Tobias Ratschiller's Blog

Earlier this year in March I went on a trip in South America and visited Argentina, Bolivia and Peru. In terms of business it was quite an eye opener for me and I'm convinced that there are huge growth opportunities in Latin America's Internet sector.

It started with the travel preparations. Booking flights or trains in countries like Bolivia or Peru is a pain. To book a flight from Santa Cruz, Bolivia, to La Paz, Bolivia, I not only had to call to make the booking, but I also had to fax in scanned copies of my passport and credit cards.

In general tourism is not well developed except in the most common touristic areas such as Machu Picchu. When we took a boat tour across Lake Titicaca, we (two persons) were alone aboard a huge ship - dedicated tour guide included! For the price we paid (I think something like $200 for the day) this could only be a loss for the tour operator.

Obviously Internet penetration is low at around 25% but even compared to countries like Ukraine with similar penetration rates I was surprised that even for example companies in Lima do not advertise their websites much on their shop signs, flyers etc.


 Of course, once you take a jeep and go for a few hours into the campo, you discover places like San Antonio: no Internet, no cell, no phone, and they got power last November.


While we have a sizeable business directory of Bolivia's Santa Cruz in our porfolio, San Antonio, or the next larger city of Concepcion (10,000 population) don't even register on the map.

The AdSense numbers reflect this general impression: CPM of South America is half of Ukraine, Ukraine is half of Europe. Still the growth of our Latin sites (we have, for example, a business directory of South America) is impressive.

It's that time of the year again. I love the time between Christmas and New Year's. Business is quiet and there is time to look back and plan for the next year (even if it usually comes down to the same Pinky And The Brain quote: "Gee, Brain, what do you want to do next year?" "The same thing we do every year, Pinky - try to take over the world!").

2010 was without big surprises in our business. I had said in the beginning of 2009 that crisis years are building years, and in 2010 we continued what we had begun in 2009. We bought over 50 developed sites this year on forums, Flippa, and via private deals. In addition, we started a promising cooperation with the Domain Developers Fund to develop some of their names in equal rights joint ventures, the first ones being TCM.org and Holland.net.

... or how we increased our CTR from January 2010 to August 2010 by over 300+%. You can literally double or triple your AdSense revenue just by optimizing the ads.

Here are some leanings that we found had the most impact on our revenue. This may be common knowledge among publishers, but it took us a while to come this far, and cost us tens of thousands of Euros along the way. Unfortunately, the AdSense Terms of Service disallow disclosing of the exact CTR or other figures, therefore limiting meaningful discussion - probably that's why we didn't find useful articles when starting!

I assume you are familiar with the basics of optimizing AdSense, such as the AdSense heatmap, setting up channels for measuring your experiments, and Google's basic color suggestions.

Split test everything. The software mantra of "Don't touch it when it's not broken" does not apply here. With AdSense, continuous improvement is key. The best way to test is split testing (variation testing), which often leads to unexpected results. Obviously different sites and different industries have different preferences and requirements. It surprised me though how different sections on one website performed significantly different. Here is an example. The Danish business directory on Denmark.net performed much better (0.5%) with a red border than the English language version did - there the red border was actually detrimental to CTR.

How to split test AdSense: Here is a quick way to implement version that serves two or more different channels on the server side with a random distribution. This is easy to do in PHP:

mt_srand((double)microtime()*1000000);
$rand = mt_rand(1,2);

if($rand == 1)
{
// serve AdSense channel A
}
else
{
// serve AdSense channel B
}

You can then easily compare the two channels and see the winner.We split test everything. Font colors, font size, borders (rounded or not?). We actually automate split testing to automatically pick the winner and continue with that until it can't be further optimized via machines.

Color optimization: Should you go for a blended palette or for ads that really stand out? Again there is lots of conventional wisdom around this area which you may completely throw out of the window once you actually start testing.We have not found a single answer to this question - except that you should test!

Placement. Probably the single most important factor of the CTR is the placement of the ads. You have to give special attention to the publisher guidelines here, for example not to place ads too nearby navigational elements. However, it can make be an easy 0.5% difference in CTR if you place an AdSense unit a little bit higher above the fold than in the center of the page. Again, they key here is testing variations.

Best performing ad units first.  Google displays the highest paying ads in the first ad unit of the page. So make sure that first unit is your best performing one! You may need to use CSS positioning to show ad units higher up than they are actually in code. For example, you may have a header ad that doesn't perform well and a in-content ad that performs well; you would now use CSS to position the header ad unit in the header of the page even if in the source code this ad unit comes second.

Section targeting: Using section targeting, you can highlight to the AdSense bot which parts of your page contain the main text body to which the ads should be targeted to. This can make a huge difference by serving ads that are actually contextual.

Direct placements: AdSense has two ways of serving text ads: In the first way, Google does all the logic of trying to understand your page and the visitor to serve contextual ads. The second way lets advertisers target your site directly and place ads in specific channels on your site (if they are enabled for it). It's important to know about this, since the CPM and CTR can be wildly different for the two scenarios. If you go to a channel in AdSense that is currently enabled for external placement, there is a link called "View 7-day earnings" which shows a report comparing the direct placement performance versus Google's. In our case, the direct placement was generally under-performing by 30-40% except on specific pages, so we disabled direct placement for many channels.

Heat maps: While basic heat map optimization is explained by Google, it took us awhile to apply it across different sites. Here is just an example of how a "minor" repositioning can add over one percent to your CTR.

Old:

New:



Just moving the information sidebar (company logo and some text) on our South America business directory from the left of the content to the right increased the CTR by more than one percent.

Demographic-based optimization:  Since most of our sites have anonymous traffic, we rarely do this, but for web properties with many signed in/identified users, this is huge. I've seen examples where just serving different colors for men and women lead to a CTR doubling. I'm sure this is taken to excellence by the big portals.

Hire an AdSense optimizer: If an expert is able to increase revenue by 20% on an account, then for sites that have at least some basic AdSense revenue (5k/month plus) this would be a worthwhile hire.

Meanwhile, just keep optimizing, it's actually one of the fun parts of the online business because every bit directly translates to more money in your or your company's pockets.

What would you do if you owned a store that is normally visited by 100 people a day, and suddenly, starting on a random day only 10 visitors come in?

You are fucked, that's what.

And still, this is what happens to small businesses and start-ups so often that you would almost call it normal. An important customer cancels their contract, the webserver goes down, a snafu messes up your inventory, or Google decides to stop sending traffic to your website. And with small overall revenue of maybe a couple hundreds of thousands to one or two million per year, all of these events have the potential to significantly affect your revenue.

Most of these have already happened in my businesses in one way or the other. Most recently, the traffic died on one of our websites. This is your worst nightmare if you have a website that is depending on organic search traffic. One day, you wake up to virtually no traffic.

It does not even have to be a Google penalty (which would be unwarranted for in our case since we always follow the webmaster guidelines), but a simple hiccup in the Google algorithm that leaves you with a shop and no customers.

So how do you prepare against potential massive drops in revenue?

The theory is easy, of course. "Secure revenue streams are diverse revenue streams", the saying goes. The ideal situation is that no income stream makes up more than 20% of your overall revenue. In start-ups and small companies, this is easier said than done. Would you turn down a customer just because the contract is 40% of your income? Of course not.

As I see it, as a small business, you should look at diversification as buying time in case one income stream goes foul. How much time do you have when your most significant income stream vanishes to find a replacement? In general, I would shoot for more than six months.


Let's say your company has 100,000 income per month, 40% of that coming from your largest customer. Your expenses are 70,000 per month, leaving a net profit of 30,000 - or a minus of 10,000 per month if your big daddy goes away. After cutting expenses, you should be off to a modest monthly loss that hopefully you can sustain for at least six months with your company's savings.


Another strategy is to "insure" against income loss by hedging against potential losses. For example, as company and investors, we hedge against the risk of traffic loss by buying stock in Google, Apple, Microsoft, etc. As site developers, we hedge by running many websites in unrelated niches.

In our case of the Death Valley of traffic, there was a Happy End, by the way. On May 19th, traffic suddenly came back to the site. Thanks, Google. :)

Looking through our Austrian business and people directories, here are a few names I just had to share with you.

Of course I wouldn't go so low to make fun of psychologist Dr Killer or dentist Dr Dick. Never would I laugh at names like Hermenegilde Fucker, Reinhold Titz, or the many people in Austria called Harasser.

But I thought it was quite weird that a hotel would call itself Pension Wanker. I'm not sure about the pub Dick Macks in Vienna, neither.

Then there are the city names. There is a city called Egg in Austria. And one called Rottenegg. In Tyrol, you can find Mutters and  Natters. Or Rum. The shortest city name in Austria is Au.

The most common names in Austria are Bauer (7,942), Gruber (6,884), Huber (5,974), Berger (5,659) and Eder (4,942).

Nine people are called End. Six are Last. None called One, unfortunately, but there is a One Up marketing agency, the One Touch fashion store and One and 2, whatever they do.

Say somebody databases are boring.

Since this month we had to develop and launch our newest batch of country tourism websites (Visit Brazil, Visit Argentina, Visit Chile, Visit Peru, Visit Bolivia and Visit Paraguay), I wanted to see some official country logos and asked a colleague to compile a list of country tourism brands.

Here it is - enjoy! I think my favorite is France.