Whenever domainers discuss sales, the topic of sell-through rate (STR) almost invariably comes up. Someone selling 5 domain names a year has a very different connotation if their portfolio is only 100 domain names, compared to another investor selling 5 names from a 10,000 name portfolio.
The NamePros Blog has considered Sell-Through Rates (STR) several times, including the 2019 article Domain Name Sell-Through Rates, but let’s look at the topic with some fresh insights.
1. STR Is A Rate
While we sometimes use the numerical value in other ways, such as a probability, it is important to keep in mind that Sell-Through Rate it is a rate. Therefore STR is, at least directly, something you calculate after the fact. For example, if you held 500 domain names over the past year, and sold 2 domain names, your STR would be 0.4% {=2/500}.
2. The Time Period Is important
While the normal assumption is that the STR is calculated on the basis of one year, the so called annualized sell-through rate, if something is just called the STR, it could be based on a different time period, like all time.
Let’s say that I listed 1000 names on a certain market for the past three years. I sold 1 in the first year, 4 in the second year, and another 4 in the most recent year. My lifetime STR would be 0.9% {=9/1000}, my annualized STR in the most recent year would be 0.4% {=4/1000}, and my annualized average STR for the three year period would be 0.3% {=9/(1000*3)}
3. STR Is Not The Whole Picture
While STR is one metric of domain portfolio performance, STR does not tell the whole story. Let’s say you have 1000 names and sold just 1 domain name the entire year. Your annualized STR would be a paltry 0.1%. However, if you sold that domain name for $100,000, and that was far more than your acquisition and holding costs, then your year might have been very successful.
If looking for a single metric to track how well you are doing I would suggest the Dollar Volume Per Listing (DVPL). That measure takes into account both your sales volume and the number of names that is spread over.
4. Don’t Focus Too Much on Industry Wide STR Values
Each domain name is different, and each portfolio is different. While it may be helpful to have an estimate for an industry-wide STR, or a subset say the industry-wide STR in a certain extension, the only STR that really matters is what you can achieve in your portfolio.
5. You Might Still Have 0 Sales
A few days ago on X, Michael Bang wrote a post with an eye-catching lead that said you would need 691 listed domain names on a certain marketplace to guarantee one sale a year. I think many just looked at the lead sentence, without delving into the long post to discern what Michael was really saying.
The central point of Michael’s post was that the expected value is not the value each investor will achieve in any given year. Let’s say a domainer has a portfolio of 100 names and their annualized STR is 1%. The expected average number of sales in a year would be 1 {=0.01*100}. However, if we had a population of 1000 similar investors, many would sell 1, some would sell 0, and a few would sell 2, 3, or more domain names in the year.
Michael was interested in how many, despite a STR of 1%, would still sell zero domain names in the year. One way to approximate that is to multiply the probability of each name not selling to find a total probability that none of the names sell (this assumes that each is an independent probability). Here is Micheal’s math, included here with his permission.
Analysis courtesy Michael Bang from this post.
We see that 99 times out of 100 (assuming a 1% STR) each name will not sell in the year, or altogether about 0.37 probability that not any of the names sold during the year. We simply subtract that from 1 to get the probability that at least one name did sell. Michael finds that even if you have a 1% STR, there is still only 63% chance you have at least one sale during the year.
There is a whole topic in math around predicting how many discrete events occur during some time interval. It is called the Poisson Distribution. Here is Michael’s analysis using that, getting a result approximately the same as the first way he calculated it.
Analysis courtesy Michael Bang from this post.
The mathematically inclined can use the Poisson distribution to estimate the likelihood for different number of sales in a year from a portfolio of a certain size for a given STR assumption.
6. Annualized STR Can Be More Than 100%
While at first glance you might think that 100% is the maximum possible annualized STR, that is in fact not correct. The annualized STR means the rate that would be sold if domain names were held for a one year period.
For example, let’s say I just started in domain investing, and acquired 10 domain names, and sold 5 of them over a period of just 4 months. That is pretty unrepresentative of the market, but if it did happen, the annualized STR would be 150% {=5/(10*(4/12))}. It is at least theoretically possible that the annualized sell-through rate is more than 100%.
7. STR to Predict Probability of Sale
Let’s say you have been in domain investing for many years, have a stable portfolio size of 2000 domain names, and, while there is some variability from year to year, your annualized STR is around 1%.
If you acquire a new name similar in quality to your overall portfolio, you can use the past STR to predict the probability of that name selling during the coming year is about 0.01, or 1%. Probabilities are normally computed on a scale from 0 to 1, but can be expressed in percentages by multiplying by 100.
8. Same STR, Different Uncertainty
Let’s consider two investors.
It is a crude approximation, but often the square root of a number is taken as roughly representing the standard deviation. In a normal statistical distribution, there is about 68% chance the value falls within 1 standard deviation, 95% within 2 standard deviations, while more than 99% fall within 3 standard deviations.
So if we go back to our two domain investors, A sold 4 names, so the square root of that is 2. That means that 68% of the time, if many similar investors were considered, the number sold would vary from 2 to 6 sales in a year, and 95% of the time from 0 to 8 sales per year. If we express those as STR, we don’t know if the real STR, long term, is from near 0 to 8%. So on the basis of this one year with limited sales, we can’t conclude much about long-term STR with any confidence.
Investor B had 400 sales, and the square root of that is 20. Therefore 68% of the time, if we looked at the same investor over many years, or many similar investors over one year, the number sold would be from 380 to 420. In 95% of the time it would be 360 to 440, with a corresponding uncertainy in STR only from 3.6% to 4.4%. Even if we insist on better than 99%, still the uncertainty is only from 340 to 460 sales, with corresponding STR range from 3.4% to 4.6%. In summary, if we take 95% of the time as the standard:
Read more on normal distributions here, though be warned that the math treatment is heavy.
9. Many STRs
@Recons.Com has suggested that it is useful to think not of just a single STR for a portfolio, but imagine your portfolio as having a number of buckets, each with their own STR. If you have a large enough portfolio, with significant number of sales. you can calculate multiple STR values.
For example, depending on your portfolio you might have a bucket for legacy extensions, one for created brandables, and one for .io or .ai names. Keep in mind the uncertainty when dealing with small numbers, though.
10. STR of Terms
If you use the Domain Insights tool available at Atom, for many names it will tell you the STR for the term based on Atom (SquadHelp) sales data. Therefore as well as considering STR for different extensions, or types of name, STR can be applied to rates that certain terms sell.
It is only part of the picture, just because names with say the term ‘health’ sold with a healthy STR does not mean that any combination of two words that include ‘health’ will sell. I don’t know for sure, but I suspect the term STR values at Atom are not annualized STR, but over the life of the marketplace. Nevertheless, I find this information really helpful when considering acquiring names.
11. Is STR Good For Anything?
So if DVPL (dollar volume per listing) is a better metric of how you are doing, why does STR matter at all? I think the most important use for estimations of STR is in deciding if a particular domain name is worth holding long time.
Multiply your best estimates of STR for this name, times the net return on likely selling price, and compare that to the annual holding cost. True, your STR will be an estimate, with uncertainty, but this framework is helpful. This overlooks any parking revenue, or considerations of changing value in a trending name.
Figure from Is Domain Name Investing Profitable? by Bob Hawkes at NameTalent.
Read more on this technique in the final section of the article Process to Rate and Price a Domain Name – Part 2.
Dividing your portfolio into buckets makes sense in this regard. For example, the annual holding costs for .ai, .io and some new extensions are higher than for legacy, so you need to make sure that is justified on the basis of estimated STR and selling prices.
Final Thoughts
I think STR is something that investors should consider, but don’t focus only on sell-through rate, it is just one of the factors that needs to be considered.
For additional reading on this topic, as well as the 2019 article Domain Name Sell-Through Rates, I would recommend What Are The Odds? Applied Probability for Domain Investing.
Thanks to Michael Bang for permission to share part of his analysis in this article.
The NamePros Blog has considered Sell-Through Rates (STR) several times, including the 2019 article Domain Name Sell-Through Rates, but let’s look at the topic with some fresh insights.
1. STR Is A Rate
While we sometimes use the numerical value in other ways, such as a probability, it is important to keep in mind that Sell-Through Rate it is a rate. Therefore STR is, at least directly, something you calculate after the fact. For example, if you held 500 domain names over the past year, and sold 2 domain names, your STR would be 0.4% {=2/500}.
2. The Time Period Is important
While the normal assumption is that the STR is calculated on the basis of one year, the so called annualized sell-through rate, if something is just called the STR, it could be based on a different time period, like all time.
Let’s say that I listed 1000 names on a certain market for the past three years. I sold 1 in the first year, 4 in the second year, and another 4 in the most recent year. My lifetime STR would be 0.9% {=9/1000}, my annualized STR in the most recent year would be 0.4% {=4/1000}, and my annualized average STR for the three year period would be 0.3% {=9/(1000*3)}
3. STR Is Not The Whole Picture
While STR is one metric of domain portfolio performance, STR does not tell the whole story. Let’s say you have 1000 names and sold just 1 domain name the entire year. Your annualized STR would be a paltry 0.1%. However, if you sold that domain name for $100,000, and that was far more than your acquisition and holding costs, then your year might have been very successful.
If looking for a single metric to track how well you are doing I would suggest the Dollar Volume Per Listing (DVPL). That measure takes into account both your sales volume and the number of names that is spread over.
4. Don’t Focus Too Much on Industry Wide STR Values
Each domain name is different, and each portfolio is different. While it may be helpful to have an estimate for an industry-wide STR, or a subset say the industry-wide STR in a certain extension, the only STR that really matters is what you can achieve in your portfolio.
5. You Might Still Have 0 Sales
A few days ago on X, Michael Bang wrote a post with an eye-catching lead that said you would need 691 listed domain names on a certain marketplace to guarantee one sale a year. I think many just looked at the lead sentence, without delving into the long post to discern what Michael was really saying.
The central point of Michael’s post was that the expected value is not the value each investor will achieve in any given year. Let’s say a domainer has a portfolio of 100 names and their annualized STR is 1%. The expected average number of sales in a year would be 1 {=0.01*100}. However, if we had a population of 1000 similar investors, many would sell 1, some would sell 0, and a few would sell 2, 3, or more domain names in the year.
Michael was interested in how many, despite a STR of 1%, would still sell zero domain names in the year. One way to approximate that is to multiply the probability of each name not selling to find a total probability that none of the names sell (this assumes that each is an independent probability). Here is Micheal’s math, included here with his permission.
Analysis courtesy Michael Bang from this post.
We see that 99 times out of 100 (assuming a 1% STR) each name will not sell in the year, or altogether about 0.37 probability that not any of the names sold during the year. We simply subtract that from 1 to get the probability that at least one name did sell. Michael finds that even if you have a 1% STR, there is still only 63% chance you have at least one sale during the year.
There is a whole topic in math around predicting how many discrete events occur during some time interval. It is called the Poisson Distribution. Here is Michael’s analysis using that, getting a result approximately the same as the first way he calculated it.
Analysis courtesy Michael Bang from this post.
The mathematically inclined can use the Poisson distribution to estimate the likelihood for different number of sales in a year from a portfolio of a certain size for a given STR assumption.
6. Annualized STR Can Be More Than 100%
While at first glance you might think that 100% is the maximum possible annualized STR, that is in fact not correct. The annualized STR means the rate that would be sold if domain names were held for a one year period.
For example, let’s say I just started in domain investing, and acquired 10 domain names, and sold 5 of them over a period of just 4 months. That is pretty unrepresentative of the market, but if it did happen, the annualized STR would be 150% {=5/(10*(4/12))}. It is at least theoretically possible that the annualized sell-through rate is more than 100%.
7. STR to Predict Probability of Sale
Let’s say you have been in domain investing for many years, have a stable portfolio size of 2000 domain names, and, while there is some variability from year to year, your annualized STR is around 1%.
If you acquire a new name similar in quality to your overall portfolio, you can use the past STR to predict the probability of that name selling during the coming year is about 0.01, or 1%. Probabilities are normally computed on a scale from 0 to 1, but can be expressed in percentages by multiplying by 100.
8. Same STR, Different Uncertainty
Let’s consider two investors.
- Investor A has 100 names and sold 4 during the year, suggesting a STR of 4%.
- Investor B has 10,000 names and sold 400 during the year, also 4% STR.
It is a crude approximation, but often the square root of a number is taken as roughly representing the standard deviation. In a normal statistical distribution, there is about 68% chance the value falls within 1 standard deviation, 95% within 2 standard deviations, while more than 99% fall within 3 standard deviations.
So if we go back to our two domain investors, A sold 4 names, so the square root of that is 2. That means that 68% of the time, if many similar investors were considered, the number sold would vary from 2 to 6 sales in a year, and 95% of the time from 0 to 8 sales per year. If we express those as STR, we don’t know if the real STR, long term, is from near 0 to 8%. So on the basis of this one year with limited sales, we can’t conclude much about long-term STR with any confidence.
Investor B had 400 sales, and the square root of that is 20. Therefore 68% of the time, if we looked at the same investor over many years, or many similar investors over one year, the number sold would be from 380 to 420. In 95% of the time it would be 360 to 440, with a corresponding uncertainy in STR only from 3.6% to 4.4%. Even if we insist on better than 99%, still the uncertainty is only from 340 to 460 sales, with corresponding STR range from 3.4% to 4.6%. In summary, if we take 95% of the time as the standard:
- Investor A “true” STR may be 0 to 8.0%.
- Investor B “true” STR is likely in range 3.6% to 4.4%.
Read more on normal distributions here, though be warned that the math treatment is heavy.
9. Many STRs
@Recons.Com has suggested that it is useful to think not of just a single STR for a portfolio, but imagine your portfolio as having a number of buckets, each with their own STR. If you have a large enough portfolio, with significant number of sales. you can calculate multiple STR values.
For example, depending on your portfolio you might have a bucket for legacy extensions, one for created brandables, and one for .io or .ai names. Keep in mind the uncertainty when dealing with small numbers, though.
10. STR of Terms
If you use the Domain Insights tool available at Atom, for many names it will tell you the STR for the term based on Atom (SquadHelp) sales data. Therefore as well as considering STR for different extensions, or types of name, STR can be applied to rates that certain terms sell.
It is only part of the picture, just because names with say the term ‘health’ sold with a healthy STR does not mean that any combination of two words that include ‘health’ will sell. I don’t know for sure, but I suspect the term STR values at Atom are not annualized STR, but over the life of the marketplace. Nevertheless, I find this information really helpful when considering acquiring names.
11. Is STR Good For Anything?
So if DVPL (dollar volume per listing) is a better metric of how you are doing, why does STR matter at all? I think the most important use for estimations of STR is in deciding if a particular domain name is worth holding long time.
Multiply your best estimates of STR for this name, times the net return on likely selling price, and compare that to the annual holding cost. True, your STR will be an estimate, with uncertainty, but this framework is helpful. This overlooks any parking revenue, or considerations of changing value in a trending name.
Figure from Is Domain Name Investing Profitable? by Bob Hawkes at NameTalent.
Read more on this technique in the final section of the article Process to Rate and Price a Domain Name – Part 2.
Dividing your portfolio into buckets makes sense in this regard. For example, the annual holding costs for .ai, .io and some new extensions are higher than for legacy, so you need to make sure that is justified on the basis of estimated STR and selling prices.
Final Thoughts
I think STR is something that investors should consider, but don’t focus only on sell-through rate, it is just one of the factors that needs to be considered.
For additional reading on this topic, as well as the 2019 article Domain Name Sell-Through Rates, I would recommend What Are The Odds? Applied Probability for Domain Investing.
Thanks to Michael Bang for permission to share part of his analysis in this article.
Last edited:
