Two sides can oppose each other strenuously and still be wrong in exactly the same way. For or against, too much of the debate about the new ICANN top-level domains (TLDs) ignores TLD signaling and uses inappropriate TLD success measures. Here I spotlight the key mistakes by concentrating on “.biz” registrations, and I put forward some possible remedies.

The opposing camps point to comparative registration data on various TLDs. Those who are against expansion of TLDs argue that “.biz” was intended for business use but was unable to compete with “.com”; ergo, they say, there is no demand for new business TLDs. The pro-expansion camp argues that trademark owners don’t need to register their brands under new TLDs, as evidenced by the small number of “.biz” registrations, and point out that only a few trademark holders have combined their brand under “.com” with “.biz” (or with other TLDs).

Both sides must consider this: As signals, “.com” and “.biz” are neither strong substitutes nor complementary. For a business, the “.com” sends two signals: longevity and global presence. The two add up to a mystique, a feel of authoritativeness that bolsters a site. This mystique is beyond the reach of a newcomer like “.biz.” A well-established outfit, one with a global presence, would choose “.com”—or so consumers tacitly assume, including consumers who might type in a site’s address when cruising the Internet. Thus, a “.biz” site largely foregoes direct navigation and typo registrations.

But the pro-expansion side can’t look to the lack of “.biz” registrations as proving there’s no danger that companies would have to register under all sorts of new TLDs just to protect themselves. Using both “.com” and “.biz” doesn’t make much sense. After all, what does “.biz” say that isn’t already signaled by “.com”?

The pro-expansion camp has used an incorrect measure of TLD failure: namely, taking a point in time and comparing the total number of “.com” registrations with the number of dual registrations—that is, the number of brand names that have been registered under both “.com” and “.biz” (see Analyzing Domain Names Registered Across Multiple Existing TLDs and Implications for New gTLDs).

Although there is no one best TLD-success measure, there are a number of incorrect ones. Arguing that defensive registrations will be unnecessary, the pro-expansion side compares the number of “.biz” registrations with “.com” registrations at a point in time. Such a comparison is, in general, misleading. For starters, as a few people have pointed out, there’s the apples-and-oranges danger of comparing restricted and unrestricted TLDs. A TLD tailored for a specific group or region—from “.eu” for Europe and Europeans to “.cat” for Catalonia and Catalans—may well be restricted to that region or group, allowing it a sharply limited pool of potential registrations. (Of course, some restrictions are artificial so as to suggest scarcity and allow jacked-up registration fees.)

Although “.biz” is not restricted, in that it is available for registration by anybody, the extension is intended to signal a for-profit business. That choice gives it a narrower scope than “.com,” resulting in fewer registrations.

Comparing the number of a given TLD registration to “.com” at a point in time is also a mistake. Usefulness can change over time; if you don’t take that fact into account, you might as well compare the current use of floppy disks with USB flashcards and decide that floppy disks never played a part in the mass adoption of computers. By contrast, “.biz” at no time played a significant role in the development of the Internet. Moreover, there might be a negative “.biz” reinforcing signal, namely that when the domain community bashes “.biz,” such action deters companies from registering them.

At a given point in time there may have been significantly more brands registered under “.com” and “.com” only than under “.com” and “.biz.” So what? Without a statistical test, you cannot conclude that companies did not at some point also register “.biz”; thus, you cannot reach a conclusion that trademark owners will not, at least initially, register their brands under various new TLDs. They can stop renewing registrations when they get a better picture of the signaling and trademark-protection landscape. Thus, the evidence that domain name registrations that include the Amazon brand are mostly under “.com” does not rule out short- or long-term demand by Amazon for any of the new TLDs.

Moreover, early adoption of a TLD does not necessarily imply long-term demand, as the “.nu” experience demonstrates. The demand for “.nu” in the mid- to late 1990s (mainly in Sweden, Denmark, and the Netherlands) was due to “nu” meaning “now” in the relevant countries’ languages and to stringent registration requirements for “.se,” Sweden’s country-code TLD (ccTLD). Furthermore, the meaning of a TLD signal can change over time. For example, the large number of “.net” registrations is for name-servers, not for-profit Internet businesses (as was originally intended).

For better TLD success measures, you can meaningfully compare the penetration level of a restricted to an unrestricted TLD by dividing the number of registrations by the target population. A second measure is to compare the aggregate market value of the TLD based on average market price of second-level domain names (or the market price premiums among TLDs) times the number of registrations. A third is to measure acceptance/penetration levels by comparing the number of annual registration for, say, the first five years for various TLDs. A fourth is to plot annual growth rates in registrations. Thus, for example, a downward TLD trend for any of these measures is valuable information.

(One tool for monitoring the performance of the various new TLDs is to use S-curve analysis, which typically captures a technology’s maturity/diffusion life cycle.)

In conclusion, a signaling framework can explain why “.biz” has not had the penetration of “.com” and why companies have registered their brands under a select few TLDs and ccTLDs. Also pundits should be more careful in their use of comparative TLD registration because numbers can lie—if you don’t ask them the right questions.