In this interview, DiVincenzo describes the four steps in detail, as well as how to implement them in your own portfolio.
Watch the full video at:
Three messages before today’s interview educates and motivates you.
First, if you’re a domain name investor, don’t you have unique legal needs that require domain name technical know-how and industry experience? That’s why you need David Weslow of Wiley Rein. Go search for David Weslow on DomainSherpa, watch his interview and you can see for yourself that he can clearly explain issues, can help you with buy/sell agreements, deal with website content issues and UDRP actions, and even help you write your website terms and conditions. David Weslow is the lawyer to call for Internet legal issues. See for yourself at NewMediaIP.com.
Second, managing multiple domain name marketplace and auction site accounts is a pain. Inevitably, you forget to sign into one and lose a great domain…or worse. Now imagine using a single, simple-to-use and comprehensive control panel to manage all your accounts. That’s Protrada. You can setup search filters, analyze domains, automate bidding, list domains for sale, and buy domains across all major marketplaces. Protrada also has a new semantic engine that builds Google-friendly websites with rich content and network feeds. Sign up at Protrada.com to get 20 free credits and start building and monetizing your domains today.
Finally, if you have questions about domain names, where should you go to ask them? The answer is DNForum.com. Not only is DN Forum the largest domain name forum in the world, but it’s the best. You can learn about domain names and the industry, buy and sell domain names, talk about domain name news, and meet other domainers just like yourself. Register for a free DN Forum account and begin advancing your skills and knowledge today. And when you do signup, send me a friend request so we can connect.
Here’s your program.
Michael Cyger: Hey everyone, my name is Michael Cyger and I’m the publisher of DomainSherpa.com – the website where you come to learn how to become a successful domain name entrepreneur directly from the experts. Then, when you have a success to share, you can come on the show and give back as today’s guest is going to do.
Here’s today’s question: How do you dissect your domain name portfolio and generate 70 percent more revenue from it by doing that dissection?
Joining us to answer this question is Paolo DiVincenzo. I met Paolo at the TRAFFIC conference in Florida last October, and he’s a great guy, very knowledgeable. Paolo is the General Manager of DomainPower, a division of Domain Holdings.
Paolo, welcome to the show.
Paolo DiVincenzo: Thank you, Michael. It’s great to be here.
Michael: How did I do on your name, by the way?
Paolo: You did great!
Michael: Nice. All right. So, Paolo, I threw out a pretty big question at the beginning of the show. You’ve spent months working on this problem. Can people with a domain portfolio of hundreds or thousands—that’s just all over the place, and we all have them. We have dogs in our domain, we have decent domains in our domain portfolio, and we might have some great ones if we’re fortunate, but a majority of us have “dogs” to ok domains.
Michael: Can people take that type of portfolio in the hundreds or thousands—in the quantity of hundreds of thousands, and get them to 70% lift using this methodology that you’re going to be talking about today?
Paolo: They can. We’ve seen by taking domains off of a single PPC feed and putting them and segmenting out those domains in individual verticals and then rotating them amongst providers, we’ve seen that 70% lift. Your individual results may vary, but if you can take away one message from this, it’s that the key to it all is segmentation. Different domains are going to perform differently in different places, and the key to getting that 70% lift or whatever you’re going to see is really breaking up your portfolio intelligently and finding out where those groups perform best.
Michael: Is there a representative domain that you can throw out from this portfolio and tell us what the domain is, what the revenue was before, and what the revenue was after you did this? After you came up with this methodology and process for monetizing it?
Paolo: I’ll tell you one domain that we had actually put a lot of work into before it started performing. This was not simply a matter of taking a domain and testing it on different platforms and seeing where it ended up; it was more a matter of taking a domain, testing it, finding out what’s failing basically everywhere, looking at why it was failing, and then making some tweaks to the domain to get it to perform.
The domain I’m talking about is spiralfrog.com.
Michael: What is that?
Paolo: It used to be a download site for free downloads. It was blocked by one upstream provider, so one major PPC provider wasn’t even serving ads on it. That was one problem. The other PPC provider just didn’t recognize the term. They were serving up ads regarding frogs and other things that were irrelevant. It was just overall not monetizing well.
So what we actually had to do with this domain was we put it on what we call our “content framework,” and we had to populate it with some content so that the search engines could read it effectively and know what it was about. And it was about downloads, not frogs. That’s an example of a highly manual optimization process, but it’s an example of where you can take a domain that’s doing absolutely nothing beforehand, and tweak it to get it to perform.
Michael: What was the revenue for the 30 day test beforehand, and what was the revenue for the 30 day test after?
Paolo: To be honest, I don’t have those particular figures for that particular domain. the 70% that I gave you earlier, that was on a whole portfolio of domains, so that was an individual’s portfolio that we took off of our standard PPC feed and then started routing it in these groups through partners.
Michael: I think a lot of people, Paolo, and we talked about this beforehand—a lot of people are hesitant to believe some of the claims that people make in the industry. It’s completely acceptable. I have problems accepting some of the claims, too. You just never know if you can trust some of the data that some providers are giving you. There’s a lot of people out on the internet that say they’ll make you money and they’re just scams. So I want to say before we start this interview that Paolo was kind enough to send me a portfolio of 9800 domains—is that how many was on the list, Paolo?
Paolo: That’s correct.
Michael: And he showed me the revenue that they were generating from parking for a 30 day test period prior to this methodology that we’re going to discuss today, and then a 30 day period after he went through his methodology and system. I can verify it. I’m looking at the numbers, and it’s actually more than 70%, but it’s a phenomenal outcome from this, and so there’s a lot of domains here.
I don’t want to read any of the domains, Paolo, like we discussed in the pre-interview, but if there’s any—and clearly they range from some domains have very little benefit and some domains a tremendous amount of benefit. Actually, a lot of the domains have a tremendous amount of benefit.
Are there any from that list that you want to just throw out as a matter of example, or do you not feel comfortable doing that?
Paolo: I don’t feel comfortable doing that, but I gave the spiralfrog example. That’s one in particular that we did significant work on and that’s representative of what we do.
Michael: You have a four step process that you use to quantify this revenue lift and quantify your process that anyone can apply to their portfolio. We’re going to go over that process. But first, I’d love to find out what entrepreneurs and investors were doing just before their current work. What were you doing prior to becoming general manager of DomainPower?
Paolo: My background in domain is actually just five months old. But I’ve been working in the internet industry since 1999 in product management roles and in senior marketing roles.
I started my career with AT & T. I quickly hopped to a startup called iplace.com. I was their VP of Product Management there, and we were acquired by homestore.com. We produced real estate products for the internet—informational products, very similar to what you see in Zillow[.com]. Like I said, ultimately we were acquired, and then shortly after that, I started my own company.
My business was acquiring small e-commerce sites and streamlining their operations. We would bring them in house, make them profitable again, and then we would scale them up mostly throughout online advertising.
I’ve got a long history in optimizing and working with assets with potential that maybe are underperforming, and I’d say it’s probably that experience that’s been most relevant to the domaining industry.
Michael: People always want to know what kind of domains the interviewees have that come on the show. Do you have a domain portfolio yourself?
Paolo: I do. I have a very small one that actually came with acquisitions of e-commerce sites. Every e-commerce site I would buy would have maybe 10 or 11 periphery domains and yet I kick myself for doing this now, knowing what I know about the industry, but I used to just routinely let all of those expire. I let some phenomenal ones expire. I wouldn’t say phenomenal, but good enough to be making me some money now.
So I have a small portfolio. I have maybe 50 domain names that I keep.
Michael: I’m sure a lot of the domainers that are watching this show love you for giving back to the community during those prior days.
The 50 or so domain names that you have—are some of those developed e-commerce sites that you just run on the side, or are they just all—are you done with the development and those are just undeveloped domains?
Paolo: I’m all done. Prior to taking this job, I actually wound down my entire business, which was the process of selling. I had roughly 12 e-commerce websites. I actually grouped those up into industries and then sold them to competitors. I no longer hold e-commerce businesses, but I do own domains that were part of those businesses.
Michael: Are you allowed to use DomainPower to monetize those remaining domains you have?
Paolo: I do. I brought them all on the platform.
Michael: Excellent. Have you seen a 70% lift in revenue as a result of bringing those to the platform?
Paolo: No, because most of them are real e-commerce sites that are selling products, so I did not see a revenue lift, but I’m sure if they were on other alternatives, they’d be seeing that 70% lift.
Michael: Let’s go through your four step process. Let me give the audience an overview of the four steps because you shared your process with me beforehand, which I appreciate, and then we’ll go through each of the steps in detail.
The first step was “Establish a baseline metric for what you’re doing.” A lot of the DomainSherpa audience realizes that I came to DomainSherpa from publishing and I used to publish about the Lean and Six Sigma methodology for process improvement.
There’s a famous quotes out there: “You can’t improve what you don’t measure.”
I commend you for having the first step be to “Establish a baseline.” You need to measure it.
Your second step is to “Group domain names.” The third is to test them, and the fourth is to optimize them. That’s the high level.
Let’s go to the first step: Establish the metric in your baseline. How do you do this?
Paolo: The most important part of this is that everybody measures visitors differently. One thing I learned in the industry—there’s sort of a rough industry standard for how you do that, but the actual execution is so different from one company to another that the number of visitors varies so greatly that you really can’t use the visitors you’re getting out of their control panel as a baseline.
Michael: I never realized this before. So when you go to Parking Company A, let’s say, and they tell you on your dashboard on their system that your revenue per thousand visitors is a dollar and you go to Parking Company B and you have some domains over there, and they tell you the revenue per thousand visitors is two dollars, you can’t say B is better than A, because they might be measuring things differently is what you’re saying.
How is that? How come they don’t define it in the same way?
Paolo: Pretty much everybody looks at a unique visitor as somebody who’s visited once within the past 24 hours. That part tends to hold between companies. It’s how people measure that’s different.
Some people are doing it with a server and when the server sees a visitor, it’s a little more open to how it determines whether somebody’s a human being or not. Some measure it through the browser which is a little more restrictive, so the browser will filter out crawlers and bots and other things that are impersonating people. So there are just different ways of measuring that totally change the results.
What we definitely see is in looking at one platform versus another—there can be a 300% difference in how they’re measuring visitors which translates into a 300% difference in revenue per thousand visitors if you’re looking at equal revenue. The variance is all over the place. You can really pay no attention to the numbers that you’re given for unique visitors. You have to establish a translation yourself.
Michael: How do you do that?
Paolo: The first step is figuring out which—if you’re testing between two platforms, it’s figuring out which platform you’re going to use as the gold standard. You have to—in one platform is the one that you are going to use to measure the others.
Michael: Is it actually a gold standard, or is it just a baseline where you’re saying you’re going to compare everything to this one?
Paolo: It’s a baseline.
Michael: So you can pick any of them, you just need to say, “I’m going to compare the other ones to this one.”
Paolo: That’s exactly right.
Michael: So you do that, knowing that not every platform calculates your unique visitors the same way and your revenue per thousand. As a result of that, you then choose one as a baseline and you compare the others to it.
What kind of sample size do people need to take in order to establish their baseline?
Paolo: For visitors, you’re probably talking about—you want to at least look at 1,000 visitors. More than a number, what you’re looking for is a timeframe. So what you want to do is you want to take a group of domains, you want to let them run at least a week on one platform, and then let them run at least a week on another platform.
Assuming they’re getting pure type-in traffic, and there’s no paid traffic to the domains, you’re going to see more or less your multiplier between the difference of uniques on one versus the other over the course of the week. The reason you want to do it over the course of the week is because weekend traffic looks totally different than weekday traffic, and you definitely want to include a weekend in your time span.
Michael: So at least a week so you get that weekend traffic because you may have something that has more traffic as a result of consumers who have more free time on the weekend versus business users during the week which you may not capture on the weekend. At least a week, but if somebody really wants to do a good test, do you suggest two weeks or a month?
Paolo: I would go a couple weeks just in case you’ve had a weird spike one day, those few weeks will sort of smooth that out.
Michael: In the data you sent me beforehand, you actually did it for a month.
Paolo: That’s correct.
Michael: Was that just to get an even bigger sample size? To have a higher confidence level?
Paolo: We’re just talking about uniques. The reason why we would let things run for a month is beyond just traffic. It takes a while for revenue to ramp up on any given platform. So if a platform’s seeing a domain for the first time, it’s going to take a little while for its upstream provider to read into that domain and serve relevant ads, and then a lot of platform providers put some work into optimizing those domains. so we really don’t see them start to plateau out on performance until about two weeks.
So a thirty day test—two weeks is about right to get a general idea of how something’s going to perform in a certain place. A month is rock solid.
Michael: If you’re going over there, you’re probably wasting some money by not having them optimized.
Paolo: Yes, exactly. You have to make a decision one way or another.
Michael: So you use one platform as the basis and then how do you compare the other ones to that one?
Paolo: You run traffic on one platform for a given amount of time, and you run it on another platform for a given amount of time, and then you figure out the difference between the visitors and that difference becomes your multiplier.
So from now on when you see visitors on that other platform, you multiply it by the multiplier to translate it to your baseline platform visitors.
Does that make sense?
Michael: No, because how do you calculate the multiplier?
Paolo: I’ll give you an example.
Let’s say you take your group of domains, you run them on one platform, and you generate 1000 visitors. Over the course of two weeks, you take that same group of domains, and you run it on Platform B over the course of two weeks. It’s 2000 visitors. Now you know every time you run domains on that other platform, you have to multiply the numbers of visitors by 0.5 in order to translate RPMs back to your baseline.
Michael: Were you saying 1000 visitors or 1000 dollars?
Paolo: 1000 visitors.
Michael: But whether you’re running your domains on Platform A or Platform B, all the domains aren’t developed, so they’re all typed in traffic. So I would assume that they would have the same count of unique visitors on Platform A or Platform B.
Paolo: But they don’t, because it depends on the way that the platform measures any visitors.
Michael: So what you’re saying is they actually should be the same, but because they measure differently, you’re calculating the multiplier so that you can figure out the baseline. So you can look at the same group of domains across them.
Michael: I got that. That makes sense. So the assumption is that you do get the same amount of type ins over a long period of time just because they’re being—just because they’re parked. There’s no development going in, there’s no search engine optimization; just the common cause sort of variation of the domain name should match up between the two.
Paolo: That’s right. And there are some caveats to that. Once in a while, you’ll pick up a domain that’s got a lot of existing links. So it was once another website that had a lot of inbound traffic, and that stuff tends to expire over time. If you know that you have one of those in your portfolio, pull it out for a test, because it’s going to skew your results.
Michael: That makes sense. While you’re testing Platform A, and then parking Platform B, you are also generating revenue at those two places.
Michael: Do you want to record that as well?
Paolo: Yes, absolutely. This test records revenue, it also records your baseline. Ultimately what you do is you use that multiplier I was talking about to establish a common denominator between platforms in terms of uniques. Then what ultimately you’re looking to derive to compare the two platforms is that revenue per 1000 visitors that you’re talking about. That is the metric that you’re trying to get accurate to compare two platforms.
Michael: I think I get the first step. You’re going to take—if you want to compare your domains on Platform A and Platform B and Platform C, you take all of them, you move it let’s say to Platform A, you record the number of unique visitors and the revenue, then you go to Platform B for two weeks, you record the unique visitors and the revenue, you go to Platform C, two weeks, you do the it the same way—and then you can come up with the multiplier based on unique visitors for the three platforms.
So if Platform A has 1000 unique visitors, and Platform B has 2000 unique visitors, the multiplier from A to B is 2. So you want to divide, later on in this process, Platform B by 2 in order to get an equal amount.
Michael: And then you’re also going to have the revenue, but we’re not doing anything with that yet.
Paolo: No, once you have your baseline, that’s something you can use in all of your tests. So this is just the initial step is to get that baseline. Then you worry about revenue.
Michael: Makes sense. Now we’re on Step 2 which is to group the domains.
Paolo: That’s right.
Michael: What does that entail?
Paolo: This is super-critical. This is really where we found we got the majority of the lift is in intelligently grouping the domains. Grouping domains that behave the same for whatever reason.
The first grouping that we do—and that I would recommend a domainer do—is industry. It’s the most common one when you ask somebody what types of domains they have, generally the answer back has to do with industry: “I have travel domains. I have credit card domains.” That is definitely the first grouping you should do.
Michael: So that means that you go through your list of 50 or 500 domains and you try and group them by industry. So you take all the automotive ones and you put them at the top, and then you put healthcare ones, and then you put education ones . . .
Paolo: That’s right. And you can get as granular as it makes sense. One thing you don’t want to do is have thousands of groupings and then end up with 2 or 3 domains in each grouping. The whole point behind grouping is to combine the visitors within a group of domains so that you can get significant data back when you go and test that group.
So you have to make sure that that group has enough traffic to where you’re going to have significant results back in a couple weeks or else you’re going to deploy a group of two domains on a platform and you’re not going to have any results back for six months. There’s not much point to doing that.
Michael: That makes sense. Now what if people’s portfolios are just all over the place? Like what if they’ve got a ton of brandable domain names? What do they do with those?
Paolo: Those ones are tough. If they’re already receiving a certain amount—a good amount of traffic, then you can test those independently. If they’re not, though, and they don’t fit into a group, there’s no such industry as ‘brandable.’ You’re kind of stuck with an amorphous blob that you’re just going to have to test as group unto itself.
Michael: So the ones that just stand alone by themselves, you’ll—those are off to the side. What about high traffic domains? Ones that you know get a ton of traffic?
Paolo: Those should be tested independently. The reason why we group domains, or the reason why we advocate grouping domains is so you get significant data back. If you’re already getting that on a single domain, by all means, just measure that domain separately, because every domain—even if it’s in a certain industry or even if you can group it with similar domains, is going to behave a little bit differently by itself. If you can measure that and optimize it, then by all means do so.
Michael: If somebody is looking at their domain portfolio and they’re parking right now, what is a high traffic domain name, in your opinion?
Paolo: That’s tricky. It’s driven by many different things. What you’re looking for is enough data to make an intelligent decision. Now, whether you want 90% confidence in that decision or 80% confidence in that decision is up to you and that’s one of the things that drives how much data you’re going to need back.
My policy is a rough guideline, I wouldn’t make a decision until you get about 400 uniques on the domain and you have a revenue that ties into about 400 uniques on the domain.
Whether you’re going to run the domain by itself also depends on how long you’re willing to wait to make a decision on it. I wouldn’t wait more than two weeks to a month. So I’d be looking for a domain that has at least 400 visitors per month.
Michael: So if it does have at least 400 uniques per month, then that’s considered high traffic and you’re going to want to test that one on its own?
Paolo: I would. Like I said, it depends on your patience and how accurate you want to be with your decision. In fact, there are a lot of online tools which I use all the time to calculate your sample size. So what they let you do is you plug in your confidence level and you say, “I want 90% confidence or I want 85% confidence,” and then you put in some sort of measure—the variability and the results. In the case of, unfortunately, pay per click, it can be 5 cents or it can be 5 dollars. There’s a high degree of variability. Then you hit “submit” and it tells you what you need for your sample size.
I would suggest playing with those and seeing what kind of level you’re comfortable with. 400 kind of sticks in my mind as a rough rule of thumb for how many uniques you need to have to make the decision.
Michael: So you’ve got your portfolio, you take out your brandable ones which probably aren’t going to get very much traffic, but you want to keep because you think they’re going to be valuable over time, that you might be able to flip them. You bought it for some amount of money and you think you might be able to get some multiple of that in the future. You put your high traffic domain names which we just discussed, which you’ll probably want to evaluate separately, and I’m going to come back to that question.
But then you’ve got the masses in your domain name which most of us have, and you’re saying the first step in the grouping step—Step 2—is to figure out the industry segmentation. So you go through and you figure out what’s automotive, what’s healthcare. There are big industry groupings out there. If people aren’t sure that they are using the right industry segmentation, Paolo, is there a list on your website that they can go look at that you know are good segments that monetize, that anyone may want to consider?
Paolo: Yes, there is in fact. When we first launched DomainPower, we built up landing pages for the high paying verticals and so we’ve got those verticals broken down. I think we’ve got 14 of them and we have individual landing pages for those 14 verticals, and those are either high pay per click verticals—so things like credit cards, mortgages–or they’re high pay per lead verticals which is another way that we monetize. In either case, parking won’t make you a lot of money in those verticals, because people who are harvesting leads are also buying pay per click ads and they’re paying a lot for those.
Really what you’re looking for in terms of verticalizing—one of the first things you want to look at is industries with a high volume of advertisers who pay a lot per click.
Michael: I just visited DomainPower.com and I’m looking for that listing of industries where clearly you have them. I can see shopping, insurance, travel, but I don’t see a full list on here.
Paolo: I think if you—would you like me to take a look?
Michael: Yes. Can you? Let’s find this for people.
Paolo: Yes, we used to have a summary on the website.
Michael: Let’s find this so that we can refer people to this list to make sure, because you guys have spent a lot of time looking at all the different industries which is not just people searching for it, but you need the volume, and you also need the advertisers that are willing to pay for that, and so I know that I’d never seen a full list from somebody. I’m sure that a lot of domainers have seen it, and I know that a lot of domainers have their own list, but for those that are brand new to this, being able to refer to a list I think would be useful.
Paolo: I’m looking and I don’t see one, so I might have misspoken.
Michael: What we’re going to do is, Paolo, can you—after this interview, we’ll post the list either below the interview, or if you want to put it up on your website, then I’ll put a link to that from the interview page.
Paolo: Yes, that would be fine.
Michael: So we’ll do that.
So the industries are defined and people can look through theirs and they can segment them out. What are some of the highest PPC verticals that you’ve found that are currently?
Paolo: Credit card is a big one. Education is a really big one. Legal is another big one. Mortgages still is a big one; it was much bigger. Automotive is another very, very good one. Travel is another good one. Those are the ones that occur to me off the top of my head.
Michael: After you’ve segmented your list, then what do you want to do?
Paolo: After you’ve segmented your list, you’ve got it into industry groupings now. Look at the individual domains within those industry groupings and the next cut that I like to do is what I call “user intent.”
I’ll give you two examples that I like to throw out. Certainly within the travel industry, there’s a high degree of variation in the meaning of domain names. So somebody who’s looking for a site called VisitRome.com is looking for totally different content than somebody who’s trying to visit a site called RomeFlights.com. The RomeFlights person is just looking to compare the cost of flights to Rome or possibly book a flight to Rome. So in that case, what they’re probably looking for is some sort of Kayak-type engine where they can punch in their departure date and get the best fares. Somebody who’s looking to visit Rome is looking for content. They’re probably looking for information on the city, major sites, and in that case something totally different is going to drive conversion.
That is the next segmentation I would do, and that segmentation totally drives different results in terms of conversion.
Michael: If I’m thinking about visiting Rome and I know that there’s a lot of cities used to visit cityname.com as their branding because they didn’t grab the city name early enough in the registration process. If I’m going to type VisitRome.com, I may want to download a PDF or an e-book, or I may actually buy a book and have it sent to me, or I may want a snapshot of the city with maps and all the critical visitor interest points identified on that map, versus the RomeFlights that you’re talking about.
So you’re going to monetize them differently is what you’re saying.
When you have your domains broken out into your industries and then you look at the five within travel, how do you notate VisitRome versus RomeFlights user intent?
Paolo: It’s just a spreadsheet exercise. So you have your groupings in a spreadsheet and then what you’re ultimately looking at once you tag those domains, is you’re looking to come away with a list of domains which you’re going to test independently.
Michael: So you would test VisitRome.com differently than RomeFlights.com?
Paolo: It would be its own group. It would be a subgroup of travel. So you would have your group “Travel,” and then your subgroup of “want to explore a destination” and your subgroup of “wants to buy a ticket.”
Michael: Is there anything else that people need to think about as they’re segmenting their domain names within those groups?
Paolo: Yes, the other big segmentation we do—there’s three that we look at primarily—the other big segmentation is geography which is . . . The way we do it is we look at where our visitors are coming from. We do that dynamically on the server and then we route the domain to where that individual visitor or that individual visitor’s geography’s going to monetize best, and that differs between platforms. It’s a little tough to do if you’re a domainer looking at a portfolio and trying to do this manually, but what you can do is CCT [00:31:00 Unclear. all these? all leads?] and segment those out. For example, if you have a lot of co.uk’s, there are certain platforms that do better with co.uk traffic than with American traffic and you can route those differently as well and test them differently as well.
Michael: So if you have a bunch of domains because you love .ca for Canada or you’re from Germany, so you bought .de’s, you’re going to want to break those out separately from everything else because there might be better options to monetize those ccTLDs—country code top level domains.
Paolo: That’s right.
Michael: Is there any other things you’re going to want to do as you’re looking at your portfolio and grouping them?
Paolo: Yes. I think I’m going to talk about this a little bit later, but ultimately what you’re going to see is that some domains—for whatever reason—just don’t perform well. Even if they’re broken out. They look differently than the rest of the group. Those are unique things. I gave you the example of spiralfrog earlier and that’s a good example of something that was just sort of mystifying and needed to really be manipulated on its own to do anything.
The three segmentations I mentioned are the big ones I was going to cover. 99% of the optimization—there’s always going to be that 1% outlier that needs a little help on its own.
Michael: So that’s step 2. You want to segment your domain names, look at the user intent within each of those segmentations, and see if you can segment them further, and then look at geos also and you’ll want to separate those further.
Michael: So step 3 is, now that you’ve done all this segmentation, you know how things should be breaking out, what is step 3?
Paolo: Step 3 is to test them. I’ve alluded to that’s what we’re going to do next, but you take the groups of domains and you look at how much traffic they’ve received, and then you determine how long you have to run them on a given platform to get significant data back in order to make a decision on how that platform is performing.
Michael: So we ran the test before and we know that certain domains get a certain amount of traffic already.
Michael: A, B, C domain dot com. You ran it in Step 1. You got your baseline, let’s say it’s 100 unique visitors within a 2 week period. So I group that into—I don’t know—some grouping, let’s say. And what you’re saying now is that I need to understand the sample size necessary to get a statistically significant data pool so I can test whether my new result is valid against my old result?
Paolo: That’s right. So you look at your group of domains. Let’s say I have a group of domains and it totals up to 50 visitors a week. You look at that and say, “That grouping might be a little bit too small. I don’t feel like waiting four months for my data to come back and make a decision.” So you might want to take that group and combine it with another group. And then once you have a group that has enough visitors to get you back data in a timeframe that you’re comfortable with—like I said, roughly speaking, try for under a month—then you take that group of domains and put it on a platform for that time period that it’ll take you to get data back.
Michael: I understand. So you’ve done this analysis on sample size using the calculators online and people can go search for sample size calculation. You’ve determined that probably about 400 visits—unique visitors—per grouping will give you a 90% confidence in the results?
Michael: So you’re saying if you took a group and you moved them to a—that was getting 50 visitors per day, let’s say—you would need to run it for four days at a platform—I’m sorry, eight days—in order to get 400 visitors to be able to say you have a high confidence interval in that new data that you’re getting?
Paolo: That’s correct. And I’ll add one more thing to that is you really want to give a group of domains at least two weeks to run on a platform because it takes that long for performance to ramp up. I wouldn’t go for shorter than 2 weeks.
Michael: So two weeks and/or 400 unique visitors, whichever is longer?
Paolo: That’s right. And the 400 visitors is just sort of a rule of thumb. I would hate for somebody to stick to that and think it’s gospel. It’s a rule of thumb I use in my head. It seems to work out that way if you’re looking for 90% confidence, but by all means get going. Play with those calculators that I was talking about to determine the real numbers for your portfolio.
Michael: Let’s just say it takes a month to get 400 visitors for me because my domains really stink—which they do. So then, I’m going to go to the same platform that I did my baseline on? Am I going to run them there again?
Paolo: No, if you’ve already got data on those domains, don’t bother. You’re looking for new data.
Michael: So I’m going to go to Platform B, and I’m going to put them there—that group of domains, but I’ve already run them at Platform B to begin with, right?
Paolo: You ran at least some of them to get your baseline. When you’re looking for your baseline for unique visitors, you don’t have to run your whole portfolio on that. All you’re looking for is a translation. You can use a representative sample of your domains to get enough unique visitors back to develop that visitor multiplier.
Michael: So you can take, say, 30 domains out of your 500 domain portfolio, leave all the rest at parking where you’re generating money, and then you can just move between them and get your multiplier number?
Paolo: That’s right.
Michael: And then after you get the multiplier, now you’re—you move forward. You’re now on Step 3, and actually you can be doing your Step 2 while you’re getting that multiplier, and now you’re on Step 3 and now you’re going to take that group of domains, move them over to Platform B and run them for an appropriate amount of time like we’ve discussed.
What kind of data are you gathering?
Paolo: You’re gathering revenue and unique visitors. That’s basically all you need. Some platforms give you click through rates which is anecdotally interesting, but ultimately all you’re trying to get through is that revenue per 1000 visitors. That is going to be what’s going to make your decision.
Michael: Is there anything else that we’re going to do in the testing phase?
Paolo: No, that’s it. Just run them for no less than two weeks, get your data back, and then determine your revenue per 1000 visitors and that’s what’s going to make the decision.
Michael: So that’s Step 3. A lot of these platform—these parking companies will allow you to group your domains, so you can move them in easily and keep them in a group and then look at the revenue per 1000 that’s coming in?
Paolo: That’s exactly right. Most platforms will let you develop what they call folders or they call them portfolios and they’ll let you segment out your portfolio.
Michael: And it’s actually going to tell me the revenue per 1000 by domain, right?
Paolo: That is right. It’s going to tell you your revenue per 1000 by domain, but don’t pay attention to what they tell you. Make your own number. It’ll tell you revenue, it’ll tell you visitors, and then you do the translation, and then you determine the revenue per 1000 visitors.
Michael: So your final step is to optimize. So I assume that you’re going to take this data and now you’re going to do something with it?
Paolo: That’s right. So you’re going to take a look at this data and you want to look for those outlier instances where something’s really just not performing. Oftentimes what’ll happen is if a platform is using a certain upstream provider, be it Google PPC, or Yahoo! PPC, that upstream provider will have blocked your domain.
There’s really—it’s not predictable which domains they block, unfortunately. The domain could have an ancient history that has nothing to do with you. It could have a history of arbitrage or something else that’s given it a bad name with one of these providers and they’ve blacklisted it and they just won’t serve ads on it.
What’ll happen sometimes is some platforms will just not serve ads on the domain, or they’ll serve ads from a third tier provider which don’t monetize particularly well. So look for those outliers that are just doing really poorly. That’s part of the optimization.
Michael: Most of my domains perform poorly. Why aren’t they all these outliers that are maybe even blocked? How do I know that a domain should be performing well, but isn’t?
Paolo: One thing you can do is you can try it on different providers. If it does really well on one and really poorly on another, there’s something to that.
Michael: So looking across the different parking companies will tell you if that’s a potential outlier because one may use a direct Google feed and Google is blocking it for some reason, but another one may use third party leads where you can’t be blocked and people are paying you for filling out a lead form and so that would be a good way to compare the two..
Paolo: You’re exactly right. It’s more likely the Yahoo! versus Google difference, so Google, they try to have exclusive providers. So if Google is blocking the domain, there’s really not a lot the platform can do for that domain. If they are blocking it, then send it to somebody who’s got a Yahoo! feed, and chances are, they’re going to do better with it. These are the sort of differences that you’re looking for.
Michael: What do you do if you do find some of those differences?
Paolo: Then you pull the domain out of its group because it’s performance is not relevant to that group, and you put it in its own group and then you call that group “Google blocked” or “Yahoo! blocked” and you send that to a Yahoo! or a Google feed provider depending on where it’s going to perform best.
Michael: Can some of these providers actually help you improve your performance within these different groups?
Paolo: They can, and that’s a huge optimization step and most of them, they have a vested interest in seeing you do well on their platform, so—
Michael: Because they make more money as well. They—most of them are taking a cut of the revenue generated, right?
Paolo: That’s right, so, in every case I would establish a relationship with the parking platforms that you’re testing out. Let them know which domains you have on their platform, let them know which domains you expect to do well, and let them optimize, because a lot of them—they can do manual work to help those domains do better.
Michael: I know you know all the providers out in the space, and I know some of the providers out in the space, because we go to the conferences, so we either meet the founders, or we meet the people that are in charge of the company. But most of the people watching this show—they don’t have access to those people. Is it as simple as saying, “Hey, I’ve moved these ten automotive, car buying related domain names to your service and I want to optimize them.” And email into the help desk or something and say, “Are there some seed terms or is there something else that I can do to improve their performance?”
Paolo: I never looked at it. I’ve never been in that position where I had to do that, but I know everybody’s hungry for customers and everybody knows that the key to a customer’s heart is going to be producing the most revenue for them. I would hope that you could—it might take a little perseverance, but I’m hoping that you could find somebody who was willing to lend a hand in doing that.
Like I said, they all have ways of optimizing. I think you had mentioned seed terms earlier. That’s a huge one, so if they see that they can seed a domain with a certain term or feed that domain to the upstream provider so that it can serve back relevant ads, that’s something they do. And then they put that the domains on relevant templates. That’s another thing that they can do.
To use your example—an automotive domain—they can make sure that’s sitting on their automotive template or whatever they’ve tested and proven to be the best-performing template for that given domain.
So there’s a lot of work that they can do, and like I said, they have an interest in producing revenue for you, so I’d hope that they’d lend a hand.
Michael: So we know our multiplier, let’s say, from parking company A to B is 1000 uniques to 1100 uniques. So the multiplier on B is 1.1 compared to A. And then we know the baseline revenue of those domains in parking company A and then we know the revenue of those domains in parking company B.
If B is making less money or more money, how do we use the multiplier to be able to compare it directly to parking company A?
Paolo: If it’s making more or less money on an absolute basis, or an RPM basis?
Michael: On an RPM basis, I’m sorry. You tell me what we want to look at. How do we compare the two using the multiplier?
Paolo: You just made the comparison. That is what you want to look at. You want to look at the RPM for the group between the different providers. Once you’ve done your translation of the visitors using the multiple we talked about.
Michael: That’s what I wanted to know. How do you take that multiplier and convert the RPMs so that it’s apples to apples?
Paolo: First, you leave the revenue alone. The only thing you translate is the denominator. Parking platform comes back with saying, “This group of domains had 10,000 visits,” and you multiply that 10,000 by your multiplier. You leave revenue where it is, and then you—whatever the calculation is on your visitors, that becomes the new denominator, and that’s how you calculate your RPM.
You don’t touch revenue. You don’t touch a translation on the RPM. You do it on just the visitors.
Michael: So they’re giving you RPM numbers, but you don’t care about those. You take the unique visitors and then the revenue that you’re generating and then you just put the revenue over 1000 visitors?
Paolo: That’s right.
Michael: You generate your own RPM. And then you want to use the multiplier on the unique visitors to bring them back to the same level?
Paolo: I’ll start from the beginning. So your platform hands you back two numbers. They hand you back $20 and 1000 visitors.
You take that 1000 visitors, multiply it by your multiplier that we came up with earlier, and let’s say the number ends up being 2000 visitors. So you take your revenue number—I forget what I said it was. $20? And you then put it over 2000 visitors and then the result of that is your RPM that you’re going to use to compare.
Michael: Just so it’s completely clear because I’m not the sharpest tack, parking company A and parking company B—when you were up at Step 1, “Establish the metric in your baseline,” about 1000 visitors, 1100 visitors . . . so parking company B has a multiplier of 1.1, right?
Michael: So then when I go to Step 4, and I take those domain names and I know what my revenue is on parking company A because they’re all parked there and I move those five automotive domains over to parking company B, and I have, let’s say 800 visitors, I’m going to take 800 and I’m going to divide it by 1.1? Or multiply it by 1.1?
Paolo: You are going—and you’re talking about 800 is parking company B or A?
Paolo: Multiply it by 1.1.
Michael: So I’m going to multiply it by 1.1 and then I’m going to put it over—I’m sorry, get it to a 1000 basis, and I’m going to divide the revenue by that and that’s going to get me the revenue per thousand visitors.
Paolo: That’s correct.
It’s good that you’re bringing this one home. I started out with the whole “Establish your baseline” and that is absolutely key. We get that, and we spend a good portion of our day doing those translations because there’s a lot of misconception in the industry when people come to you with their revenue per thousand visitor numbers that those—that’s a tried and true number, that they vary so much you really have to start at the beginning to do the calculation.
Michael: Exactly. And everybody’s defining it differently, which is something that I didn’t realize was happening.
So it is important to make that determination on a revenue per 1000 basis and not just trust what they’re giving you.
Paolo: That’s exactly right.
Michael: In general, what have you learned? After going through this exercise with your 9800 domain names and then others that you’ve done after that?
Paolo: I’ve learned that the key to it all—the really big difference—is different domains perform differently in different places. It’s really that simple. If you’re looking to make the most out of your portfolio, the secret is not looking for performance across the board. It’s not taking from one provider to another to the next and seeing how it does. It’s really breaking it out, and that’s where the lift is.
It is incredible. You would think that there would be, say, one dominant provider who uses a Google partner as an upstream—that is just not the case. There may be somebody who dominates in 80% of the cases, but there’s going to be 20% of the cases where it makes sense to take domains and put them somewhere else.
There really is a high degree of variability. To be honest, at this point I’ve been in this job for 5 months. We’ve had our decision engine in place for 2 months. At this point I haven’t really figured out the reasons behind those variations, but I can definitely confirm that they exist.
Michael: We ran an interview with Howard Hoffman a few weeks ago. Howard has spent a lot of his career doing exactly this. We didn’t go into how he does that, but he has determined the exact same thing. Gambling and casino-related domain names will perform on a certain platform better than any other platform. So knowing which domains are performing where will give you that edge that you need in order to be able to generate enough revenue to pay for the domain registration for the year and / or get a multiple above that.
Which types of—what was sort of a shocker for you, Paolo? Which types of domains would you think would perform better across, but then don’t? They ended up being dogs?
Paolo: I can’t—I mentioned that spiralfrog one already three times. That’s an example of an individual domain. But the one that took me a little while to grab onto was definitely this user intent segmentation. I was using all the classical segmentation—vertical is the oldest, vertical in our industry is really the oldest classification around, and then geography is an old one. But this user intent one—that was a big “Ah ha!” moment when we started to look at these different industry classifications and how individual domains were performing and very differently between them. It was a realization that just because you can stick something into a vertical doesn’t really mean that you understand what the user is looking for when they get there.
Michael: This process makes perfect sense to me now. I can follow it. If I were to do this for my portfolio, it would take a while because now that I’ve—now that I understand my portfolio better, which is great, and it’s going to help me run my business better, and it’s going to help me monetize my domains better, it’s going to help me sell them for more money in the future because I understand what I own better—that’s actually a lot of work. There are a lot of parking companies out there, right?
Michael: How many parking companies would you say there are? Let me ask you this: How many monetization platforms do you have plugged into DomainPower?
Paolo: We have 31 in DomainPower right now. Now those include platforms. Your typical parking providers and those also include different monetization models. We have some direct advertiser relationships. We have zero click providers which I’m not sure if you’re familiar with those, but basically somebody types in a domain and if you have a bidder for the traffic on that domain, we automatically redirect to that bidder.
So we’ve got—our monetization partners run the gamut. I can tell you major partner providers—there’s at least 25 of them that I can think about there. You’re right. There’s a lot of people to choose from in terms of testing.
Michael: So two weeks and then 25 providers per two weeks, plus I have to move them all and everything. I don’t want to make this into a promotional commercial for DomainPower, but clearly if people have watched, or listened, or read this far into the interview, they’re interested in this topic, they understand it better.
I want to understand more about what DomainPower does. You’ve used this four step process on your own domain portfolios and Domain Holdings has a lot of domains. I can come to you with my domains and you can do this for me and then I don’t have to do all the work myself. Is that correct?
Paolo: That’s correct. When we do it for you, the bulk of our work is automated, so we build what we call the monetization engine which I’ve talked about previously but what it does is it basically takes your domains in and runs them through this whole process that I’ve described. It takes your domains, puts them in the sub groupings that I talked about, and then it starts to rotate them on different platforms and actually pulls in real revenue data, does the visitor translation that I was talking about earlier and determines where they’re doing best.
Where it differs a little bit—from other standard testing that you might be able to do, is we also have a whole warehouse of predictive data based on what we’ve seen with similar domains in the past that gives us an accurate gauge of how we think a domain’s going to do and where it’s going to do best. And we basically send the bulk of its traffic to where we think it’s going to do best right off the bat.
So it’s not a long ramp up to get it to speed. We start it up pretty high, and then we optimize it from there.
Michael: It makes sense that I can give you a portfolio of 1000 domain names and you can segment it by keywords related to the verticals, and by the geographic factors like the ccTLDs or the IDNs, but how do you get the user intent in an automated programmatic fashion like you?
Paolo: We don’t. That user intent part is a manual piece and that’s something we do just looking at groups of domains within industries and then segmenting them out. I don’t know if there’s a way to do that—I’m sure there is a way, but we certainly have it figured out, and that is a manual piece.
Michael: You’re taking my domain portfolio, you’re actually adding a lot of your technology to it, you’re adding a manual review of every single domain name to it. Do you charge for this?
Paolo: We take a rev share, so a share of the revenue that we receive. One other manual piece that we do also is we talk with all of our platform partners and we negotiate the best rate that we can with them, so in addition to using our technology, you’re also using our buying power. We bring a lot of bulk to these providers and they usually get us pretty good rates as a result. We get rates that would be hard to get if you’re just coming in off the street and working with them indirectly.
Michael: I believe that. You’re in the industry. You’ve got much more clout than I do with my thousand domain names because you’ve got tens or hundreds of thousands of domain—how many domain names do you have in the DomainPower platform?
Paolo: We have hundreds of thousands.
Michael: But then I have to give you a portion of that revenue for the value added effort that you’re putting into this. Hopefully you’re making me more money, so clearly you deserve some of that. Do you disclose what that revenue share is?
Paolo: We do. If you—when you come in and sign up on the website, it’s a 60/40 rev share in your favor, and then if you are a significant customer, then it’s negotiable.
Michael: So if I’m bringing you 100,000 domain names, it’s negotiable?
Michael: I’m going to go come on your system. You’re not going to charge me anything to sign up. You’re taking my domain names, and then you figure out on these different platforms based on all the factors that we talked about, plus your predictive index—your data on the hundreds of thousands of domain names that you’ve had in the past—where best to put it.
Let’s say that I give you a grouping of domain names that are education related and based on prior experience and all the factors, you put it at parking company A for a monetization platform, let’s say A, because it could be a lead, it could be any number of ways people are actually buying that domain traffic and rerouting it to another website.
How do you test that over time to make sure that that’s still—like six months from now or a year from now—the best monetization platform for my domain name group?
Paolo: That is a great question. We never make the assumption that something that was true a month ago is going to be true indefinitely. I’ll tell you the way we do it internally. We always have a certain percentage of traffic as test traffic. So our decision engine actually routes traffic to the providers that do the best with it and the percentage that it routes to those providers is proportionate to the extent to which they outperform other providers. So if you’re just killing it—if a platform’s just doing awesome across an entire group of domains—we’re going to send 99% of the traffic to that given provider. If they’re doing okay versus Platform B, we’re going to send maybe 60% to that provider.
There will always be at least 1% test traffic just to make sure that that performance holds over time.
Michael: I give you my domain names, you’re optimizing them over time continually and I don’t need to worry about it.
Michael: What else about DomainPower do you do that somebody might want to know about?
Paolo: We do have other services besides our monetization platform. We also do premium brokerage. So if you have domain names that were—you’re looking to sell, we have a team of nine people who sit in an area of the office called The Pit and they make calls all day and they try to find users and get you top dollar for those domains. So that’s another one of our services.
Michael: Anything else about DomainPower that you want to talk about?
Paolo: No. We’re having a new website coming out soon, so probably by the time this video airs, we’ll have our new website up and we just encourage you to call us. What we like to do is talk with you and find out how you’re doing on other platforms and we will be straight up as to whether we think we can beat it by using our monetization decision engine.
Michael: That sounds great. Here’s my final question for you, Paolo. I have seen other—I know yours is the monetization decision engine, but there are competitors out there that do similar types of things. They either rotate your traffic or they have said that they will optimize the process.
How should a user decide between your platform or their platform when they’re plugged into, say, 30 monetization systems? Is that the kind of thing that they should try your system for six months and then try their system for six months and then pick the one that generates the most revenue, or . . . How would you suggest to them that they make that decision?
Paolo: I think you should always be—ultimately we’re all after making—any domainers looking at making the most money they can off their portfolio. So by all means, use that objective metric. Test us against them. No problem at all.
I will tell you what we do a little bit differently than a lot of people. I mentioned the notion of predictive analytics. So we take a whole history of data, we map your domains to similar domains, and then we use our knowledge about those other domains to start your domain at a high yield by sending it to partners that we think are going to perform the best for you. That’s a little bit different. I’d say a lot of platforms start assuming they know nothing about a domain, and so it really starts from scratch in terms of testing. It tests 50% of its traffic everywhere until it can get some earnings.
So we’re a little—in my opinion, we’re a little bit quicker at getting that initial ramp up. The other place where we differ is that we don’t rotate individual domains. We group domains because based on our experience, the average domain on our platform gets one visitor per day. It’s going to take you a year to get significant data back. The only answer in order to optimize and to have enough data to make intelligent decisions is to group domains up. So that’s something that our platform does which is a little different.
And then the other thing we do is we have accounts directly with other platform providers and we use our buying power to negotiate up their rates. So we’re not just bringing your accounts into our testing platform. These are our accounts, and we’ve negotiated up the rates.
Michael: I just want to reiterate again because you were kind enough to share in confidentiality with me some of your data that you collected before and after using this four step process, and it is a phenomenal lift.
When people—oftentimes, they look at some of the headlines that I put on the website—and I can say that I’ve looked at this data. I’ve seen your domain names, and I will maintain it confidential, but I’m appreciative that you provided it to me so I can say that I’ve given it the Mike Cyger stamp that I can see what you’ve done with these domain names and it’s a great process and technology that you’ve built.
Paolo: Thank you very much. It’s been great talking to you, Michael.
Michael: So if you have a follow-up question, please post in the comments below, and we’ll ask Paolo to come back and answer as many as he can.
If people want to follow you, Paolo, are you on twitter? Can they follow what you’re doing as you’re going through this process and making more money for customers and continuing to get into domain parking and monetization?
Paolo: Yes, sure. Our company has a twitter account, I believe the handle is Domain Holdings and we post there often. We post some of the premium domains that we have for sale as well as updates on the platform
Michael: Great. And now I’m going to urge the audience as I always do at this point. If you received value from this interview, please take a moment to write a comment to Paolo and say thank you for sharing your four step process, thank you for providing us some insight into the industry that so often is just a black box. Another way you can say thank you is to like this on Facebook so you can share it with your network if you have a lot domaining friends on there or entrepreneurs on there. Tweet it. Google+ it. Any of those social networks are great.
Paolo, if somebody has a question about the process that they don’t necessarily want to post here, is there an email address they can reach you at?
Paolo: Yes, they can reach me directly at Paolo dot domainholdings dot com.
Michael: That sounds great. Paolo DiVencenzo.
Paolo: That’s P-A-O-L-O. I just realized it’s not as intuitive as it is to me.
Michael: It is not. I’m glad you spelled that for our transcribers also.
Paolo DiVincenzo, General Manager of DomainPower at Domain Holdings, thank you for being at DomainSherpa, sharing the details of your platform and helping others become successful domain name investors.
Paolo: Thanks so much, Michael. I appreciate the opportunity.
Michael: Thank you all for watching. Well see you next time.
Watch the full video at: