Brandbot is the Artificial Intelligence which powers much of Brandpa. At the moment, Brandbot is responsible for:
- Appraising domains when you submit them
- Optimising what domains appear first in search results
- Writing text to describe a domain
How Brandbot appraises domains
We have gathered data on millions of domain names, from our own sources and from third parties. We also use data from our sales, visitors, offers, historical trends, search volumes, and more. All of this is used as training data for Brandbot, to help it predict the highest price that a domain is likely to sell for.
It’s easiest for Brandbot to value something that looks like something it has seen before. For example, it has a pretty good idea about short pronounceable domains, high value keywords, and so on. If you throw a random string of letters at it, Brandbot will usually conclude that domain is unlikely to sell for much.
Because it’s an AI it doesn’t have the broad understanding of a human, so it will struggle with misspellings, puns, and implied meanings. For example, Brandbot wouldn’t have guessed “covfefe” could be valuable.
Brandpa sells a lot of brandable domains, so we’ve aimed to teach Brandbot what makes brandable names ‘sound good’. For example “eryson.com” sounds better than “zaxscd.com”. You can see how Brandbot scores the sound of a given domain, as the “natural” percentage:
Brandbot also considers the consonant/vowel/number patterns (e.g. “hello” = CVCCV), number of syllables, words, phrases, alliteration, phonemes, and more. We have extensive data on how much these factors tend to impact sell through rates and valuations.
Another factor is ‘recognition’, whereby more common phrases, words, and numbers are more likely to sell. For example, numbers like “365”, “99”, or “101” are more valuable than random numbers like “98342”. With words, we also apply an analysis of the likely sentiment, frequency, and connotations.
Even considering all of this, there is no perfect way to appraise a domain, and Brandbot won’t always get it right. If you have trouble with Brandbot’s suggestions, you can always ignore or override them.
How Brandbot optimises search
Search results are ordered by Brandbot. It learns relationships between keywords, allowing it to make smarter suggestions that don’t just match the text you type in.
Brandbot also learns from a user’s behaviour what domains are more likely to interest them. For example, if a user favourites a domain, we may prefer to show them other domains which are similar. We also use this information to selectively advertise to some prospective buyers, with domain names that they paid particular attention to.
How Brandpa writes text to describe a domain
Currently, Brandbot uses a relatively simple approach, combining fragments of human-written text together, depending upon the keywords and industries applied to a domain. We do some extra work to ensure this makes grammatical sense.