and blocking will turn out to be like virus creation and blocking, with the good and bad guys playing the opposite roles, depending on whose side you are :) .

September 2, 2007

Affiliate Programs And Retargeting

I recently wrote a lengthy article on What is Retargeting?. Subsequent to it, my recent experience with some of the affiliate programs got me into thinking a bit into the future direction of advertising and affiliate programs.

With the growing cost of online PPC advertising, large companies (alexa rank below 5000) are realizing that they can make use of their own affiliate programs that are more sophisticated and context-aware and pay based on PPA (pay-per-action). Typically the PPA programs offer more commission than the money earned through PPC. But the rate of conversion is typically less.

However, by offering these unique PPA affiliate programs that are context-aware, these companies are able to do re-targeting for free! Yes, that’s right. Take Amazon for example. Any shopper who is actively looking to buy something, if he wants to buy on Amazon but not sure of the pricing, he would search further on the web and might land up on one of Amazon’s affiliate program. After a bit of searching and realizing that there is potentially no better deal elsewhere (combining tax & shipping), then the user might land up back at Amazon through one of it’s affiliates. If the user ends up buying the product, it’s a win-win for Amazon and the affiliate. If not, it results in retargeting for free for Amazon!

Another interesting thing is, some companies don’t care about how much traffic is really brought to their site by their affiliates. On the other hand, Commission Junction recently sent an email to me stating that they would deactivate the account as the volume from my sites is very low. Yeah, right! Like I care? I will write another article about the future of context-aware affiliate programs and what that means to companies like Commission Junction and LinkShare.

Blog at WordPress.com.