Summary

  • ARIN's Board-adopted 2026–2028 strategic direction calls for 95% of registry customers to be under Registration Services Agreements.
  • Its ARIN 58 data-accuracy report identifies 1,135 organizations that already have some resources under agreement and others outside it. The public materials give no current numerator, denominator or rule for counting those mixed holders toward 95%.

A percentage can tidy a board slide while leaving the registry untidy. ARIN's strategic direction sets a concrete ambition: resolve legacy-holder issues and put 95% of registry customers under Registration Services Agreements. At ARIN 58, the Board report repeated the target among the 2026–2028 plan's key actions. It also said the Board had aligned on meaningful performance indicators. The indicator itself has yet to appear beside the goal.

The snag is sitting in ARIN's own Data Accuracy presentation. Its outreach pipeline contains 10,218 organizations. Of those, 9,083 have no existing agreement relationship. Another 1,135 are mixed: their organization records contain resources already under an agreement and resources outside one. Under an “any RSA” count, all 1,135 can appear in the successful column. Under a “fully covered organization” count, they remain work to be done. Both calculations can be described in English as customers under agreement. They describe different registry states.

A third group makes the denominator harder. ARIN lists 2,660 organizations with no user contact information on their records: no email, phone contact or address. They cannot be treated as ordinary outreach prospects. The goal does not publicly say whether they are inside the registry-customer denominator, counted separately as unresolved records, or excluded until a holder is established. Moving a record between those definitions could improve a percentage without improving the address book.

ARIN also reports 372,877 /24-equivalents of space and 16,212 legacy CIDR blocks outside agreements in Q3 2026. Uncontracted space fell 81% from Q1 2024, with a large step in Q4 2025 when 2,768 prefixes totaling 1.28 million /24-equivalents moved under agreements. That is substantial registry work. It is measured in addresses and blocks. The strategic target is measured in customers. A large holder can move the space chart dramatically while changing a customer count by one; thousands of small records can move a customer count while barely touching the space chart.

The distinction has an operating edge. ARIN's legacy-services page says an uncontracted legacy holder can still maintain registration data, reverse DNS and DNSSEC. Access to ARIN's RPKI and authenticated IRR requires an agreement. Thus the choice of what counts toward 95% also controls what the Board tells itself about service reach. An organization with one contracted block and another outside the contract is already in the door for some resources. The public scorecard needs to show the resources still in the hallway.

ARIN could publish the answer in one line of arithmetic: define the eligible organization population, show how many are fully covered, state how mixed and no-contact records are treated, and repeat the count each quarter. Its present cohort table is useful precisely because it exposes the unfinished cases. The Board's headline number should use that candour. Otherwise 95% is a destination on a map that omits the road markings.