Part 8 – Portfolio Management – Traffic Domains

Part 8 – Portfolio Management – Traffic Domains

The current series of articles has really provoked some interesting discussions with a number of domain investors. What I have noticed is the high level of emotion that surrounds the whole domain traffic and optimisation debate. Many people have leapt to conclusions rather than look at the data and try to understand what it is saying.

Let me say once again that domain monetisation is not dead. The reason for this is genuine domain traffic contains valuable leads for businesses and advertisers are more than willing to pay for those leads. The goal of optimising your domain traffic is to best match the right advertiser to the right piece of traffic at the right time.

Like any industry, the advertising buying market is very dynamic with wide ranging payout levels at different times during the year, day and even sometimes second. So when I talk about optimising domain name traffic we need to really be as close to real-time as possible. Given the volume of data, the only way to route traffic to the highest paying solution is via algorithmic switching. So is all of this work really worth it?

I recently published some numbers over at the NamePros forum and I thought that it would be worthwhile digging into them here. For full disclosure I should state the data is from my company ParkLogic and we use algorithmic switching of domain traffic.

The data is real and was used as a part of the optimisation process for a particular client. Results for new clients will vary depending upon the optimisation levels and the domains in question.

ParkLogic Results

What the screen capture shows is if we saw at least 100% of the baseline traffic (ie. The same) then we provided a 184% uplift in revenue. At least 80% of the traffic we provided a 162% uplift and for the entire portfolio there was a 127% uplift versus the baseline. In other words, if we saw similar levels of traffic to the baseline data then we knocked the results out of the ball park.

The baseline data came from a portfolio that was currently being optimised across multiple monetisation solutions via Above.com. I have nothing against Above as ParkLogic provides an entirely different service compared to them.

Just for the record, there were some domains where the baseline outperformed our results. This could be for statistical variations across time etc. We would recommend to any client where the baseline outperformed ParkLogic to take those domains away from us and re-baseline them to ensure the results hold true. ParkLogic does NOT win for every domain ALL of the time (versus a baseline) but I can say that we DO send the traffic to the winning monetisation solutions at a particular point in time.

What the data clearly shows is when you apply a level of discipline to optimising a domain portfolio that we do then the improvement can be considerable. The problem is most people try to optimise their portfolio themselves and have convinced themselves that they are saving money in fees.

This is a false economy because the data clearly shows that any fee would be more than covered by the uplift. In addition, the fact that you now have all of your time available is a huge saving in itself!

I don’t normally toot my own horn but ParkLogic really knows its stuff. We’ve been optimising domain portfolios for nearly a decade and when someone comes along and says they’ve built their own optimisation system it’s hard not to roll my eyes. Over the years I’ve heard many approaches……everything from DNS round-robin through to an automated system that leaves traffic at one place for a few days and then moves it to another for testing. I hate to say it but all of them are sub-optimal.

As an example, let’s tackle something really simple, incorporating zero-click real-time bidding (RTB) solutions into the traffic stream. Essentially RTB platforms let you poll them for what they are willing to pay for a piece of traffic. So they may come back with 12 cents as an answer. If you look at the traffic for a domain you may have an earnings per click rate of 10 cents and immediately jump at the 12 cents…..that is actually a really silly thing to do.

What we need to understand is that the 10 cents that is being earned at parking companies is an average across a period of time (typically a day) while the 12 cents is at a point in time. In other words, the 10 cents could be made up of a 20 cent click, a 5 cent click and a 3 cent click which provides an average of 10 cents.

What you don’t want to have happen is the RTB networks take the 20 cent traffic and pay you 12 cents while leave the 5 and 3 cent traffic behind. This would effectively reduce the yield from the traffic. So after around 6 months of testing various methodologies we worked out a system for incorporating RTB networks in an effective manner.

I know this was a simple example but it does give a glimpse into some of the complexities around routing traffic to the highest paying solution. It also illustrates that if you get your maths wrong then you could lose your shirt!

1830 Hits
0 Comments

What's Going On With PPC? - Part 1

What's Going On With PPC? - Part 1

I regularly dive into "big data" to try and better understand what is happening with the performance of a client’s account and to determine whether action needs to be taken. It was at this point in time that I was faced with some very puzzling numbers to muse over.

It was always assumed that the click through rate (CTR) was an indication of user intent. The goal of any optimisation strategy was to maximise the CTR by making the advertisements more relevant to the domain so that the user then clicked. This seems to make logical sense but it’s clear from the recent round of data that something else is at play.

Escrow.com

Likewise, the Earnings Per Click (EPC) was a measurement of advertiser competitiveness. The greater the demand for the keyword the higher the price paid by advertisers….it’s essentially a reflection of the Google auction process.

In the past, domainers would select higher paying relevant keywords to maximise both the CTR and EPC. For example, I may set “Mortgage” as the keyword rather than “House Auctions” for a real estate domain because “Mortgage” had a higher EPC and around the same CTR as “House Auctions”.

This whole context sensitive mapping of the domain to the keyword was effectively thrown out when Google migrated to psychographic user based targeting. In other words, Google is now trying to match advertisements to the user and not necessarily the domain.

For example, if you’ve been searching for vacations in Bali recently then when you can go to parked golf related domain you will often see vacation advertisements. This seems to make sense until I pulled out the stats for a couple of accounts…..and this got me questioning what is really happening.

Both samples have thousands of domains and lots of traffic and are not being distorted by a single domain. The lines are a level 4 polynomial trend line for both CTR (blue) and EPC (red). Incredibly, both charts have a similar profile where the CTR and EPC effectively mirror each other and create an inverse impact. So what does this all mean?

Around two years ago Google migrated all of their parking partners to using what is now known as Custom Ad Frame (CAF). Essentially this means that all parking pages are now largely served by Google and not the parking companies themselves. The goal of this change was to open up Google’s premium advertisers to the domain channel….which is good news for domain investors.

The problem I have is that the CTR component of the graphs suggest that Google is playing a really strange game. They appear to be constantly moving from accurately targeting users with advertising to doing it really badly. Although some variations are expected, the wild swings from May to Sept are a bit perplexing. This seems to fly in the face of the Google ethos to always provide the user with better and more accurate results.

In both cases the overall CTR since May has dropped significantly while the EPC has risen! What this suggests is that during the summer people click less often but advertisers pay more. This is the reverse of what I would expect. During the summer period I would expect users to click less (distracted by the sunshine) and since advertisers are on vacation many of them would exit the auction process. This reduces demand and you would expect EPC rates to decrease and the traditional summer downturn would result.

What the data suggests is that as better quality advertisements (ie. higher paying) are displayed then people click on them less. This seems unlikely and combined with the shapes of the graphs it suggests a level of manipulation of the data behind the scenes.

So the question really needs to be asked, "What is CTR and EPC?" Unless we now live in a world where users can discriminate their clicking behaviour by magically known the EPC rates then there has to be something else at play.

I plan on unpacking this further in the next articles in this series as I explore what is actually going with domain traffic.

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

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. He has also recently published his first science fiction book, Battleframe.

Michael is passionate about working with online entrepreneurs to help them navigate their new ventures around the many pitfalls that all businesses face. Due to demands on his time, Michael may be contacted by clicking here for limited consulting assignments.

1726 Hits
0 Comments