Running a Proper Domain Test

I’ve been in the business of managing domain traffic for nearly two decades, the last of which has been as one of the co-founders of ParkLogic. In this article I will share with you how a proper analytic traffic test should be constructed and the pitfalls that so many people fall into….which ultimately robs them of increased revenues.

Let’s imagine you have a reasonable amount of traffic revenue at your current monetisation partner. By reasonable I mean at least $250 per day. For those of you that think this is impossibly high then you can stop reading hear and continue to believe that domain traffic monetisation is dead. Everyone else can keep on reading.

Let me say from the outset that if you have left your domains with the same parking company for more than a few months then I can almost guarantee that you’re leaving money on the table. For a start, we see that around 60% of all domains achieve greater results from non-Google sources.

In constructing a test, you need to have around $100/day of revenue (based on the previous month’s stats) from a couple hundred domains. Some domains should be doing a few dollars per day, while others should still have traffic but do no revenue. This will provide a good statistical sample that will help verify whether the new monetisation partner is actually performing or not.

Any new partner worth working with should then ask for these numbers. If you think you’re being clever and not providing them by asking them to perform their “best” then all you’re actually doing is delaying the optimisation cycle. The new company will have no idea where to concentrate their resources unless they have the numbers.

The new company should then use the baseline numbers you’ve provided as a benchmark and provide a detailed report of where they’ve won and where they’ve lost. There should also be reasons why they have won or lost for each domain.

For those of you that think it's really smart to inflate the baseline number then think again. Let's imagine the new partner almost achieves the "new" amended baseline....are you going to leave the domains with them? All the data says you should take them away. If you don't take them away then they'll know you've just started the relationship out with a lie....not a smart thing to do!

I remember having a conversation with a domain investor that was running a test and they were surprised when I told them to not expect a particular domain to keep on performing the way it was. We were winning versus the baseline but I knew that the domain was highly seasonal….so how could we claim a “win” in this situation? We couldn't be certain the upturn would continue.

The test should take around 6 weeks to complete so that low traffic domains have an opportunity to shine. Typically, the results will be initially low but don’t panic…..over time you should see them ramp up.

Not being privy to how other companies behave, I can only share how ParkLogic does a traffic test. When a new customer provides baseline data, we adjust it to a daily value for each domain. This benchmark is then agreed to with the client and we begin the optimisation process.

As a part of our optimisation we run a real-time auction in about 100 milliseconds between many different global advertising networks. If any of them bid more than the very best traditional parking solution, then they win the traffic otherwise the best Google backed solution receives the traffic. There's a LOT more to it than that but I hope this gives you the idea.

Over the last 12 years ParkLogic has developed a sophisticated algorithm to route the traffic while at the same time ensure that there is a sampling regime in place (which is another story in itself!) that checks whether the winner really did pay the most. This system works incredibly well and allows us to focus on more manual optimisation techniques for specific high value domains.

We request to have a meeting with the new client every two weeks where we provide daily charts of the key metrics and a detailed summary of the performance based upon traffic levels. The new client is also provided with my direct contact details if they have any questions between the scheduled meetings.

In the reports we also display which domains we’ve performed better on and which we’ve performed worse. Nothing is hidden. In some rare cases we’ve even advised the client to take away a few badly performing domains and return them to where they were before. We believe in taking seriously the best interests of the domain investor.

When running a traffic test, never forget that you’re actually comparing data from two points in time. For example, if we compare baseline data from May versus July then the current data is likely to suffer due to the northern hemisphere summer. Likewise, November has an uptick for cyber-Monday and black Friday compared to October.

In fact, most domains exhibit some sort of “global” or “local” seasonality. For instance, I once had a domain that did nothing all year until the first two weeks of March where it went crazy! It was a domain that related to the world day of “Pi” (the mathematical symbol) on the 14th March.

We had another customer that had a lot of education related domains that did poorly in June compared to the baseline data provided for May. Really understanding individual domains is often the key to unlocking their full potential value.

I hope this article has helped you consider how you are monetising your domain names and whether it’s time to run your own test. Drop me a line if you would like me to assist in any way.