After publishing the article, “Getting Dirty in the Domain Data”, earlier this week I ended up having an interesting discussion with a domain investor. I thought that it would be worthwhile continuing to pull apart the data from the previous post to help many domain investors understand why optimising traffic across multiple monetisation solution is so beneficial.
I will be referring to the data from the previous article so you may wish to read it if you haven’t done so already.
Sampling by Changing the DNS
Many domain investors sample different parking providers by changing the DNS. This method is fraught with many problems that largely stem from comparing results from single sources across different periods of time. Some of the challenges are:
1. What are you measuring?
Revenue is not a good measurement of success as there will be different levels of traffic at different points in time. Since parking companies count traffic different you can’t rely on the produced Revenue Per Thousand Visitors (RPM) numbers.
2. Data Distortion
When testing different parking providers over different periods of time you can get massive distortions in the data from seasonality in both the domains and the time of year. In the example from the previous article the domain had massive results in May to July due to it being a travel related domain.
3. Traffic Leakage
There is a propagation delay each time you change the DNS and this creates more traffic leakage and no new information. In some extreme cases where the TTL (Time To Live) for the domain is long the DNS may not update for months for some users.
The Cost of Information
Some investors believe in splitting traffic equally across multiple solutions and then at some point in time sending all the traffic to the monetisation company that pays the most. This is one of the worst ways to optimise domain traffic and here is the reason why.
There are a lot of strategies around sampling but they basically boil down the single question, “What did the information cost?” In other words, if I was earning one dollar with one company and then sampled another company and found they were paying 90 cents then the information cost me 10 cents.
Minimising these "information costs" is crucial to optimisation. From the example domain, A.COM, in the previous article if we sample the traffic equally across the different parking companies then the portfolio would have earned $1611 versus $2217 or 38% less overall (ignoring direct advertisers).
Every domain needs its own sample regime. At the most simplistic level domains with less traffic should be sampled less often compared to domains with high levels of traffic. In each case, what you are after is a statistically significant result that allows you to decide where to route the traffic. If you don’t have a statistically large enough sample, then you’re guessing.
Real-Time Decision Making
All traffic routing decisions need to be made on a real-time basis. Based upon the data, decisions need to be made literally milli-second by milli-second. I can only speak for my company, ParkLogic, as we use dynamically changing data from multiple inputs to alter not only the routing decisions of traffic but what is displayed on the page and ultimately which advertisers are engaged.
As an example, we track over 250 different metrics for every domain every day and process this data to alter how the traffic is routed. Layered over the top of this daily data we then can then incorporate external dynamic data such as geo-based weather.
Everything must lead to a decision....otherwise it's just intellectually interesting but pointless.
Winning Solutions Constantly Change
If you sample other solutions (however you decide to do it) and then lock that solution in for an extended period of time, then you will be losing. The data from the previous article clearly shows that even for a single domain the winning parking company changes constantly (see below table).
For example, if we routed ALL the traffic through to Voodoo (average winner) and applied Voodoo's payout rates each month then Voodoo would have paid out $436 for the ten-month period. The domain actually earned $4531 for the same period of time (including direct advertisers). The reason for this was a combination of an advertiser paying a lot for the traffic in May-Jul and other parking solutions beat Voodoo the majority of the time.
This doesn't mean Voodoo is bad....as they actually did win for a couple of months. Remember the data is summarised on a monthly basis and can only testify to the fact that the same behaviour exists at the daily and even changes milli-second by milli-second.
Benchmarking Results Must be Done Simultaneously
I mentioned this point briefly when discussing the problems with sampling via DNS but it is important to reiterate it. Testing new solutions must be conducted at the same point in time otherwise distortions in the results will occur and incorrect decisions made.
Let’s imagine I used the domain’s revenue results in June as the baseline data and compared this to any new parking company in September. I could erroneously conclude that the new company was hopeless! Remember that A.COM (in the previous article) is a travel domain and has extraordinary performance in June.
I’m in a discussion right now with a customer where about one hundred of their domains just aren’t performing. I’m not worried about this customer leaving ParkLogic as we are both working through the data to understand why their performance is down.
Too many domain investors immediately bail on their existing partner and whip their domains out somewhere else in the vain hope they will perform better. This syndrome has a saying, “The grass is always greener on the other side of the fence.” In other words, you will always think somewhere else is better than where you are.
My advice is don’t do a knee jerk reaction and move your domains. Sit down and dig into the data and really understand what’s going on with the traffic. We are in an industry that is built upon data and if you wish to get abnormal returns then it’s vitally important that you get your arms around it or work with a partner that can help you do so.
I hope the few items I’ve raised here in this article will help give you a fresh perspective on your own domain portfolio. Over the years I’ve found that earning more from domain traffic is not always the solution that investors are after. What they want to know is they are maximising their returns and there is proof that this is being done.
Anyone can have a good or bad month but knowing that there are systems and experts in place that are monitoring and understanding the results is really where it’s at. This is particularly the case if you must report to investors or a board. Having the data to confidently know that everything that can be done is being done often alleviates the concerns of the most aggressive directors!