This articles continues directly from the previous in the series on conducting a traffic test (click here to view). I then create a summary tab which pulls all of this data together so that at a glance you can see what’s going on. Since we create these packages of analytics for clients on a regular basis you’ll have to excuse me if I refer to the testing company as ParkLogic.

In the summary, you need to know the total amount of revenue (daily) earned by the testing company and the baseline. This will provide the lift above the baseline. In the case of the screen capture there is a $190.94/day increase which has then provided a 127% uplift versus the baseline. This is quite a good result.

I then like to find out what was the maximum amount that ParkLogic was winning by per day. This is so that clients that wish to take all of the domains away can then leave the winners with us. In this case it’s $457.54 per day which isn’t that bad considering the baseline for ALL of the domains was $715.40.

In fact, if you leave the winners with ParkLogic and send all of the losers back to where the baseline source then the result should be $1172.93/day versus a total baseline of $715.40. In this case the result is a 64% uplift in revenue and only assumes that the baseline produces the same results as previously attained.

When I was speaking at NamesCon in January I said that I would like to show the attendees the best computer game I’ve ever played. It has all the aspects of both a tactical and strategic game with a bit of time travel for good measure. Everyone leaned forward with expectation…

I then put Excel up on the big screen in the room. It’s safe to say there was a bit of laughter. Let’s think about it for a second, Excel allows us to answer questions about the performance of our domains like no other application can. We can view the past, compare it to the present and even predict the future.

Right now we are going to mobilise the data that we have arrayed from the previous articles in the series so that we can understand what is happening with our domain traffic. This will be a VERY high level view of our traffic test but I believe that it will help get some answers to our questions.

The first thing we need to do is to ensure that all of the baseline data is in the same currency as the testing monetisation source. Once this is done then we then need to convert everything to daily data. The reason why we need daily data is that it gets rid of the problems associated with 28/30/31 days in a month and also allows us to later compare the data against daily from the new testing source.

Next, create another tab in Excel and call this “Latest Data”. When capturing data make sure that you have multiples of 7 days so that any variability caused by the weekends can be minimised. All of this data needs to be brought back to a daily format so that we can then compare it to the baseline.

Make sure you check the DNS settings of EVERY domain in the test so that domains not pointed correctly can be removed. There’s no point in penalising a test by having domains with baseline data and no data from the testing source.

Now create a tab called “Analysis”. In this tab you need to add you complete list of domains in column A, Column B, C and D are for the baseline views, revenue and RPM. Columns E, F and G are for the new monetisation company’s views, revenue and RPM.

Now that the data is sorted out you are now in a position to conduct some analysis.

In the previous articles in this series I discussed the importance of baseline data and what you should be measuring. I’m now about to dive into the most important metric that you can ever use to evaluate the success of a traffic test, the normalised revenue per thousand visitors.

I want to apologise to all those readers where mathematics isn’t your strength….that’s OK. I will try and make this article as simple as possible so that it gets across the point.

Many domain owners have learned to pay attention to the revenue per thousand visitors (RPM) that is produced by the various parking companies. The reason for this is that it takes into consideration the variation in the traffic levels for each domain. So really what is RPM? The formula for RPM is the following:

Revenue / visitors x 1000

This makes sense until you get under the both revenue and visitors. For a start, visitors is actually filtered traffic and since each parking company filters traffic differently than this number changes for each company. Another way of viewing visitors is:

Raw Traffic x Parking Company Filter

Then there is the Revenue number. What revenue number should you be using? The estimated numbers, the number confirmed 2 days later, the number less clawbacks, the number less account adjustments etc. There are about 7 different revenue numbers that need to be examined for each domain!

For example, over the years we found that some monetisation companies would say that they will pay more for domain traffic during the month and then do an account adjustment at the end of the month. This meant that they were effectively bribing the traffic routing algorithms to during the month and then taking all the money back at the end. To understand who the winner is you need to take this type of behaviour into account.

So let’s look at our formula for RPM now:

(7 Different Revenue Numbers) / (Raw Traffic x Parking Company Filter) x 1000

Because we don’t know what the parking companies use to filter their traffic let’s imagine that we can actually count the Raw Unfiltered Traffic that we send each company for each domain. Let’s also imagine that we are able to sort out the revenue and with a bit of magic work out the actual revenue number for each domain each day. The formula then simplifies and looks like the following:

(Adjusted Revenue)  /  (Raw Traffic)   x  1000

This is the normalised RPM (nlRPM) and it allows you to directly compare any monetisation company against another. What we do is count each bit of traffic that we send each company each day and then measure the adjusted revenue that the traffic generated. When you do this for all companies you get a number for each company so that you can then know which one is actually paying the most.

Let me say from the outset that this starts to get REALLY complicated! This is also what you need to consider if you want to know who is actually winning your traffic at any point in time.

Thankfully, at my company, ParkLogic, have servers and algorithms that all the automatic mass calculation of all of these numbers. We then use these numbers to route the traffic to the winning company. Each day, we have servers that all they do is process data for about 15 hours to get to the nlRPM.

So let’s imagine that we have a domain that has a huge nlRPM and it’s smashing the baseline data out of the ballpark. Do we claim victory? Heck no! Even when you have the normalised data you need to understand WHY the domain is winning.

For example, let’s imagine that you have an education related domain and you are comparing the baseline data in July versus September. I can almost guarantee that the nlRPM will be higher in September as school’s back and this will attract the educational advertisers!

To put everything into context, the nlRPM is like the science of domaining….you have to have this number really know how you are doing. Understanding why the nlRPM is changing is the art….this is where experience comes into play.

I think that the gold rush provides a really good analogy for traffic monetisation. In the past, there used to be gold lying on the ground everywhere and you didn’t have to do anything to pick it up. Today you have to drive a shaft 3 miles deep and run side passages that follow the seam of gold. This is what I’m talking about with the nlRPM. The gold is still in the mine but you just need to dig it out and this is what I do day in day out.

Please leave a comment or send me a message if you would like me to run a webinar on how to run a properly constructed traffic test.

---------------------------------------------------

Michael Gilmour has been in business for over 32 years and has both a BSC in Electronics and Computer Science and an MBA. He was the former vice-chairman of the Internet Industry Association in Australia and is in demand as a speaker at Internet conferences the world over. Michael is passionate about working with online entrepreneurs to help them navigate their new ventures around the many pitfalls that all businesses face.

“What are you measuring” seems like an obvious question when you are conducting a traffic test with a new monetisation company. From experience, most people get the answer to this question completely wrong.

When you are running a traffic test you are NOT seeing if you can earn more money. This may sound strange but it’s the simple truth. What you should be actually measuring is the capabilities of the new company to optimise your traffic for a better result. More money (although important) is merely a by-product of the optimisation activity.

For example, we recently had a client place their domains with ParkLogic for optimisation and as time went by we just couldn’t beat their numbers. In fact, it seemed that the more effort that we put into the optimisation the worse thing became.

It was at this time that I had an epiphany (ie. a brainwave!). It just so happened that the client provided their baseline data in Euros from back in August and we were being benchmarked against these numbers. Since the ParkLogic system uses $USD we dutifully converted the baseline data from Euros to $USD when we received it.

The below graph tells you the story of the relationship between the Euro and USD during this period of time. The Euro has fallen from $USD1.36, when we did the initial conversion for the baseline data, to $USD1.09 (bank rate). That was a fall of 25%!

So the fact that we were providing the client around a 5% uplift wasn’t the real result at all. The real result was 5% + 25% = 30%. To clearly show this we applied today’s exchange rate to the baseline data so that we could clearly measure our optimisation efforts and nullify the impact of the exchange rate.

Another example is we were optimising a domain portfolio and the results were absolutely stellar! In fact, we were beating the baseline data by about 250%. Naturally, the client was really happy with the results but something was annoying me about the numbers….

A quick comparison analysis showed that one domain was getting a number of clicks paying $80+ each. We removed the impact of this domain as we did not believe that it was sustainable and redid our analysis. The uplift was still around 135% but I was more comfortable with this number being sustainable.

In both these examples, only paying attention to the money earned column would cause you to completely miss the actually impact of the optimisation effort. It’s not just the data but interpreting the data that is critical to getting the most out of your domain portfolio.

Another simple example is when you have domains that are seasonal in nature that affect the result. These should be removed from any analysis to determine whether the new company is actually adding value. They have the potential to either inflate or deflate the numbers and cause you to come away with a completely incorrect picture of the new monetisation company.

I live a breath numbers all day every day and I would like to say that I’ve seen everything…..but I know that I haven’t. When measuring the impact of any change you must first of all have a clear baseline and then have a metric that you can actually use to measure the impact of that change. It’s the art combined with the science that produces results….and both are constantly evolving.

In a future article I will begin to unpack what metric is critical for determining whether a new monetisation company is performing or not.

---------------------------------------------------

Michael Gilmour has been in business for over 32 years and has both a BSC in Electronics and Computer Science and an MBA. He was the former vice-chairman of the Internet Industry Association in Australia and is in demand as a speaker at Internet conferences the world over. Michael is passionate about working with online entrepreneurs to help them navigate their new ventures around the many pitfalls that all businesses face.

Doing properly structured domain traffic tests are absolutely critical to extracting the complete value from your domain traffic. What it also means that you need to get down and dirty into the numbers to really understand what’s going on. I’m going to unpack what I mean by this and how we typicaly conduct traffic tests at ParkLogic.

First of all, let me say that doing a properly constructed traffic test is not for the faint hearted and is NEVER as simple as changing a nameserver.

I was speaking with a domain owner that has been in the industry for many years and still believes that changing the nameservers for their domains is how you start a traffic test. It’s one step in a larger plan and is rarely done straight away.

The first stage in any test is to establish what you are measuring against. For example if all of the domains are currently parked at Company A then the least you need to do is download the last months data to benchmark the success or failure of the test. This set of data is known as the baseline.

There is no point in messing with the baseline data and scaling it up to make the new company “work harder”. I’ve seen data multiplied by a factor and then sent through as “baseline” data. I've even been provided as a baseline the best the domains have ever performed in their history!

Let’s think about this for a second. Let’s imagine the test beats this manipulated data. This is good news because it means the new monetisation company is awesome! The question of what do you do if the test fails is much harder to answer. You either look like an idiot because you leave the domains with the new test….which means the new company knows that you messed with the data or you move the domains away and take a monetary hit. This is even more stupid. So please don’t mess with the baseline data.

Next, hold the baseline data very loosely. Baseline data by its nature is a snap shot at a particular point in time. You need to not only look at the numbers but understand what they mean. This will involve doing an analysis at the domain level and understanding why there are wins and losses.

For example, there is no point in holding the new monetisation company to account for education domains if the test is being conducted in July and the baseline data is from May. Of course education domains will perform better then! In your own thinking you need to remove these types of domains from any test.

All baseline data needs to be reduced to daily numbers. This helps average out the traffic and revenue data across an extended period of time and allows the new company to snapshot a week of data, turn it into daily data and compare it to the baseline.

In many instances you should also screen capture all the high value domain names at the current monetisation company. I would normally recommend leaving this up to the new company. These screen captures can be used to help unravel why some domains may be performing worse than others during the test.

The absolute minimum requirements for baseline data is views and revenue. If you don’t have either of these then you really shouldn’t run any test at all…..it’s just a waste of time.

What we have often done is told the client to leave their domains exactly where they are. We then ask them for access to their current monetisation company account and change the nameservers to ParkLogic. We then route 100% of the traffic back to the existing monetisation company.

This allows us to establish URLs (raw traffic), Views, clicks and revenue. From this we can establish a normalised RPM (revenue per thousand URLs). This number is the ONLY number that will clearly display who is winning in any future test.

After a week of running traffic to establish a normalised baseline we then test a percentage of the traffic elsewhere….but more on this later. By creating a baseline in this method you have a completely accurate measurement of success or failure.

---------------------------------------------------

Michael Gilmour has been in business for over 32 years and has both a BSC in Electronics and Computer Science and an MBA. He was the former vice-chairman of the Internet Industry Association in Australia and is in demand as a speaker at Internet conferences the world over. Michael is passionate about working with online entrepreneurs to help them navigate their new ventures around the many pitfalls that all businesses face.