tips Discounting domain prices - 2023 update. (Don't discount much)

[twiki]

Some of you might know that in the past I did a lot of posts on discounts, and pricing to clear.

This is sort of an update post, as to what I am doing in 2023. Been testing lately, side note I always test things.

Note: This might not apply to everyone so do your own best judgement. I'm just sharing what I understood and I am using now.

Okay, now to the point.

In the past I used to sell a lot of domains at $199, $299 and whatever prices. Also at $75 in some cases. This was pre-2022. And quite a lot of other folks here have followed suit with good results in general. But in 2022, all this has stopped. Something did change in the market.

So what I do now? (except from a few tests here and there, and NP sales):

Never discount much under $1K, if at all.

And let me explain why.

There are 3 obvious types of buyers of domain names if you ask me:

1) Retail buyers (that will pay top dollar)

2) Lowballers (in lack of a better name; anyway buyers that will only ditch a couple hundred out, for any reasons including they don't have more; usually NOT investors )

3) Other investors.

The difference in my perception is that investors no longer pay $199 or $299 on a domain since 2022 started. Or at least most of them. And if you do the math in the current market, you'll know why. They'd be making losses.

So what do you do now by discounting that much? You actuall give the option to lowballers to get your domain for cheap before the right buyer comes, and you might definitely miss on that top retail sale.

I currently use 2 prices, $999 and $750 currently and still doing some alternate testing (haven't decided yet which brings more).

Please also note that if you by chance get 5 sales at $199, that's just $1k. You might feel you sold a lot, but no, that'd be just one discounted sale of 1k. It's a perception thing that might fool some of us (it definitely "tries to play me" as well).

You also might get hit with more fees for more payments.

And if you plan to renew them? That's entirely dumb pricing in such case, as overall at 1% or 2% yearly average sales ratio, you are bleeding money fast. Especially in the current market where domain prices went up and sales went down; so yearly ratio might be significantly smaller, you definitely need to price high.

Besides, domain prices haven't went down lately. And no chance they will by a lot. Because it doesn't make sense economically for most investors to reduce prices now.

Now the caveat is, everyone loves a good deal.

So if you had your domain at say $2K, dropping it to $1500 or $1700 might bring you a very happy buyer.

But do not discount it to $250 or $150 or whatever; as that will likely only bring you loss overall. Assuming though that you made your research well and that your prices are accurate.

Happy sales in 2023!

Later edit: Most of my domain are .COMs of 1-2 words and not liquid type, to be more specific. Either meaningful or brandable .COMs.

 

[okaydomains]

Actually, you can test in a meaningful way if you have thousand+ names priced, let's say at $1500-2000 currently and giving around 1-1.5% STR.

Just take 500 of those randomly and price at $199 and another 500 (randomly) at $750 or $999. Let it stay that way for few months and see which bucket did better. Even better, if you also have the third bucket where the prices remained unchanged for benchmarking.

My background is not in statistics, but I don't think that's a big enough sample size or enough sales to produce statistically significant data, especially if you add a third bucket. I also don't think it's easy to find two equivalent names, especially when limited to your own portfolio.

It seems that the only people with enough data to produce useful results are the marketplaces. I'm not sure where that leaves us.

 

[okaydomains]

Evidently some of the names were on the radar of potential buyers and when they saw the big discount they pulled the trigger and bought.

This is one of the things I like awake at night wondering about.

If I lower the price and it sells, was the buyer someone who previously passed on the name because it was too expensive? Or did the earlier visitors see the high price and move on, never to return? Is it a brand new visitor who is now making an impulse buy? If so, would they have paid more?

I think we are not meant to know the answer to this, at least until we get better analytics from Afternic.

 

[Recons.Com]

My background is not in statistics, but I don't think that's a big enough sample size or enough sales to produce statistically significant data, especially if you add a third bucket. I also don't think it's easy to find two equivalent names, especially when limited to your own portfolio.

It seems that the only people with enough data to produce useful results are the marketplaces. I'm not sure where that leaves us.

I understand. That is why a large portfolio is important to draw meaningful conclusions. Marketplaces could have helped, but they seem to have no knowledge, desire or motivation.

Regarding equivalent names, that is not required. Let's say you have 1500 names that made 15 sales in 1500-2000$ range. That tells you that you have a decent portfolio of pretty much hand-reg to closeout quality. Now if you randomly distribute them into 3 buckets, all very good to bad ones within the mix would be pretty evenly distributed just because that is how random works across large sample. So all buckets would be the same initially and the only difference would be the new prices.

 

[twiki]

My background is not in statistics, but I don't think that's a big enough sample size or enough sales to produce statistically significant data, especially if you add a third bucket. I also don't think it's easy to find two equivalent names, especially when limited to your own portfolio.

It seems that the only people with enough data to produce useful results are the marketplaces. I'm not sure where that leaves us.

Here's about buckets,

I had split my portfolio this year in 2 buckets. 2000 were in bucket A, 3000 in bucket B. Sort of off numeric as it was by accident, but anyway.

Through 2,5 months, bucket A had no sales, while bucket B had several, in the tens range (all other sales actually in 4-fig range). Note, the names are not sorted by strength among the two.

Would you think that is accurate or not to draw some conclusions on?

Cause it's the same thing as running an 1000 names A/B test over 5 to 6 months.

(edited a bit for clarity)

 

[Recons.Com]

Here's about buckets,

I had split my portfolio this year in 2 buckets. 2000 were in bucket A, 3000 in bucket B. Sort of off numeric as it was by accident, but anyway.

Through 2,5 months, bucket A had no sales, while bucket B had several, in the tens range (all other sales actually in 4-fig range). Note, the names are not sorted by strength among the two.

Would you think that is accurate or not to draw some conclusions on?

Cause it's the same thing as running an 1000 names A/B test over 5 to 6 months.

(edited a bit for clarity)

I am not sure I understood what you are saying about the sales results. And also how exactly you split the names between the buckets.

If you split the names using random function in a software (excel or something else) and then either did not change anything else or changed the same, the results in the buckets should be proportionately similar.

 

[okaydomains]

Thank you @twiki and @Recons.Com for the help. Statistics is not my strong suit, but I have read a decent amount and have some related experience.

Regarding equivalent names, that is not required. Let's say you have 1500 names that made 15 sales in 1500-2000$ range. That tells you that you have a decent portfolio of pretty much hand-reg to closeout quality. Now if you randomly distribute them into 3 buckets, all very good to bad ones within the mix would be pretty evenly distributed just because that is how random works across large sample. So all buckets would be the same initially and the only difference would be the new prices.

I see - so you're saying that with the random selection the differences in quality will even out in aggregate. That makes sense to me, but I'm still not sure that it's enough names for this to work.

Through 2,5 months, bucket A had no sales, while bucket B had several, in the tens range (all other sales actually in 4-fig range). Note, the names are not sorted by strength among the two.

Would you think that is accurate or not to draw some conclusions on?

Cause it's the same thing as running an 1000 names A/B test over 5 to 6 months.

Ok, so here might be the crux. I like A/B testing and have done quite a bit of it on ecom sites during my consulting days.

But what I learned was that you need a *ton* of traffic to get meaningful results. Most webmasters don't have it, not even close. And I think that's something the A/B testing industry tends to hand wave away.

So I don't think most A/B testing that's happening is statistically rigorous, and it's just confirming someone's gut feelings.

But to answer you directly - no, I don't think 4 sales over 2.5 months is enough to draw *statistically meaningful* conclusions from.

I'm not saying you didn't learn anything or your conclusions were invalid. I'm just saying that no, I don't think you had enough traffic to make that determination statistically.

I believe @AbdulBasit.com has commented on this before, and feels that for a real test you need to move all of your names and leave them for a while (at least six months). That sounds more accurate to me.

 

[twiki]

Thank you @twiki and @Recons.Com for the help. Statistics is not my strong suit, but I have read a decent amount and have some related experience.


I see - so you're saying that with the random selection the differences in quality will even out in aggregate. That makes sense to me, but I'm still not sure that it's enough names for this to work.



Ok, so here might be the crux. I like A/B testing and have done quite a bit of it on ecom sites during my consulting days.

But what I learned was that you need a *ton* of traffic to get meaningful results. Most webmasters don't have it, not even close. And I think that's something the A/B testing industry tends to hand wave away.

So I don't think most A/B testing that's happening is statistically rigorous, and it's just confirming someone's gut feelings.

But to answer you directly - no, I don't think 4 sales over 2.5 months is enough to draw *statistically meaningful* conclusions from.

I'm not saying you didn't learn anything or your conclusions were invalid. I'm just saying that no, I don't think you had enough traffic to make that determination statistically.

I believe @AbdulBasit.com has commented on this before, and feels that for a real test you need to move all of your names and leave them for a while (at least six months). That sounds more accurate to me.

The test actually was over 1 year. This are the stats for 2023 and those weren't 4 sales but much more.

The other results over 2022 yielded the same statistical result, minor difference (95% belonging to B).

 

[twiki]

Thank you @twiki and @Recons.Com for the help. Statistics is not my strong suit, but I have read a decent amount and have some related experience.


I see - so you're saying that with the random selection the differences in quality will even out in aggregate. That makes sense to me, but I'm still not sure that it's enough names for this to work.



Ok, so here might be the crux. I like A/B testing and have done quite a bit of it on ecom sites during my consulting days.

But what I learned was that you need a *ton* of traffic to get meaningful results. Most webmasters don't have it, not even close. And I think that's something the A/B testing industry tends to hand wave away.

So I don't think most A/B testing that's happening is statistically rigorous, and it's just confirming someone's gut feelings.

But to answer you directly - no, I don't think 4 sales over 2.5 months is enough to draw *statistically meaningful* conclusions from.

I'm not saying you didn't learn anything or your conclusions were invalid. I'm just saying that no, I don't think you had enough traffic to make that determination statistically.

I get what you mean though. You need a large enough sample for the set to have any precision. Hundreds of sales ideally.

Might also be correlation vs. causality... different things.

But in domaining, I'm afraid we often have to do with the scarce data we have.

I believe @AbdulBasit.com has commented on this before, and feels that for a real test you need to move all of your names and leave them for a while (at least six months). That sounds more accurate to me.

Indeed, Abdul has said that. He also stressed it out to me directly, 6 months not less.

 

[magnificentdomains]

Interesting. I am wondering how long your test ran for the stage where you saw great improvement and how long for the stage where it flattened out. And also, how many names were involved?

Again, for pure comparison, one bucket of randomly selected names should remain unchanged. Then you can compare to both the increased phase and flattened phase. This would allow to weed out the possible seasonality or general economy effects.

This test was on about 6-7k names priced at 2248. I dropped the price to 1462 and saw a very unusual, and noticeable increase over about a 40 day span. Initially I thought sub $1,500 must be the magic price point. Over the next few months data flattened out to normal sales volume. Currently those names are @ 1,988 and the rest of the portfolio ranges from 3-40k with a handful above that. I'm forever tinkering with price.

 

[twiki]

This test was on about 6-7k names priced at 2248. I dropped the price to 1462 and saw a very unusual, and noticeable increase over about a 40 day span. Initially I thought sub $1,500 must be the magic price point. Over the next few months data flattened out to normal sales volume. Currently those names are @ 1,988 and the rest of the portfolio ranges from 3-40k with a handful above that. I'm forever tinkering with price.

I found that that often tinkering with price isn't really helping sales but the contrary.

Though it helps in the beginning when you don't know exactly what price to use.

 

[magnificentdomains]

I found that that often tinkering with price isn't really helping sales but the contrary.

Though it helps in the beginning when you don't know exactly what price to use.

Yes I know what you mean. I don't mean to imply I am constantly changing prices. I would say on average once a year I will make a major category price adjustment. I have read up a bit on pricing psychology and always looking for a better price point given the economic climate. Although I do see myself using the new % option in afternic to run sales etc.

 

[poweredbyme]

Stop thinking like a seller, start thinking like a buyer.
You see an expensive domain which you can't afford. What would you do?

Almost nobody would regularly check if there is a price change unless the price is high enough.
Because time has a value.