Hi all,
Phew!
I've finished a major algorithm update. You have been heard and I'd like to thank everybody for giving me valuable feedback during the first phase of the public BETA test. I hope you continue to help me make this tool better by communicating with me during this second phase of beta testing.
I'm very pleased to release EstiBot BETA v2.0
To put it short, I've made a major update to the system and changed quite a lot of things. This is a lengthy post so go directly to EstiBot.com if you're not in the mood for my ramblings
Some important changes:
-backlinks are now considered in the valuation. This gives a more accurate estimate for old domains and existing websites. However, there are so many more variables that determine the value of an existing website, that you should keep in mind that the estimate is at best only a pointer, an indicator of potential. Keep in mind that when EstiBot valuates google.com at USD 208,908,904 - of course it does not mean that you can buy Google at that price, it just means that it's damned valuable and you'd better believe it
-Number of backlinks (google) are now displayed for the users convenience
-the user is now informed whether the Searches/mo field represents exact searches or what I call "search potential" in case exact searches were not found using the currently available, very limited tools. This should increase the usability of EstiBot as a keyword research tool. EstiBot can valuate most domains equally accurately whether exact searches are found or not.
-the viewable database has been removed. I still use it for beta testing purposes but I have no interest, intention or time to look at let alone register individual domains when hundreds of names are punched in every day. I destroy the database at regular intervals after I'm done testing
-the "appraise another domain" button has been added to the results page by public demand
-I have made numerous filters, error checking routines, statistical updates and so on in order to increase the overall accuracy. I won't bore you with the details
The algorithm is getting very complicated....
I have attempted to keep EstiBot optimized for the low- to mid end domain range, because I feel this will be most useful to the majority of fellow domainers. This means that I aim to make it as accurate as possible in the zero to $20,000 range or so. Actually, low end domains are much, much more unpredictable than high end domains as to their monetary value. It would be much easier to optimize for the high end, but I like a challenge as much as the next guy
It still does a pretty good job in the high end range, too, but any valuation of above 10,000 or so means that you have a potentially valuable domain name and it might just be worth getting a professional appraisal. To the best of my knowledge, I sincerely feel that SEDO appraisals are best.
Remember that it's anybody's guess whether any given domain will ultimately sell at $10,000 or $90,000, and similarly in the other ranges, $1,000 or $9,000... so what you are getting with EstiBot is a ballpark valuation. The rest depends hugely on the motivation of both buyer and seller. Anything above $100,000 is just a rough pointer.
Here are the results of my first test run today. First, I went to
http://namebio.com and clicked a search for the 100 most recent domain sales (excluding Bazaar sales because those are often sold at much below value).
Then I ran those names through EstiBot Beta v2.0, and here are the results in the order in which they appeared on NameBio, warts and all - I did not remove any unfavorable results. I will comment some of these also..
I'd like to thank Justin / DomainSpade for this invaluable tool called NameBio.com...your site has been crucial in the development of EstiBot, it has made testing so easy - I've added a link to your site from my main page as a modest way to give something back.
As a general comment, I am not of the opinion that a known sales price is necessarily the best valuation of a domain name. Usually the buyer almost by definition thinks that the domain name is more valuable than what he is prepared to pay for it. Conversely, some buyers seem to pay hugely exaggerated prices for domains, at SnapNames for instance. So the actual, known sale price is not necessarily the real value of a name, there are so many more factors. Of course, it does give a very good general indication of value.
Here is the data from today. The first number is the EstiBot valuation, and the second number is the actual known sales price.
webmosis.com 17 2,500 <--- unpredictable brandable, end user stuff
MxGamer.com 115 200
dinosaurs.mobi 1,496 500 <-- perhaps EstiBot is too mobi optimistic?
TravelSay.com 164 3,000 <-- end user magic again
MassageWeb.com 605 700
PearlFunding.com 6 850 <-- machine cannot see the putative beauty in this name
FatDrug.com 1,515 2,500
pzow.com 174 1,500 <-- EstiBot likes LLLL.coms but is conservative and it's good to remember that there is huge end user potential in LLLL's
hks.mobi 236 210
BelleBella.com 430 995
aaaTunes.com 0 299 <--- who'da guessed?
TelephoneBills.com 2,830 2,000
GlobalEye.com 920 4,000
PromotionalGames.com 1,475 250
r1.org 2,194 1,397
677.net 749 1,250
ylm.de 78 672 <-- .de is strong, but difficult to valuate, big variance
YoungAthlete.com 2,687 300 <-- nice buy
WhatMatters.net 6,633 1,000
ShareStory.com 5,737 1,888
GymMachine.com 3,237 1,000
MiniLaptops.com 662 1,000
YogaWeb.com 631 2,600
SpaceSystems.com 17,232 2,000 <-- amazing domain name a very good buy in this age of rocket boom and private space business
LawyerSoftware.com 4,327 1,000
SchoolOfFinance.com 3,126 1,600
SecretaryJobs.com 9,271 3,000 <-- a very nice buy
kazbah.net 31 488
bu0.com 321 260
xy4.com 500 412
studentenkredit.net 58 333
PimpedMyRide.com 451 398
fff0.com 151 151 <--- speechless
OK, there is quite a bit of variance in this very short list. But when statistically tested, EstiBot reaches now a very high predictive value. Also, the next sale of the above names may well be closer to the EstiBot valuation
I've tested the system on thousands of domains. When I look at the above list with its fluctuations, I am quite happy, it's just as can be expected for a short sample in this unpredictable world of domain sales.
Here is another test from today, more in the high end range:
rex.com 160,435 395,000
vn.com 110,440 100,000
part.com 674,645 100,000
15.com 236,935 100,000
amt.com 114,542 100,000
vodka.com 205,262 3,000,000
character.com 284,756 100,000
microwave.com 418,821 95,001
announce.com 18,133 90,000
06.com 90,898 80,000
moka.com 14,447 72,223
bcf.com 15,893 71,200
ofertas.com 15,694 76,214
kut.com 48,349 65,000
pat.com 131,345 63,000
pog.com 27,265 20,000
SpringBox.com 12,282 20,000
yule.com 14,917 20,000
freo.com 3,311 20,000
As a general comment to this, I'd say that EstiBot is very rarely wrong in determining the overall ballpark. If it says your domain is worth >10,000, a thorough human evaluation is definitely warranted.
Also this demonstrates how high end domains are more predictable than low end domains.
Anyway, there it is, enjoy! Please do report major malfunctions to me as it is still in beta
Thanks and good luck
Josh
