The previous two articles in this series^{1, 2} have outlined techniques for ‘mining’ brandable domain names—that is, domain names of potential interest to entities looking to launch a new brand name and associated website—from the enormous dataset of unregistered names (determined via zone file analysis). The key element of the identification process is the implementation of filtering techniques to identify a (relatively) short-list of candidate names fulfilling certain criteria or comprising particular name patterns, which can then be manually reviewed for marketing and branding suitability. The previous articles have addressed two techniques which can be applied to this process; namely the use of phonotactic analysis to identify ‘readable’ (or wordlike-‘sounding’) names, and the use of algorithms to generate (alternatively spelled) variants of specific keyword- or brand terms of interest.

In this follow-up, I consider the use of combinations of these techniques to identify a special class of domain names; those where the name (strictly, the second-level name, or SLD—i.e. the part to the left of the dot) is a term ending with ‘-ize’ (or variants). ‘ize’ names are popular in branding, and can convey a sense of modernity and dynamism³. There is also branding wisdom suggesting that names which function as a verb (or have a name structure which is amenable to this) can aid with memorability⁴. Even just amongst the set of companies offering branding and marketing-related services, we find a number of ‘-ize’s (Localize (local-ize.ch), Merchize (merchize.com), Partnerize (partnerize.com), e-tailize (e-tailize.com), Visualize (visualize.design), Color-ize (color-ize.com), etc.).

Domain names of this type can take a number of forms, including actual dictionary terms, or ‘sensationally’ spelled variants—which can also apply to the suffix itself; the option to use ‘ise’, ‘ize’, ‘yse’ or ‘yze’ variants—all of which ‘read’ similarly—provides an additional degree of flexibility to the search process (and can also be a consideration when looking to address a target market—‘ize’ spellings being more common in the US and ‘ise’ in the UK, for example). Other categories of brandable ‘ize’ names include neologisms, phrasal names (e.g. ending in ‘wise’ or ‘size’), or other plays on words (e.g. where a word ending in ‘-ify’ can be ‘translated’ into ‘-ifize’, as a soundalike of ‘-ifies’).

Overall, the total numbers of unregistered names are very large, demonstrating the importance of an effective filtering approach. For example, the analysis identified 65,864 unregistered domains of length 6 characters and ending with ‘ize’ (or ‘ise’, ‘yse’ or ‘yze’), 1.8 million 7-character instances, 47.5 million 8-character instances, and 1.235 billion 9-character instances (outside the set of registered examples, which run alphabetically from aaacruise.com to zzsunrise.com).

The filtering process used to identify brandable examples can include combinations of any of the techniques discussed previously, or others—including (for example) searches for domains matching certain consonant (C) and vowel (V) patterns, with (for example) ‘VCVC-ize’ or ‘CVCVC-ize’ perhaps more frequently producing credible names than other combinations. In this analysis I consider the prevalence of exact matches to, or phonetic / sensationally-spelled variants of, dictionary ‘-ize’ words, as an illustration of the types of attractive available domains which can be identified.

As with the previous articles, a group of the domains with greatest brandable potential was identified through manual review of the filtered short-list, and registered as a ‘test-set’, allowing an assessment of their estimated values based on an AI tool (Table 1).

These findings provide a further illustration of how credible brandable domains can readily be identified from the pool of unregistered names using suitable filtering and analysis tools and techniques. It is also noteworthy that, even with names as short as eight characters, there are instances of unregistered dictionary terms to be found.

In practice, the production of a short-list must be guided by the would-be potential brand owner, based on their preferences on a potential name, its structure, and their area of business (and associated potential keywords), but these analyses have shown that bespoke filtering processes can relatively straightforwardly be applied to the universe of names—through the use of effective algorithms—to identify sets of candidate names meeting specific requirements.